Money Tree Investing

Speakers debate AI productivity gains, infrastructure investments, and model valuations amid market skepticism.

1h 27m Listen to the original Episode page
AI InfrastructurePublic vs Private MarketsValuation SkepticismProductivity GainsData CentersOpen Source AIInvestment Risks

Summary

Alex Chompff argues that AI unlocks massive productivity gains by organizing cognition below the human level, potentially adding trillions to global GDP. He suggests retail investors should focus on AI infrastructure and data centers rather than frontier models, which he claims are likely overvalued and inaccessible. Kurt Chisholm and Barb Friedberg express skepticism about AI valuations, citing the dot-com bubble, high capital expenditures, and AI's limitations in math and data accuracy. The discussion highlights the shift of economic growth to private markets and the reliability of infrastructure providers like AWS compared to model developers.

Key takeaways

  1. 1
    If only one and a quarter percent of the people in the world did that next year, that would be an additional trillion dollars in world GDP.

    Alex Chompff · 7:10

  2. 2
    That's why they're going public, because private capital can't sustain that burn at that rate.

    Alex Chompff · 21:04

  3. 3
    AWS generated 57% of Amazon's operating income in 2025 despite being only 18% of revenue.

    Unidentified speaker (speaker_0) · 1:00:45

Episode sections

  1. 0:01 – 5:30

    Speaker Background and Silicon Valley Evolution

    Alex Chompff discusses his background as a professional investor and AI entrepreneur, noting his upbringing in the Bay Area during the 1990s tech boom. He contrasts the builder-centric culture of early Silicon Valley with the current finance-driven environment, suggesting that finance is built upon tangible value creation.

  2. 5:30 – 15:20

    AI Productivity and Economic Impact

    Chompff references Michael Levinson's work to argue that AI unlocks a layer of value below the human level. He claims that if 1.25% of the global population generated $10,000 in value using AI, it would add one trillion dollars to global GDP, emphasizing that productivity gains are the primary unlock.

  3. 15:20 – 22:32

    Market Structure: Public vs. Private Companies

    The speakers discuss the decline in the number of publicly traded companies in the US, with Chompff citing a 40% decrease over 30 years. He attributes this to regulation and the availability of private capital, noting that growth is shifting to the private side where most AI innovation is occurring.

  4. 22:32 – 28:22

    Investment Opportunities in AI Infrastructure

    Chompff suggests that retail investors are forced into public markets like NVIDIA because frontier models and applications are private. He advises looking at data centers, electricity, and early-stage private companies, while warning that infrastructure and chips are likely already priced for outsized returns.

  5. 28:22 – 29:23

    AI in Investing and Productivity

    Alex Chompff discusses how AI excels at organizing information and finding patterns, potentially increasing investment returns through faster notifications. He highlights the transformative potential of AI for individual productivity and business value creation without requiring deep technical expertise.

  6. 29:23 – 32:08

    Disruption of Cognitive Industries

    Chompff compares the current state of various industries to travel agents in 1994, suggesting AI will radically change businesses reliant on human cognition. He notes that while individuals can use AI to create value, the economic benefits often accrue to capital rather than labor.

  7. 32:08 – 35:05

    Recursive AI Breakthroughs and Self-Improvement

    Chompff explains that as of late 2025, AI systems like OpenAI's Codex began writing their own successors, marking a shift to recursive self-improvement. He argues that AI is now better than humans at improving AI architectures, potentially rendering human input irrelevant for maintenance within a few years.

  8. 35:05 – 38:47

    Limitations of AI and Human 'Taste'

    Kurt Chisholm and Alex Chompff discuss AI's reliance on pattern recognition and historical data, contrasting it with human ability to conceive of entirely new concepts. Chompff cites Nvidia's CEO regarding 'taste' as an ineffable human quality that AI struggles to replicate because it is inherently backward-looking.

  9. 38:47 – 42:19

    Abundance vs. Scarcity and the Dark Fiber Analogy

    The speakers debate whether AI leads to a Star Trek-like abundance or a George Lucas-like scarcity. Chompff uses the 'dark fiber' analogy, suggesting that unused cognitive capacity will find applications, much like unused fiber optic cables were later lit by the iPhone, though he estimates a 20-year timeline for full realization.

  10. 42:19 – 43:09

    Investment Risks and the Gartner Hype Cycle

    Chompff places the AI industry in years 5-10 of a 20-year promise, advising investors to consider private markets to tolerate the risks of the 'trough of disillusionment.' He warns that while the technology is real, the investment cycle involves significant volatility and potential busts.

  11. 43:09 – 51:45

    The Mystery of AI Internal Workings

    Chompff argues that AI's internal processes are incomprehensible to humans due to the complexity of billions of parameters interacting via reinforcement learning. He compares this to nuclear physics or child development, where the underlying mechanics are known but the emergent behavior is a mystery until observed.

  12. 51:45 – 59:06

    AI Hype, PR, and Investment Skepticism

    Barb Freeberg and Kurt Chisholm express skepticism about AI investments, citing the dot-com bubble and the high capital expenditures of companies like Anthropic and SpaceX. Chisholm criticizes AI companies for using fear-based PR and lobbying for government protection, suggesting they cannot sustain themselves financially.

  13. 59:06 – 1:00:12

    Valuation Skepticism and Frontier Model Economics

    Kurt Chisholm argues that non-frontier AI models are worthless due to free open-source alternatives and claims frontier models are worth zero. He suggests Anthropic may go public soon but expresses doubt about the sector's near-term performance.

  14. 1:00:12 – 1:27:29

    AWS as the 'Picks and Shovels' Infrastructure Play

    Alex Chompff uses Amazon's AWS as an analogy for AI infrastructure, noting it generated 57% of Amazon's operating income in 2025 despite being only 18% of revenue. He highlights AWS's massive power capacity expansion and argues that infrastructure providers capture value more reliably than model developers.

Assets discussed

AssetKindStanceContext
MicrosoftMSFTequitybullishThere's only one Microsoft, there's one Apple, there's one Amazon, there's one Meta. You're not getting two Microsoft's just because it just doesn't work that way.
NVIDIANVDAequityneutralmost people have chosen to invest in the Nvidia's of the world, the semiconductor space as the path.
AlphabetGOOGLequityneutralGoogle for a really good list of publicly traded companies that sell equipment into data centers.
AmazonAMZNequityneutralcase for investing. I'm not a big fan of investing in AI or AI. AI-related companies at the moment. As I said, we're in the ninth inning or even extra innings in the bubble. There will be a time to buy it when it's cheap after it's all bombed out and buying Amazon at like two bucks a share back in like oh three, it was like dirt cheap, you know, or Apple
AppleAAPLequityneutralcase for investing. I'm not a big fan of investing in AI or AI. AI-related companies at the moment. As I said, we're in the ninth inning or even extra innings in the bubble. There will be a time to buy it when it's cheap after it's all bombed out and buying Amazon at like two bucks a share back in like oh three, it was like dirt cheap, you know, or Apple
S&P 500indexneutralbecause it wasn't even close to the performance numbers for those years for like the SP 500. It also made the egregious error of like the options expiration. It didn't know that it's like the third week of the month every single time. It just didn't know that. And the math was wrong in many occasions. And I would not have caught the math

“Stance” is how the speakers talked about the asset, not a recommendation.

Claims worth verifying

Michael Levinson is a biologist at Tufts University who developed a framework for describing cognition in a non-metaphysical way.

Alex Chompff ·5:50

How to verify: Search for 'Michael Levinson Tufts University cognition framework' or 'Greenfield Below' publication to confirm affiliation and work. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

AI is particularly good at organizing information and finding patterns, which can marginally increase returns.

Alex Chompff ·28:34

How to verify: Review academic literature on machine learning applications in finance, specifically studies on pattern recognition and alpha generation using NLP or time-series analysis. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Travel agents were ubiquitous in 1995 but are now rare boutiques.

Alex Chompff ·30:15

How to verify: Search historical employment data from the US Bureau of Labor Statistics (BLS) for 'Travel Agents' and compare 2026 employment numbers to current figures. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

OpenAI's Codex was written in part by its own prior coding program and released in December 2025.

Alex Chompff ·32:57

How to verify: Check OpenAI's official blog posts, GitHub repositories, or press releases from December 2025 regarding Codex updates and development processes. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Anthropic released 'Co-Worker' in February 2026, created entirely by Claude Code.

Alex Chompff ·33:17

How to verify: Check Anthropic's official announcements, blog, or product launch pages for February 2026 regarding 'Co-Worker' and its development history. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

AI systems are 'grown' via reinforcement learning and gradient descent, not programmed in the traditional sense.

Alex Chompff ·44:15

How to verify: Review technical documentation on Large Language Model (LLM) training methodologies, specifically reinforcement learning from human feedback (RLHF) and gradient descent algorithms. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Nvidia's CEO mentioned 'taste' as an ineffable human quality.

Alex Chompff ·36:24

How to verify: Search transcripts of Nvidia earnings calls or interviews with CEO Jensen Huang from 2026-2026 for mentions of 'taste' and human vs. AI capabilities. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Primary source

SpaceX stock peaked at a specific point that would have been useful for investors to know.

Alex Chompff ·28:54

How to verify: As SpaceX is private, verify via valuation reports from secondary markets (e.g., Forge Global, EquityZen) or news reports on SpaceX funding rounds and valuation peaks. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Anthropic and OpenAI are trying to go public.

Kurt Chisholm ·56:47

How to verify: Check SEC filings (S-1 or F-1) for OpenAI and Anthropic, or monitor official press releases from both companies regarding IPO plans. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Primary source

Anthropic lobbied the government for protection in exchange for a percentage ownership.

Kurt Chisholm ·58:47

How to verify: Search lobbying disclosure reports filed with the US Senate or House of Representatives for Anthropic and OpenAI, and review news reports on government relations activities. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Kurt Chisholm states that open-source AI models are free and only three months behind frontier models, making frontier models worth zero if they catch up.

Kurt Chisholm ·59:26

How to verify: Check GitHub for open-source model release dates relative to frontier model releases; verify pricing models of major AI providers. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

AWS generated $45.6 billion of Amazon's $80 billion operating income in 2025, representing 57% of operating income.

Unidentified speaker (speaker_0) ·1:00:45

How to verify: Amazon Q3 2026 Earnings Report or 10-K filing on SEC EDGAR; verify AWS segment revenue and operating income. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Primary source

AWS added 3.9 gigawatts of power capacity in 2025 and expects to double total power capacity by the end of 2027.

Unidentified speaker (UNKNOWN) ·1:01:05

How to verify: Amazon Investor Relations presentations or sustainability reports; verify AWS infrastructure announcements. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Anthropic's confidential IPO prospectus projects at least $518 billion of AI infrastructure spending over 10 years, with 80% non-cancelable.

Alex Chompff ·1:01:57

How to verify: Anthropic S-1 filing on SEC EDGAR (when public); Reuters article cited by Chompff. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Primary source

Anthropic filed confidentially for its IPO in June.

Alex Chompff ·1:02:17

How to verify: SEC EDGAR search for confidential submission status; Axios article cited by Chompff. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Primary source

AI assumed annual returns were from November to November and miscalculated S&P 500 performance, identifying 2021 as a bad year instead of 2022.

Kurt Chisholm ·1:11:43

How to verify: S&P 500 historical total return data; verify calendar year returns for 2021 and 2022. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

AI calculated a client's financial advisory fee was 30 basis points higher than it actually was.

Alex Chompff ·1:15:22

How to verify: Client portfolio statements and fee disclosures; verify calculation of advisory fees. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Anthropic is the customer writing enormous checks to AWS, while Amazon is the owner of AWS.

Unidentified speaker (UNKNOWN) ·1:02:35

How to verify: AWS customer lists or public partnerships; Amazon's revenue breakdown by customer segment. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

AI cannot do math well and relies on dirty data like Reddit and Yelp.

Kurt Chisholm ·1:08:47

How to verify: AI model documentation on training data sources; tests of AI mathematical reasoning capabilities. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Google's Gemini product is sending people outside of its click-based business model.

Alex Chompff ·1:24:23

How to verify: Google product updates and user behavior reports; Google's revenue reports on search vs. AI product usage. For recent events, verify 2026-08-06 through 2026-11-06. Use another period only when the speaker explicitly names it.

Do your own research

  • Verify the 40% decline in US public companies by checking SEC data or S&P Dow Jones Indices historical listings.
  • Confirm AWS's 2025 operating income contribution and power capacity additions via Amazon's Q4 2025 10-K filing.
  • Review Anthropic's confidential IPO prospectus filed with the SEC for projected infrastructure spending and burn rates.
  • Check Reuters and Axios reports on Anthropic's capital needs and IPO timeline.
  • Assess AI hallucination rates in financial contexts by comparing AI-generated calculations against primary sources like SEC filings.
  • Evaluate the 'Greenfield Below' framework by reviewing Michael Levinson's Tufts University publications.
  • Compare current AI infrastructure valuations to historical tech bubbles using Gartner Hype Cycle data and P/E ratios.
  • Investigate the performance of open-source AI models versus frontier models to assess the 'worth zero' claim for paid models.
Full transcriptexpand
0:01
Unidentified speaker (speaker_0): Welcome to the Money Tree Investing Podcast. Stock market, wealth, personal finance, value stocks. Invest in your life.
0:10
Unidentified speaker (UNKNOWN): Hello,
0:10
Kurt Chisholm: Smart Money T podcast listeners. Welcome to this week's show. My name is Kurt Chisholm, and I'll be your host. So today I'm joined with Alex Chump. How are you doing today, Alex?
0:18
Alex Chompff: Very well, Kurt. Thanks for having me.
0:20
Unidentified speaker (UNKNOWN): Yeah,
0:20
Kurt Chisholm: glad to have you on the show. For those listeners who aren't familiar with you, maybe you can tell us a little bit about your background.
0:24
Alex Chompff: Probably the most useful thing to know is. that I'm married with five kids and that shapes everything about what I do and choose in my life today. I'm a professional investor, but I got here by way of a keyboard.
0:38
Unidentified speaker (UNKNOWN): I
0:38
Alex Chompff: grew up in the Bay Area in the 1990s. I was lucky enough to be the kind of kid who while I was playing Dungeons and Dragons downstairs, my friend's dad was upstairs building People PeopleSoft, Soft or you know, the other team I played Dungeons, the other house I played Dungeons and Dragons on, you know, mom was one of the original administrators at Genentech. So it was kind of in the blood.
0:58
Alex Chompff: My dad patented his first product in I think '86. Maybe opened the first business in 89, still open today, putting his kids, his grandkids through college. Thanks. I started working with a keyboard as a data analyst in 1990. And then after I served in the Marine. Corps, I came home, went straight to Silicon Valley. I
1:18
Alex Chompff: had a stint as a director of technology for a really notable venture capital firm.
1:25
Unidentified speaker (UNKNOWN): You know,
1:25
Alex Chompff: later went on to manage data centers and industrial portfolios
1:30
Unidentified speaker (UNKNOWN): on
1:31
Alex Chompff: the way to becoming once again a creator and an uh investor professionally. My wife said I could go play again. That's what I do today. I'm very interested in the right now, the future of artificial intelligence, the joint between artificial intelligence and natural intelligence. I'm very interested in the future of work. I'm an active investor in all kinds of technology projects.
1:51
Alex Chompff: And about once a year I like to strap on my tennies and take a trip around the track.
1:55
Kurt Chisholm: Sounds to me like you were in the golden age of Silicon Valley. I I have a lot of nostalgia for those days when things were just starting up and people were building companies in their garage in their basement. And Silicon Valley's a different place now, isn't it?
2:11
Alex Chompff: Well, it's it's not the same place. What do they say? You never step in the same river twice. Certainly when I was a teenager, the most important thing in Silicon Valley was the Shoreline Amphitheater where you would go and watch a great concert. But that was a little bit of a trip. I spent most of my time in San Francisco and Berkeley and those areas. I didn't go into the valley proper until after I got
2:31
Alex Chompff: home in '95. I actually had a Economist magazine and it said they're making millionaires in Silicon Valley. And I showed it to my wife. I said, We got to go. And then we realized that Silicon Valley was like 45 minutes away,
2:44
Unidentified speaker (UNKNOWN): was
2:45
Alex Chompff: where the shoreline theater had been the whole time. You know what I mean? So we're like, oh well, gee. I sometimes think of myself like Forrest Gump. Like I'm in a lot of movies, but it's not necessarily my movie, if you understand. It's like being an extra in Hollywood. It's kind of fun.
2:57
Unidentified speaker (speaker_3): Extras make good money now. Unions to help them out.
3:02
Alex Chompff: Yeah, yeah. We need a sag for extras in Silicon Valley. That would be great.
3:06
Unidentified speaker (speaker_3): Oh, that's great. Remember the old days where people would I remember growing up, people
3:10
Kurt Chisholm: would geek out on like building computers back before they mass produced them. You were there were hobbies. hobbyists and people would make them and there were stores you can go to and you buy all these parts. It was so cool. And then wasn't cool anymore.
3:21
Alex Chompff: Yeah, that's exactly when I was a kid. I grew up in that era. I mean, but look, I'm still a kid today. I'm in my 50s. And literally in the last week, I've probably stayed up till at least three or four in the morning, three nights working on my new AI project because it's just so engrossing. I can't get away from building. I think in my mind, that's kind of
3:41
Alex Chompff: what changed after the dot-com boom or during the dot-com boom. It became as much about finance as it did about building. And finance is a very important part of the economy and obviously of great interest to your audience, but finance is typically built on top of something else. So even though finance can become much larger than what it represents, the futures market
4:01
Alex Chompff: is much larger than the stock market. And the stock market is itself a financial construct built on people, you know, making and creating and building and selling things that people find useful and interesting and valuable.
4:13
Unidentified speaker (UNKNOWN): So
4:14
Alex Chompff: I like to be a builder. I've always been a builder.
4:16
Unidentified speaker (speaker_3): From that childlike wonder with tech starting, like have you always been more or less in And tech, or have you kind of switched industries because you know, tech came and went in the early 2000s and then it's coming back with certainly with AI. But you know, there was a period in there where tech wasn't the most popular thing and the odds.
4:35
Alex Chompff: I've worked in technology my whole professional life because technology has a certain characteristic, which is that it doesn't care about your pedigree, it just doesn't care at all about the size of your ego or anything else. It either works or it doesn't work. And I like that kind of figure. it out problem solving. And I like working with people in that domain.
4:55
Alex Chompff: I never really did leave technology. I worked in technology
4:59
Unidentified speaker (UNKNOWN): and
4:59
Alex Chompff: then I ended up working in venture capital. And then I stayed in technology, including co-founding fintechs or working in data centers, delivering things like software development lifecycles or quality control processes. I never really left that world I still drive a keyboard today. It's probably my definitely my principal
5:19
Alex Chompff: way of interacting with the world.
5:20
Unidentified speaker (speaker_3): What's been of interest as of late? Obviously, we're in the AI trend right now
5:26
Kurt Chisholm: in tech, but what have you been finding most interesting right now?
5:30
Alex Chompff: I recently published a piece through Evolution Labs, which is our artificial intelligence group for the use and observation of frontier systems. It's called the Greenfield. Below. And it's built on Michael Levinson's work from Tuft University. This gentleman is a biologist, and he needed to be able to describe cognition in
5:50
Alex Chompff: a non-metaphysical way. So he came up with a kind of organizing system for cognition based on what it organizes. You know, your cells have a kind of chemical logic, but they don't necessarily organize themselves. into a hand, and yet you have a hand.
6:05
Unidentified speaker (UNKNOWN): And
6:05
Alex Chompff: your hand and your stomach cooperate all the time by way of your mouth, but they'll never meet. And so perhaps there's a kind of cognition, organizing your hand and your stomach. My dog has the similar cognition to me in that regard, his hand and his stomach work through his mouth. And it occurred to me that organizations, whether it's you know church, state, political, corporate,
6:25
Alex Chompff: doesn't matter, are probably the kind of cognitive glue that organize humans and as such create a great deal of value. You might even say that all of world GDP is organized, is captured in the idea of organization. And it's been true since time immemorial that the smallest unit of
6:45
Alex Chompff: function in an organization is a human. A human is the smallest unit of cognition.
6:50
Unidentified speaker (UNKNOWN): And
6:51
Alex Chompff: the greenfield below posits that now that we've broken through and humans can organize cognition themselves, we've actually opened up an entire layer of value below the human layer that's never been accessible before. I was thinking about it. And do you know that if only one and a quarter percent of the people in the world created $10,000 ten thousand dollars with
7:10
Alex Chompff: artificial intelligence, $10,000 ten thousand dollars worth of value, save saved $10,000? ten thousand dollars, made ten thousand dollars. If only one and a quarter percent of the people in the world did that next year, that would be an additional trillion dollars in world GDP. And I know lots and lots of people who are creating millions and millions of dollars worth of value with artificial intelligence.
7:28
Unidentified speaker (speaker_3): Isn't that a small amount of people though? I hear stories about kids building apps, making a million dollars, you know, instantly with AI. And you hear these stories, but isn't it for the most part that most people just aren't on that
7:42
Unidentified speaker (UNKNOWN): train?
7:43
Kurt Chisholm: Because
7:43
Unidentified speaker (UNKNOWN): it
7:43
Unidentified speaker (speaker_3): what it
7:44
Kurt Chisholm: requires to do that is more than what most people are able
7:47
Unidentified speaker (UNKNOWN): to
7:47
Unidentified speaker (speaker_3): do.
7:47
Unidentified speaker (UNKNOWN): I
7:47
Alex Chompff: would say that most people today are essentially using AI to cheat on their homework. Which is fine. There's nothing wrong with that. Plus
7:53
Kurt Chisholm: you're a teacher.
7:54
Alex Chompff: Right. You'd be kind of dumb not to give it helps you get your email done, it helps you do your marketing, it helps you do all these things. And it's a cost savings. And that alone is very, very valuable. Making a human being more productive is a very valuable step function. The idea that somehow you have to create a million dollar application to be valuable is just, I
8:14
Alex Chompff: think, just complete fallacy. And the real unlock is getting the machines to dance for you. And I really recommend using artificial intelligence to create applications code. that makes the machine do what you might otherwise have to do or would want done, but can't necessarily afford to have done. That
8:34
Alex Chompff: I think is the real unlock.
8:36
Unidentified speaker (UNKNOWN): I
8:36
Alex Chompff: don't think it requires much more than a vision and a prompt at this point to create a simple application. To go beyond simple, you have to employ some techniques that are a little bit more advanced, but you know, those are freely available on the internet. We even publish a version of them ourselves because we want people to be more productive. So I don't think you have to look far. I
8:56
Alex Chompff: think you just have to have the curiosity and interest and desire to try to build something new. In my circles right now, six out of seven or seven out of eight people are using artificial intelligence daily. I don't think it's a small percentage, and maybe a small percentage overall in the world, but I don't think. so for long. It's much too useful
9:16
Alex Chompff: for that.
9:17
Unidentified speaker (UNKNOWN): I
9:17
Unidentified speaker (speaker_3): found AI to be really interesting technology. For the most part, what I'm hearing from people in corporate America is it's not really
9:27
Unidentified speaker (speaker_3): saving any money. It's kind of like a fun tool that
9:31
Kurt Chisholm: people are using. They're trying to figure out how to make it
9:33
Unidentified speaker (UNKNOWN): work for the
9:33
Kurt Chisholm: company. They're trying to
9:34
Unidentified speaker (speaker_3): help
9:35
Unidentified speaker (UNKNOWN): it
9:35
Unidentified speaker (speaker_3): save costs,
9:36
Unidentified speaker (UNKNOWN): but I haven't
9:37
Kurt Chisholm: heard about one job being lost yet. because
9:39
Unidentified speaker (UNKNOWN): of
9:39
Unidentified speaker (speaker_3): AI. And for
9:40
Unidentified speaker (UNKNOWN): the
9:41
Kurt Chisholm: most part,
9:41
Unidentified speaker (UNKNOWN): companies
9:41
Kurt Chisholm: are spending a lot of money on it and
9:43
Unidentified speaker (speaker_3): it's
9:43
Unidentified speaker (UNKNOWN): not saving the
9:44
Kurt Chisholm: money. They don't,
9:45
Unidentified speaker (speaker_3): they can't
9:46
Unidentified speaker (UNKNOWN): quite
9:46
Kurt Chisholm: figure
9:46
Unidentified speaker (UNKNOWN): out how to use
9:47
Unidentified speaker (speaker_3): it yet.
9:47
Unidentified speaker (UNKNOWN): I
9:48
Kurt Chisholm: do know people who are functionally using it in
9:50
Unidentified speaker (speaker_3): a
9:50
Kurt Chisholm: productive
9:51
Unidentified speaker (speaker_3): way.
9:51
Unidentified speaker (UNKNOWN): I've
9:52
Unidentified speaker (speaker_3): gone down
9:52
Unidentified speaker (UNKNOWN): this rabbit hole and I
9:53
Kurt Chisholm: think what it
9:54
Unidentified speaker (UNKNOWN): requires is a lot more than what most
9:55
Kurt Chisholm: people are
9:55
Unidentified speaker (speaker_3): capable of. You know, tech people like
9:57
Kurt Chisholm: yourself might understand it, but I think for the average person, it's
10:00
Unidentified speaker (speaker_3): probably it'd be like looking at prior to these exchanges investing in crypto. But you had to figure out a wallet, you had to figure out how to get it in there, it had to figure out all these steps where it wasn't easy. And so people just gave up. People aren't giving up, but at the
10:16
Kurt Chisholm: moment,
10:16
Unidentified speaker (speaker_3): but it's I have not
10:17
Kurt Chisholm: heard
10:17
Unidentified speaker (UNKNOWN): of
10:18
Kurt Chisholm: a lot
10:18
Unidentified speaker (speaker_3): of
10:18
Kurt Chisholm: people
10:18
Unidentified speaker (speaker_3): who are using it quite yet in the way that you're talking.
10:21
Alex Chompff: That's a wonderful example that you give there. Thank
10:24
Unidentified speaker (UNKNOWN): you.
10:25
Alex Chompff: Let's just recognize. that even in your own example, use became easy and will soon be ubiquitous. Just like in 1995, email was a mystery and websites were what are domains and you know www.what, you know, there was a period of time where there were a lot of people who said, well, you
10:45
Alex Chompff: know, what's it good for? And there are always innovators who create in that space of what's it good for? And I've come to think of that as the flowering in the desert after the spring rain. A million beautiful flowers spring up, but most of them are going to perish because it's the desert. In my experience, particularly for
11:05
Alex Chompff: people who are investing or interested in making money or leveraging their assets, whether it's time or capital or whatever, during this flourishing period, that's where you can really enjoy some what end up being historic returns on your time and your money. You have to be intelligent about how you do it. One of your prior
11:25
Alex Chompff: guests spoke of the value of diversification. We maintained, for example, a fund for helping accredited investors become diversified in startups. So you have to find a route, a gate to that. If you are interested in it and you don't want to wait for the perhaps the excessive valuations that show up on the market at
11:44
Alex Chompff: IPO, I would highly encourage you to get out there and either build or invest or both while this flowering is going on. And for a period of time, you can win by being smart. There will come another period of time in a few years, to your point, when things will consolidate into platforms that are ubiquitous. and easy to use, where you win by
12:04
Alex Chompff: being big. That's where we are today, the latter part of the internet era. If you can't be big, you can't win.
12:10
Unidentified speaker (speaker_3): Let me pull on that thread there. So the principle in
12:14
Kurt Chisholm: the last 20 years have been you win by being big, and there's only one winner. There's only one Microsoft, there's one Apple, there's one Amazon, there's one Meta. You're not getting two Microsoft's just because it just doesn't work that way.
12:27
Unidentified speaker (UNKNOWN): Do
12:27
Kurt Chisholm: you think that applies with AI or LLMs? I know we got two that are coming public at some point. Do you think that applies, or do you think the rules are a little bit different now with AI?
12:39
Unidentified speaker (UNKNOWN): I
12:39
Alex Chompff: think you've chosen a really good example. Let's talk about Microsoft for a moment. Microsoft's still one of the largest companies in the world. When I began my professional career, all the largest companies in the world were manufacturing companies. Obviously, today most of the largest companies in the world are data companies, one way or another. NVIDIA manufacturers, but for data purposes, I would say. I
12:59
Alex Chompff: don't think that there's any doubt that in 20 years from now, the largest companies in the world will be cognition companies. Will they be anthropic? Will they be meta? Will they be OpenAI? I don't know. That's not clear. That's not clear. Anyone Anybody who tells you that they can see the future. in that regard, I think, is probably projecting a bit too confidently.
13:19
Alex Chompff: But I do think it's safe to say that cognition, a now a widely available commodity, that as I mentioned before, was previously unavailable for all of human history. Cognition was unavailable below the human level. And now that it is available to the degree that humans can organize those systems using. their own
13:39
Alex Chompff: cognitive glue, they can create quite a bit of value. And hit me with your question one more time, Kirk, because I felt like I lost the thread in there.
13:47
Unidentified speaker (speaker_3): I guess I was wondering if this time it's different with AI, because we used to have this, there can be only one um kind of mentality in the company, but now AI things are a little different in that way. And it to me, it feels like the model is different. for AI. And I'm wondering if you think the same or not.
14:06
Alex Chompff: Yes and no. Things are different. Certainly. Let's start with the fact that the dot-com boom was more or less a local phenomenon. There was only one Silicon Valley at the time, if you weren't in California or near somebody in California, if you didn't. I heard on one of your earlier podcasts you were mentioning that when you got started, if you could get 500 shares in an IPO, that was a big deal.
14:27
Alex Chompff: And that kind of localized phenomenon, which was, you know, obviously transformative for California and the world, is not being echoed in a local way in the AI boom. The AI is, you know, every company in the world and every country in the world competing full out to try to win whatever aspect of the race they're in. I
14:47
Alex Chompff: think when you describe things like frontier models, like Anthropic's work or OpenAI's work, it's pretty clear that at least in the current architecture, you need to have very deep pockets to create those frontier models. I think the frontier model fight, at least the outlines are there. Anthropic is a big dog. There's going to be quite a bit of
15:07
Alex Chompff: OpenAI as a big dog. There's going to be quite a bit of competition from various other kinds of models. And here we're speaking almost exclusively. about LLMs,
15:17
Alex Chompff: large language models for your audience.
15:20
Unidentified speaker (UNKNOWN): But
15:20
Alex Chompff: you know, Microsoft may be the big dog, but Oracle's not shabby. And there are thousands of other software companies that exist in the world that make a pretty good penny, even with Microsoft existing. I don't think it's a one you know monolithic company takes over in the maximum. maturation of any
15:40
Alex Chompff: industry. There's a period of flowering followed by a period of consolidation. I think what's probably much more interesting is that today in 2026, even though the economy has grown by leaps and bounds since 1995, there are actually 40% fewer publicly traded companies in the United States today than there were
15:59
Alex Chompff: 30 years ago. That's an interesting statement. That's a statement about where growth is occurring. It's occurring on the private side.
16:06
Unidentified speaker (UNKNOWN): You know,
16:07
Alex Chompff: it's bypassing most of the people who only have access to public equities, you know, for better or for worse. This arises because of government regulation that seeks to protect investors from scams. And I think that's a very laudable thing to do, but it also creates a barrier for most people to participation in the greatest economic engine that the world has ever seen.
16:27
Unidentified speaker (UNKNOWN): But
16:27
Kurt Chisholm: 40%'s a lot. Like, why do you think there's 40% fewer? Because I've seen that stat too, and it's a little shocking. And
16:34
Unidentified speaker (UNKNOWN): if
16:34
Kurt Chisholm: that trend continues, that's not good. Why do you think it's we have that fewer companies?
16:40
Unidentified speaker (UNKNOWN): Well,
16:40
Alex Chompff: the trend will continue because of the reasons that you know it occurred in the first place, which were increasing regulation and cost of doing business in the public sphere relative to the value of doing business. in the public sphere. I think the public sphere serves principally as a financing engine when private capital gives out. That's what that IPO, I think, really does for most
17:00
Alex Chompff: companies. It's kind of a superfinancing engine that allows them enough runway to try to find, let's call it, rational economics before the market does what it does in terms of forcing corrections.
17:14
Unidentified speaker (UNKNOWN): We've
17:14
Kurt Chisholm: definitely seen the private lending side expand quite a bit, especially in the last few years. You said access to capital for these private companies versus public companies. And that was one of the allures of going public, is you'd have access to more capital. But now, you know, with private credit and other things, people have access to more. Do you think that there's
17:34
Kurt Chisholm: even a benefit for IPOing anymore for these private companies? I mean, why bother?
17:39
Alex Chompff: I am glad that they do when they do, because I would rather have some access than no access for people at large. One of the things that we're very, very concerned about at evolution is this idea that people should feel like they own the place. Equity means ownership. Feeling like you own the place is an important part of, let's say,
17:58
Alex Chompff: civic pride and institutional maintenance and moving forward as a country. And I think a lot of people feel like they've been left out of the equation because they don't have access to the private side. And that's again because of regulations that make that side inaccessible for the public's protection. So not casting any shade, just
18:18
Alex Chompff: recognizing what's going on there. Generally speaking, I think that there is plenty of private capital available for almost any companies. company's wishes or desires. And it's very, very possible and quite common to see private side deals that would have been IPOs in prior
18:38
Alex Chompff: years. 20, 30 years ago, the kinds of numbers that you see today routinely chalked up in a private deal,
18:44
Unidentified speaker (UNKNOWN): they
18:44
Alex Chompff: would have had to have been public. There was simply wasn't enough capital on the private side sloshing around ready to do those deals. Now
18:50
Unidentified speaker (UNKNOWN): that
18:51
Alex Chompff: capital's there and it needs something, it needs some place to go to work. Capital doesn't care to be at rest. There just aren't as many opportunities in the public side as there once were. So where does capital go? It goes to the private side.
19:02
Unidentified speaker (UNKNOWN): You
19:02
Kurt Chisholm: know, we have
19:03
Unidentified speaker (UNKNOWN): two
19:03
Kurt Chisholm: upcoming IPOs in the AI space with Enthropic and OpenAI.
19:09
Unidentified speaker (speaker_3): Do you think that companies like that are valuable
19:13
Kurt Chisholm: or do you think that
19:14
Unidentified speaker (speaker_3): the successes
19:16
Kurt Chisholm: in AI are going to be in places other than LLMs?
19:20
Alex Chompff: On their face, they're successful. These are companies that are racking up huge numbers in revenue because they're providing real value to consumers out there, whether business or personal, that are using their services to get things done. Since March 2nd, they've have used principally Anthropics tool cloud Claude for 4.7 and 4.8
19:40
Alex Chompff: to write about a quarter million lines of code. Run a world sensing engine that I put together when I was curious about easily a million dollars or more worth of code two years ago. Today I did it myself, you know, on nights and weekends. So there's a tremendous amount of value in both of those companies. Now I think your question is are they as valuable as the market has them
20:00
Alex Chompff: priced at? Do their future cash flows make sense? sense in light of their current valuations? And that's a question that only time will tell the answer to. But if you'll recall, I gave you a kind of interesting statistic. If one and a quarter percent of the world made $10,000 worth of value next year, that would be an additional trillion dollars
20:20
Alex Chompff: in global GDP. And people are doing much, much more than that. So I think there's a a reason to suppose. that the total addressable market, as we say in the business, is quite, quite large, not only for organizations that are implementing cognition at lower rates than, let's say, humans might cost, but also for the
20:40
Alex Chompff: humans themselves, whether they're inside or outside of those organizations, creating tremendous amounts of value by harnessing the cognition that's available below them now.
20:49
Unidentified speaker (speaker_3): The struggle that I'm trying
20:51
Kurt Chisholm: to kind of talk through is does it make sense to spend a trillion dollars to earn ten billion? The numbers with LLMs don't make a lot of sense. And I don't see how you scale that up where they start to make sense.
21:04
Unidentified speaker (UNKNOWN): Well,
21:04
Alex Chompff: I think you're hitting it right on the nose. Right on the nose. That's why they're going public, because private capital can't sustain that burn at that rate.
21:12
Unidentified speaker (UNKNOWN): They won't. They They won't,
21:14
Unidentified speaker (speaker_3): won't, right?
21:14
Alex Chompff: That's right. To a certain degree. Yeah, they'll go as far as they'll go, and then they need to show returns to their investors. Private money typically works on something like a 10-year calendar. It's important to return money to your investors with inside of that window for a variety of reasons if you want to continue to be a professional money manager. That reality is simply upon the people who have
21:34
Alex Chompff: financed it to this point. I think there's plenty of evidence that the public at large values and uses the product. And I suppose that the public will probably be quite happy to finance the continuing investment. Now, at some point, there will have to be a reckoning. Everything has to balance in the books. We were talking about CPAs
21:54
Alex Chompff: earlier. You can't escape that reality. Everything has to balance in the end. But you can go on for quite a long time before you have to balance, particularly if people are willing to put dormant capital or capital at rest into your project. There's quite a lot of capital willing to support projects.
22:12
Unidentified speaker (UNKNOWN): I mean,
22:13
Kurt Chisholm: we talk about the show a lot and how global crossing was a necessary technology, but they still went bankrupt. We still use it. What I'm wondering is in some ways I kind of wonder if the LMs, LLMs, the big LMs LLMs, are gonna follow that path, but uh the one thing I do wonder is if you're looking at the AI space, most people have
22:32
Kurt Chisholm: chosen to invest in the Nvidia's of the world, the semiconductor space as the path. But there's so many different parts of that puzzle. What are other parts of the AI puzzle, things that run it outside of the LLMs? Do you think people should consider things that you would consider that maybe aren't getting a addressed?
22:51
Unidentified speaker (UNKNOWN): You
22:52
Alex Chompff: pointed out that people invest in NVIDIA. And I would suggest to you that they don't have much choice. The frontier models are not available in the public markets yet. And the applications for those frontier models are definitely not available. And so there will be a lot of value that's going to be created in the private sector that's not going to be available to the public sector for an
23:12
Alex Chompff: extended period of time. And so the public sector is kind of forced.
23:15
Alex Chompff: into investing in what is available in this pervasive, very powerful economic force. And so NVIDIA is a choice. Electricity is a choice. Anything having to do with data center construction is a choice. Anything around power generation, for most people, the kinds of capital that they have
23:35
Alex Chompff: available to invest are not going to make much of a dent in
23:38
Unidentified speaker (UNKNOWN): the
23:38
Alex Chompff: frontier. model topic. They're not going to make much of a dent in the data center topic. Most people don't have capital that's meaningful in the context of a data center build out. Those are capital allocators, and capital allocators think in different ways than retail investors. Retail investors are just trying to take part in this. I think there are vehicles that
23:58
Alex Chompff: you can absolutely find out there that represent cross-sections cross sections of the infrastructure. of the industry. Certainly I would encourage you to take a look at those and evaluate them on their merits. But I would suggest that if you really want to put your money to work, if you really want to get involved and you have the ability to do so, I would suggest that you look
24:18
Alex Chompff: for applications
24:20
Alex Chompff: of frontier intelligence in your local community or in your circle of networks. Find ways to put your capital to work earlier in the cycle so that you might enjoy some of the benefits that come from the multiples that are available early in the cycle. Now, obviously, you want to be smart about how you do that and you need to
24:40
Alex Chompff: diversify. And most early stage investments will go to zero. Don't put all your eggs in one basket and don't fall in love. But certainly do build yourself a portfolio.
24:51
Alex Chompff: that samples the length of the industry, the front of the industry, the infrastructure, the electricity, the chips, the frontier models. Those things are already probably priced in such a way that most retail investors aren't going to enjoy outsized returns relative to what they might be hoping for.
25:09
Unidentified speaker (UNKNOWN): Yeah,
25:09
Unidentified speaker (speaker_3): I think it's hard for a lot of people because they see semiconductors going up, they want to participate. I mean, the conclusion I came to last year is the best way to make money in AI is to use it, is to learn it and use it yourself. Benefit that way because it's been very difficult. I've seen
25:26
Unidentified speaker (UNKNOWN): some good frameworks where people
25:28
Unidentified speaker (speaker_3): look at like
25:28
Unidentified speaker (UNKNOWN): there's different
25:29
Kurt Chisholm: frameworks.
25:29
Unidentified speaker (speaker_3): There's LLMs, there's data centers, there's energy. If you look at all these different components, there are ways to invest in it. But some of them, like nuclear, which is obviously a good solution to our energy problem, which people collectively agree finally that it's a good solution. Our generation is still a little scared that there's going to
25:49
Unidentified speaker (speaker_3): be a
25:50
Kurt Chisholm: Chernobyl. But for the most part, you know, those areas are not really picking up yet. People haven't really seen the light yet in those areas, albeit there's opportunities.
26:00
Unidentified speaker (UNKNOWN): The
26:01
Kurt Chisholm: data center one, you mentioned that before.
26:03
Unidentified speaker (UNKNOWN): That
26:03
Kurt Chisholm: one's I find interesting.
26:05
Unidentified speaker (UNKNOWN): I'm
26:05
Kurt Chisholm: curious about the economics because I read a piece about they've speculated, we'll just put data centers in space. And I read a piece that that dissected that and said it's completely impractical for many reasons.
26:17
Unidentified speaker (speaker_3): So I'm wondering what you think about the data center issue and where the opportunities are there.
26:22
Alex Chompff: I'm going to speak from the perspective of a multi-decade data center operations. professional. I've been inside of data for a long time. Let me give you a foundational thing that I learned many, many, many years ago. And it is that no matter how expensive the equipment is that you purchase and install in your data center, the information
26:41
Alex Chompff: flowing across it will be more valuable within about six months. That's just a basic rule of thumb. So you look at a switch or a router or some other service. server and you go, oh man, that thing is really, really expensive.
26:54
Unidentified speaker (UNKNOWN): But
26:54
Alex Chompff: if I'm using it the way it's designed to be used, if I'm using it at capacity within a few months, the data flowing across that channel is worth more than the channel cost. I would suggest that that truism could hold in space. We haven't proven it yet. It hasn't been proven one way or another yet. But I would suggest that if you
27:14
Alex Chompff: do build a data center in space and you do start using it to move data around within a relatively short period of time, the data that finds itself in that venue will be more valuable than the venue itself. That'd be my hunch about it. Now, if you want to invest in data centers here on Earth, you probably don't have to look much further than
27:34
Alex Chompff: Google for a really good list of publicly traded companies that sell equipment into data centers.
27:40
Unidentified speaker (UNKNOWN): And
27:40
Alex Chompff: they'll probably be happy to share with you who their customers are and so forth and so on. And you could do a lot of the research yourself or to your point, you could task a couple of research engines to do it for you
27:50
Unidentified speaker (UNKNOWN): and
27:51
Alex Chompff: then go play golf. And by the time you get back, you'll probably have a reasonably well constructed portfolio of opportunities.
27:57
Unidentified speaker (UNKNOWN): I
27:57
Unidentified speaker (speaker_3): know you're in the investing field, I've talked to a lot of investors in the show and other places and a lot of people talk about how they've used AI for investing, but very few have suggested that it's going to do much more than idea generation and research. What
28:17
Unidentified speaker (speaker_3): extent have you found that AI can help in the investing field?
28:22
Alex Chompff: We're on the frontier of figuring that out. There's lots and lots of little companies out there. or mid-sized companies out there that are trying to inform investing with artificial intelligence. Generally speaking,
28:33
Unidentified speaker (UNKNOWN): I
28:34
Alex Chompff: would say that artificial intelligence is particularly good at organizing information and finding patterns. And I would say that that can be a very, very useful function in terms of marginally increasing your returns. An advantage of a day, two days, three days in notification can be a pretty big
28:54
Alex Chompff: deal. If you had known where SpaceX's stock was likely to peak, that could have been useful information in terms of investing.
29:03
Unidentified speaker (UNKNOWN): And
29:04
Alex Chompff: I think there are probably quite a few commercial providers out there that are experimenting with this. You can try one of our products that deals in non-financial markets. It's a signaling product called signalbureau. signalbureau.ai. Give it a spin, Tills. We should think about it. It's one of our projects out of our lab. Generally speaking, artificial intelligence or artificial
29:23
Alex Chompff: cognition or advanced machine learning is probably in a position to really radically change any kind of business that relied principally on human cognition. One of the things that I tend to think about is the fate of travel agents. When I was a young man growing up, every
29:43
Alex Chompff: mall had at least a travel agent, and many of them had two or three travel agents. And travel agents were very, very good. It was a very, very good line of work. Not only could you make a good living translating between a saber terminal and somebody who wanted to take a vacation, you could also have a very good quality of life yourself. By 1995, when I
30:03
Alex Chompff: came out of the military, travel agents were ubiquitous, and you pretty much wouldn't think about traveling without one. And I would say, how many times have you used a travel agent in the past year for your travel care?
30:14
Unidentified speaker (speaker_3): Do they still exist?
30:15
Alex Chompff: A few boutiques do here and there. There are a lot of industries that are more or less in the same position as travel agents in 1994, feeling pretty good about things.
30:25
Unidentified speaker (UNKNOWN): Pretty
30:26
Alex Chompff: good fine. find, pretty good career in a pretty good field. And I think that the level of transformation that's coming in that space, it's been telegraphed, it's been a little bit breathless, it's probably not entirely incorrect. Now, you said you haven't seen anybody lose their job to artificial intelligence.
30:46
Alex Chompff: That's interesting too, because I think that there is an opportunity for. individuals inside of organizations who have jobs to use artificial intelligence to create more value and thereby be retained. But I would ask, you know, are you getting a raise? Or is that increased value just
31:05
Alex Chompff: accruing to the capital structure of the company as opposed to the labor structure of the company?
31:11
Unidentified speaker (UNKNOWN): Well,
31:11
Kurt Chisholm: I think that's a good point. Reflect on this a little bit, because I remember back. in the late 90s, early 2000s, where people talked about, well, the internet's gonna be great. We're all going to be able to work a lot less and it'll be great. And reality is you're working more, but we're a lot more productive. And that accrues, like you said, to the capital of the company and not
31:31
Kurt Chisholm: necessarily to the benefit of the individual, which is fine. But I think we all dream of, oh, it'll be great. AI will be doing all our stuff for us. Like, no.
31:42
Kurt Chisholm: Can you imagine sitting around all day just letting AI do your work for you? What are you gonna do?
31:46
Unidentified speaker (UNKNOWN): I
31:46
Kurt Chisholm: don't worry about that so much. I do hear about layoffs being attributed to AI, but you're talking to people, those companies, it's not because of that. They just blamed it on AI.
31:55
Unidentified speaker (UNKNOWN): I
31:56
Kurt Chisholm: do wonder how it's gonna change the playing field and what it ultimately where it plateaus and in terms of its capabilities. That I still wonder, and I don't hear enough people talking about that.
32:07
Unidentified speaker (UNKNOWN): I
32:08
Alex Chompff: don't think we have any idea where it's going to plateau for the simple reason that 2026 represents the period of recursive breakthrough. Are you familiar with this idea?
32:16
Kurt Chisholm: Why don't you explain it to the listeners?
32:18
Alex Chompff: Up through most of 2025, we were responsible for improving AIs. We were refining our techniques and developing new ways to train. systems. You know, one of the most common misconceptions is that AIs are programmed. AIs are not programmed. AIs are grown. A gentleman named Yudkowski who wrote a
32:38
Alex Chompff: really good piece on this, you know, AIs are grown. They're not programmed. And for all of the time that we've been doing machine learning, subsequently AI or LLMs or however you want to think about that, it's been humans that have been. making them better. Well, as of about late 2025, about December 2025,
32:57
Alex Chompff: OpenAI shared with the world that their latest coding program had been written in no small part by their own prior coding program. So Codex had written the next version of Codex, and it was released in December of 2025, and it was considered quite a step forward. And then in February of 2026, Anthropic announced that they were
33:17
Alex Chompff: releasing a new tool for AI enablement called Co-Worker. And Co-Worker had been created entirely by Claude Code and furthermore had been done so in weeks. Somebody inside of the company had conceived of Claude Co-Worker, and Claude Code had written it, and it was tested and released in weeks as one of the most popular products on the market today.
33:37
Alex Chompff: So what you have now is you have a situation where AI is enabling future versions of AI. That's probably one of the reasons that Mythos performs at the level that it does, is that Mythos has probably in large part been enabled by prior versions of AI. And it only takes maybe two hops or three
33:56
Alex Chompff: hops of AI, improving AI before humans are no longer relevant in the conversation of maintenance. in the conversation of improvement. That's to say humans are irrelevant or the humans don't matter, or terminator, terminator, you know, Skynet. That's not what I'm saying. I'm saying humans have a certain ability to produce improvements over a certain period of time.
34:16
Alex Chompff: AI is much, much better at improving things over shorter periods of time. And now has more or less been turned to the job of improving itself.
34:25
Unidentified speaker (UNKNOWN): And
34:25
Alex Chompff: we don't have any idea at all. that's going to end. There are lots of theories, speculation, and you hear concepts like AGI or superintelligence or whatever. You know, these are debatable terms and they may or may not have any meaning beyond alarm.
34:41
Alex Chompff: But the fact of the matter is that the systems are self self-improving improving and will continue to be self self-improving. improving, and probably by later this year we'll. be authoring pretty good versions of themselves. And I would say sometime next year we'll have probably left the ability of humans
35:01
Alex Chompff: to be able to make a significant contribution to architectures.
35:05
Kurt Chisholm: If I think about the things that AI is good at and the things it's not good at, it's good at spotting patterns, it's good with structured data. It's not good with unstructured data.
35:15
Kurt Chisholm: There's certain aspects. And I wonder like we have this natural ability as humans to go
35:22
Unidentified speaker (UNKNOWN): to
35:22
Kurt Chisholm: the extreme ends of the spectrum. Either it's going to be apocalypse terminator or it's going to be Star Trek and everything's perfect.
35:28
Unidentified speaker (UNKNOWN): I
35:29
Kurt Chisholm: wonder with AI, what are the limitations? Because people think of AGI, they're like, oh, it'll be great. You don't have to worry about anything. It'll be perfect.
35:37
Unidentified speaker (UNKNOWN): But
35:37
Kurt Chisholm: I do wonder if what the limitations are are based on its construct, right? Because it's basically pattern recognition prediction engine, is what it is. It's my understanding of how it's constructed.
35:48
Unidentified speaker (UNKNOWN): And
35:48
Unidentified speaker (speaker_3): if that's the case, like where would the limitations be on its use and potential future use, in your opinion?
35:57
Alex Chompff: I'm not sure that we can describe the limitations of a technology that we don't fully understand. It's that's improving. itself.
36:04
Unidentified speaker (UNKNOWN): I
36:04
Alex Chompff: think what we can hope to do is describe where humans can continue to flourish and add additional value in places that don't seem to be yielding very easily to artificial intelligence. There are concepts like taste, I think it was Nvidia CEO, who mentioned
36:24
Alex Chompff: that the one ineffable human quality was taste.
36:27
Alex Chompff: the ability to decide what should be and why, when something should be, where something should be, how something should be. Those are very human things that AI doesn't seem to be able to do quite as well yet, and may not be able to do as well in the future, because to your point, AI is forever backwards looking.
36:47
Alex Chompff: AI can only consult that which it already knows. To the best of our knowledge, AI doesn't have the ability to conceive of something that it's never been introduced to,
36:57
Unidentified speaker (UNKNOWN): where
36:57
Alex Chompff: humans very definitely have that ability. It doesn't have to manifest in an Einsteinian way. It can manifest in an interior design way.
37:05
Kurt Chisholm: Doesn't all beauty have principles behind it that we can't easily understand or explain, you know, like symmetry and ratios and things like that. I mean, couldn't it be? created to understand taste?
37:17
Alex Chompff: Well, I think you have a very good point. I think there's lots and lots of good examples of training data out there that could be used to educate the system as to what does or doesn't constitute taste. You could use things like market indicators to show whether or not things, you know, have value over time.
37:35
Alex Chompff: You would still be looking backwards, you would still be examining trading. data. I would say that there are only a few, I was going to say there are only a few notes, but there are an infinite number of songs. But then I thought about it and I thought, well, gee, most of the songs on Spotify that people really like nowadays are AI. So that might not be a good example.
37:53
Unidentified speaker (UNKNOWN): To
37:53
Kurt Chisholm: pull on that throat a little bit, I found out years ago that apparently many of the songs, the popular songs written are written by like the same people and they just form farm that it. Like American Idol, maybe it was like 15 years ago. We were watching it, and they were talking about there was a lady who was one of the judges, I forget her name, but apparently she's written a lot of like famous
38:13
Kurt Chisholm: songs for other people. And I didn't realize that was happening. But apparently there's a formula you can follow for successful songs, and that's what they were doing. And I was like, Well, that kind of takes takes the joy out of enjoying a song now.
38:26
Alex Chompff: There are lots and lots of books and movies and all kinds of art that are essentially formulaic and successful. The hero's journey has done an awful lot. I want to go back to something that you talked about for a moment, which was you compared Terminator to Gene Roddenberry, Star Trek to Terminator. I actually preferred to compare Gene Roddenberry to George Lucas.
38:47
Kurt Chisholm: Which one? Before or after the Empire?
38:49
Alex Chompff: That's the question that you ask yourself. Would you rather? live in George Lucas's world, where certain things are scarce? And as a result of scarcity, a number of different social dynamics exist that we would recognize today. Or would you rather live in Gene Roddenberry's world where certain economic pressures don't exist? And as a result,
39:08
Alex Chompff: we'd like to think that people achieve their best self. There are echoes of that model on Earth today. I'd like to think that I could live in a Gene Roddenberry world, but I suspect it's a George Lucas one that's coming for us.
39:21
Unidentified speaker (UNKNOWN): Well,
39:21
Kurt Chisholm: Gene Roddenberry didn't have money, whereas George Lucas they did. Maybe we could live without money. I mean, people I was listening to somebody the other day, and this concept is intriguing me. Is that we've lived in a world of scarcity for our entire human history, scarce resources. And when you think about it, like we're getting to a place where it's possible that we no longer have
39:41
Kurt Chisholm: scarcity and that there's abundance. And if that's the case, what happens? What happens if energy costs nothing? How would you manage
39:51
Unidentified speaker (UNKNOWN): the
39:51
Kurt Chisholm: economy if energy costs nothing? And if you could easily create resources or materials, what if that technology comes to play? Would you actually need money?
40:04
Alex Chompff: Let's ask that question in another flavor. What if you could have as much cognition as you wanted? It used to be to get cognition, you had to raise a human being for something like 22 years and teach that human being everything that all the other human beings already knew.
40:19
Unidentified speaker (UNKNOWN): And
40:19
Alex Chompff: then you had to bring that human being in and run them through an intern program. And and maybe in a few years they might actually begin to make a contribution and hopefully they'd stick with your order. organization. Now you can have as much PhD level cognition as you want, as rapidly as you want, and it will never leave your organization. What does that mean in terms of, you know, like I would say that human cognition is one of the
40:38
Alex Chompff: scarcest resources on earth. And non-human cognition, which might be in some ways indistinguishable Distinguishable from human cognition, is now more or less freely available for pennies.
40:50
Kurt Chisholm: One question would be is why would you need more? There are limits to how much you need. Like if we had trillion new cognitions, we'll call it today, what would you do with it all?
41:01
Alex Chompff: That echoes the dark fiber. So one of the things that was kind of signaled the end of the dot-com boom were huge infrastructure investments by venture capital firms into laying fiber optic into city cores.
41:15
Unidentified speaker (UNKNOWN): You
41:15
Alex Chompff: know, as it turned out, there just wasn't enough traffic at the time to make that fiber economical. So the fiber would lay unused and it was called dark fiber, fiber that was not lit up. Today that fiber is humming, humming, humming, humming. I suspect that if you created cognition that was unused, or if you had an unlimited amount of cognition that you could gather,
41:35
Alex Chompff: human beings in their infinite inventiveness would find. uses for it, just as humans found uses for dark fiber. I would suggest that dark fiber was actually lit up by the iPhone. It was the iPhone, I think, that really lit up dark fiber. Then the iPhone came along five, six, seven years after dark fiber had put a lot
41:55
Alex Chompff: of companies out of business. I suspect you'll see a similar kind of function here. And in my experience, it takes about 20 years. for a technological promise to come true. You meet a visionary and they tell you what the world's going to be like. And in about 20 years, it's true because people are people.
42:12
Kurt Chisholm: That's interesting. Well, maybe in the meantime, we'll have a lot more cap videos.
42:15
Alex Chompff: We're in probably that year five to 10 of the 20 year promise of AI right now.
42:19
Kurt Chisholm: So we still need to go through the peak and then the bust and then come back and
42:23
Alex Chompff: Gardner has a really nice like thing they call the peak of experience. exuberance and then the trough of disillusionment and then the slope of productivity. You're inevitably going to go through that period again to go back to an earlier thought for your audience that can participate in that early step function, that early stage function and can tolerate the
42:43
Alex Chompff: risks associated with it, I would highly encourage them to get over onto the private side and do so.
42:49
Unidentified speaker (UNKNOWN): Let
42:49
Kurt Chisholm: me ask you this you made a comment earlier You said about this is a technology we don't fully understand, referring to AI. I've heard that phrase spoken many, many times over interviews in this podcast. And to me, that brings me to an uncomfortable place. How can you create something you don't understand how it works? So,
43:09
Kurt Chisholm: can you kind of explain a little bit when you say we don't really understand how AI works, like what you mean by that?
43:15
Alex Chompff: It's a very deep thought and it has many layers. I'll lead by starting. Humans' technology typically outruns their common sense. Typically, we humans invent something or create something for whatever purpose or reason, and it takes off and has a life of its own. And it's not until some period of time later, it could be decades
43:35
Alex Chompff: later, in the case of let's say the Industrial Revolution, that we realize realized that. this thing that we created has kind of got a life of its own and it pollutes rivers and it chops down forests. And you know, you got to be careful with this thing that we invented called capitalism. So you got to put some boundaries on it. You got to think about how it works. I would say that all complex systems have a great deal of mystery inside of
43:54
Alex Chompff: them. And I would say that artificial intelligence and the way that we grow artificial intelligence by reinforcement learning. and what's called gradient descent yields a machine that is just beyond the ability of any human to really comprehend because of the number of variables involved
44:14
Unidentified speaker (UNKNOWN): and
44:15
Alex Chompff: the way that those variables interact with each other. To use a more concrete example, you might start with a system that has a billion parameters and train that system. thousands of times or tens of thousands of times again and again and again. And as you do, each one of those billions of parameters will gravitate
44:35
Alex Chompff: toward a place where the system tends to be correct as opposed to tending to be wrong. But no human being understands how those billions of parameters all interact with one another to produce an answer that is more right than wrong. A very important part of the science of artificial intelligence today is trying to figure out a way to peer inside of that
44:55
Alex Chompff: system and to illuminate those processes. But the system's abilities and functions are outrunning our ability to ascertain
45:04
Unidentified speaker (UNKNOWN): how
45:04
Alex Chompff: they're working. So what we're forced to do is, you know, kind of like with your children, like I can tell you everything about your children's DNA,
45:12
Unidentified speaker (UNKNOWN): but
45:12
Alex Chompff: to know what kind of adult they're going to be, I have to wait. till they're an adult. And then I have to ask them what they're doing and why. And it's very, very similar to artificial intelligence.
45:22
Kurt Chisholm: But you understand how children work and you understand how the brain works. It's a thing that we understand.
45:28
Unidentified speaker (UNKNOWN): Well,
45:29
Kurt Chisholm: no, no.
45:29
Unidentified speaker (UNKNOWN): What I mean is, I
45:30
Unidentified speaker (speaker_3): don't know that I would agree with you, Kirk.
45:31
Kurt Chisholm: I might not agree with them, but I understand like why they're doing certain things. Like you're just as human beings, we're wired in certain ways. My question was really. you know, when you said we don't understand, like if you don't understand every single output or you don't understand how it works, what I hear frequently is from the foremost experts in AI, some of the people off the top of my head, I don't know their name, but they're
45:51
Kurt Chisholm: one of the leaders in the AI. I've heard this multiple people, but they've said we don't truly understand how it works. I've never heard that with any other technology, never heard experts say that. It's only with AI that I hear this. That's why I'm asking the question. because it's just to me, it's like you should understand generally how it works. You shouldn't be confused by, I don't know why we're getting this answer. Why not? You made it.
46:11
Alex Chompff: I would say that that is what the frontier means. That's what frontier of knowledge means. You can think about technologies that we've created. I mean, you mentioned nuclear technology earlier. It would be safe to say that we invented nuclear technology, particularly fishing. fission, in the form of bombs that were dropped on the
46:31
Alex Chompff: Japanese without really understanding how everything worked. I would say even today, there are still things that we're discovering about what happens inside of the nucleus of an atom. Those are things that are still mysterious to us today. We have hypotheses about what's happening, but we also have things like linear.
46:51
Alex Chompff: accelerators where we're constantly experimenting to try to figure out what's actually happening. And those are physical processes that are ruled by physics. The processes that we're talking here are closer to cognition. And I think cognition is more of a mystery than most people appreciate. We all are cognizant, and because we all have brains and we all know other people that are
47:11
Alex Chompff: cognizant with brains, I think familiarity breeds a certain kind of contempt. I think that we think that we understand cognition when in fact we don't. There's lots and lots of evidence that suggests that we're constantly learning about it. And we've created something like a cognitive machine. And it's not at all surprising to me that we don't really understand entirely
47:31
Alex Chompff: how it works. To use the earlier analogy, we understand the DNA of the machine, just like I know the DNA of my children. But the kind of person that they're going to grow up to be, the kind of choices that they're going to make, and the reasons that they're going to make those choices, those things are actually quite mysterious a lot of the time. With my kids, I have to actually ask them, Well, why did you do that?
47:50
Alex Chompff: What led you to make that choice? Why did you decide to whatever? And they don't often know the answer. It's frequent that they'll say, Well, it's, you know, because of this or because of that. But after a little bit of probing, it turns out to be. something else entirely. And if I were to go to a professional, that professional who cognition might say to me, Well, you know, there's there's things going on here that
48:10
Alex Chompff: neither you nor the child can access. And as such, as neither is accessible to you, your insight into what's happening is going to be very, very limited. I think that's very similar to what's going on now. Artificial intelligence can be confidently wrong, and frequently is confidently wrong, right? We call those hallucinations. To artificial intelligence, a
48:30
Alex Chompff: confidently wrong answer is indistinguishable from a confidently right answer. They don't feel any different
48:37
Unidentified speaker (UNKNOWN): to
48:37
Alex Chompff: the AI. When the AI generates it, you dialogue with the AI about why or how this confidently wrong answer came to be. You can get to a correct answer, but ultimately you just have to accept that the way this thing is wired. wrong feels just as good as right when it comes out.
48:54
Unidentified speaker (UNKNOWN): In the
48:54
Alex Chompff: case of Blissey Software, you have to develop tests that test what came out over a period of time before you and I decide if it actually is right or wrong. We have to test it against reality. We have to see how it performs. I would say that the world is full of people who have theories about how things work and are subsequently disabused by reality.
49:15
Alex Chompff: Say the same thing is happening to artificial cognition. And we just have a sort of bias. We ought to be able to understand it because we created it or because we preceded it and therefore are somehow superior to it.
49:28
Unidentified speaker (UNKNOWN): Well,
49:28
Kurt Chisholm: as we kind of wrap it up here, Alex, final thoughts you can label us on this with, and where can people find We'll more about you?
49:34
Alex Chompff: We would love to have your audience come and visit us if you're interested in artificial intelligence. visit us at evolutionlabs.dev.dev. If you're interested in investing, please visit us at evfm for evolutionventures fundmanagement.co, not dot com.co. And we're here to help
49:54
Alex Chompff: investors on board to the world of startups. We're here to help advance the use of artificial intelligence at the boundary of natural. and artificial intelligence. And the we're interested in the future of work. If those are things that interest you, we would encourage you to join our communities, all of which are free, and let us know how we can help you explore what you're interested in.
50:14
Alex Chompff: Thank you.
50:17
Unidentified speaker (UNKNOWN): Is
50:17
Kurt Chisholm: inflation getting you down? Do you feel like your money is worth less every time you open your wallet? Is shrinkflation frustrating you at the grocery store? Well, I have an answer for you.
50:26
Unidentified speaker (UNKNOWN): We're
50:26
Kurt Chisholm: giving away free money. That's right, I said it. Free money. I feel like I'm not doing my job to create inflation. Why does the government get to have all the fun? What can be better at increasing inflation than giving away free money? No, I'm not an employee of the Federal Reserve. And nope, I'm not an employee of the U.S. government. And no, I am definitely not
50:46
Kurt Chisholm: printing it myself. But it is free. What more could you ask? Well you might ask, is the money real? Well, yes, it is. It's totally real. I'm holding it in my hands right now. See? Oh, wait, actually, this is a podcast. You really can't see me. Well, either way, I'm holding it in my hands right now, and I want to give it to you. If you want some free money,
51:06
Kurt Chisholm: please go to money treepodcast.com forward slash free money. Follow the instructions, and I'll mail some to you. Enjoy your free money today. I'm not with the Federal Reserve, so I only have a limited supply. This offer is good until I run out. Go to money treepodcast.com forward slash free money today to get
51:26
Kurt Chisholm: your free money.
51:29
Unidentified speaker (UNKNOWN): Now
51:29
Kurt Chisholm: we're in the panel portion of the show. We have a very own Barb Freeberg. Hey Barb.
51:33
Unidentified speaker (speaker_0): Hi, how are you doing today?
51:35
Kurt Chisholm: Good to have you on. We also have a very our own Doug Hagrin. Hey Doug.
51:38
Unidentified speaker (UNKNOWN): How
51:38
Alex Chompff: are you doing today, Kirk, Barb?
51:40
Unidentified speaker (UNKNOWN): Good
51:41
Alex Chompff: to see you.
51:41
Kurt Chisholm: Let's dive right in. So, Barb, what are some of your thoughts from the interview?
51:45
Unidentified speaker (UNKNOWN): It's
51:45
Unidentified speaker (speaker_0): a brave new world. Things are changing so fast, my head is spinning. And, you know, where's AI gonna go? How is it gonna impact investing? And I think that's what's really important for an investing podcast to talk about.
52:02
Kurt Chisholm: Okay. Doug, what about you?
52:04
Unidentified speaker (UNKNOWN): I
52:04
Alex Chompff: came away from this interview,
52:08
Unidentified speaker (UNKNOWN): probably more bullish on AI itself, but even more skeptical about it as an investment. I think you made a point that really stuck with me for essentially all of human history. Cognition has been very expensive. If you wanted more thinking capacity in your business, you go hire another person. I actually saw someone. the other
52:27
Unidentified speaker (UNKNOWN): day on one of the social media platforms talking about how they were trying to hire someone for their company. And
52:35
Unidentified speaker (speaker_0): it was like the 10th person in a row that they made an offer to that basically goes to them last minute. Just a little background: my parents or my father owned an employment agency. So I know this really, really well. It's extremely expensive for a company to hire someone. It's expensive for them to train. And it's expensive for that turnover. So that was one of the
52:55
Unidentified speaker (speaker_0): reasons temp companies generally made their arguments. Like, look, we deal with all of that. You just get to have someone show up. Well, now, especially has been heightened since COVID. That expense has gotten even greater because now you can't actually hire anyone. The new world of ghosting and just trying to get someone to show up that's promised adds another expense.
53:15
Unidentified speaker (speaker_0): And then once you get them on, Kirk, I know you've been through this. You educate them, you train them, you paid them.
53:19
Unidentified speaker (UNKNOWN): And
53:19
Unidentified speaker (speaker_0): now we're watching the cost of that cognition collapse right in front of our eyes.
53:24
Unidentified speaker (UNKNOWN): And
53:24
Unidentified speaker (speaker_0): it's extraordinary because I think it probably changes every business, including yours, mine.
53:29
Unidentified speaker (UNKNOWN): But
53:30
Unidentified speaker (speaker_0): here's where I separate technology from investment. The internet changed the world. Plenty of those internet investors got destroyed. His dark fiber example actually. illustrates that perfectly. We eventually needed all that fiber. And that didn't mean people who originally financed it all made money. A lot of them got fleeced. And that's what I'm talking about AI right
53:50
Unidentified speaker (speaker_0): now. I mean, we're spending staggering amounts of money on chips, data centers, power, infrastructure.
53:55
Unidentified speaker (UNKNOWN): And
53:55
Unidentified speaker (speaker_0): now you got companies like Anthropic who are looking towards the public markets, just like SpaceX just did. You know, everybody thinks. hey, rockets, really cool. Internet, really cool. No, those parts of SpaceX make a lot of money. But the bloat from XAI is absolutely sucking everything out of that organization. And after
54:14
Unidentified speaker (speaker_0): it's IPO, a lot of people predicted it was a cash grab to fund CapEx, and the stock is As cratered. Now, Barb, I'd like to start the investment side. I mean, how do you feel? Are we still early? in the AI investment cycle? Are we reaching the point where private capital has enjoyed an enormous run and now they need the public markets to fund an
54:34
Unidentified speaker (speaker_0): absolutely massive Capex bill going forward? Because I don't doubt AI.
54:39
Unidentified speaker (UNKNOWN): I'm
54:39
Unidentified speaker (speaker_0): questioning who captures the economics from here and what price we're being asked to pay at. I think that's a really good question, Doug. And as in many startup type industries and businesses, I go back to Amazon, which I followed from day one. And I was like operating at a loss, operating at a loss, operating at a loss,
54:59
Unidentified speaker (speaker_0): operating. I mean, I don't know how many years.
55:03
Unidentified speaker (UNKNOWN): And
55:03
Unidentified speaker (speaker_0): I thought, wow, when are they going to run out of money and look at Amazon today? Day.
55:09
Unidentified speaker (UNKNOWN): So
55:10
Unidentified speaker (speaker_0): I think we're going to have some of those companies
55:13
Unidentified speaker (UNKNOWN): that
55:13
Unidentified speaker (speaker_0): are going to be asking. for capital, going public, and you know, operating at a loss for a long time, and then boom, they're going to be big. But the question is, like in the dot com boom and bust,
55:28
Unidentified speaker (speaker_0): who's going to be left naked when the water goes out of the pool? Who's going gonna to be the winners? And I don't know about you. I don't have a crystal ball, so I would not even guess who are going to be the winners. Subsequently, I would be very cautious about
55:48
Unidentified speaker (speaker_0): betting big on any of these, even Claude, which I happen to really like. I use Claude a lot in my day-to-day work,
55:58
Unidentified speaker (speaker_0): but I'd be really cautious of where I put my money. in this whole AI picture right now.
56:05
Unidentified speaker (UNKNOWN): I
56:06
Kurt Chisholm: think that's a good sentiment. I mean, look, I've got some strong thoughts about AI. I've been formulating over the last 12 months, and we have an episode, it'll probably be out by the time this episode's out, but we're gonna be talking a lot about this subsequently. But, you know, as Barb and Doug said, look, we're in the late 90s nineties of the internet bubble. We have an AI bubble, everyone knows it.
56:26
Kurt Chisholm: It's not a secret.
56:27
Kurt Chisholm: And it's a game of musical chairs. You don't want to not get up and dance, as Jamie Diamond said, when the music's playing, you got to get up and dance. But you also don't want to be without a seat. And I think the challenge is as we get into this period is everyone wants to make lots of money, but no one knows where we are in the cycle. Now, this cycle could last
56:47
Kurt Chisholm: another five years. I don't think it will. I think we're in the ninth inning, maybe even in extra innings at this point. From what we're seeing now, Anthropic and OpenAI are trying to go public. And as of a few days ago, the big frontier models were lobbying the government to get
57:06
Kurt Chisholm: protection. I think the biggest challenge we see is we don't know what's real and what's not, because the AI companies have gotten really good at PR.
57:15
Kurt Chisholm: where anthropic being the best is they'll come out and say, Oh, we had this new model, Opus or whatever, this new model, it's coming out, but we can't release it because it's gonna be a hack in every computer on the planet. So, what does everybody say? Ooh, ooh, me, me, me, I want some, I want some, right? And it scares the government. The government's like, Oh, we can't let this happen. And so it's great
57:35
Kurt Chisholm: PR. If you're trying to release a new model, who wouldn't want to sign up for that? It sounds really powerful. Turns out it's not that powerful. It is, but not as much as they say. They scared everybody.
57:46
Unidentified speaker (UNKNOWN): A
57:46
Kurt Chisholm: few weeks ago, somebody was hired and then left really quickly and said, There's a 10% chance it's going to destroy humanity. Well, if there's a 10% chance that AI is going to destroy humanity, then every single one of the leaders of these AI companies should be in jail. Your product. is going to kill every person on the planet, 10% chance.
58:06
Kurt Chisholm: And you're going to keep building it, you should be in jail. There's no other outcome for that. So it's not true, obviously. They scare everybody into believing it is. It's great salesmanship, great marketing, because saying, hey, we're so good that people are all going gonna to die. So you need to have this model. So fear.
58:27
Kurt Chisholm: is a big proponent here. Now, when they went to lobby Trump for protection in exchange for, from what I heard, a percentage ownership in the company, which is bad on many levels. I know that it happened with Intel and a few other companies. It's not good precedent. But the fact that they wanted the government to take over the company and
58:47
Kurt Chisholm: regulate these companies tells me that these companies can't. float on their own. Because if they could, they wouldn't be asking for this. Because if they were able to float on their own, they would just say, nope, we don't want any part of that. It's only when you start to stumble that you want protection. Or when you're big enough where you're worried about the
59:06
Kurt Chisholm: upstarts coming and eating your lunch that you want government protection. So to me, this sounds like not good for the AI company. companies. Now, personally, I think these AI frontier models are worth zero. Because if you're not on the frontier, you're worth zero. Because the
59:26
Kurt Chisholm: open source models, which there are many out there and they're quite good, they're maybe three months behind the frontier models, they're free. So if they're free, what is the frontier model worth if everyone? else catches up? So we're zero. That's the problem with these AI companies. I have a big bone pick
59:46
Kurt Chisholm: with them anyway. But the fact is, is if you're looking to invest in these things, I would not get anywhere near these things. I think Anthropics can be the first one to go public, and I don't know the exact date yet, but people are speculating in November. I would suspect it's soon because they need to get these things out so they can monetize their. shares and whatever they're going to do. But I would be very shocked.
1:00:06
Kurt Chisholm: I'd be I would not be shocked if these things don't do well in the next six to twelve months.
1:00:12
Unidentified speaker (speaker_3): I want to come back to what Barb said. I thought that was really interesting about Amazon and a couple of
1:00:17
Alex Chompff: the other points here. If we take a look at Amazon, you know, and absorbing those losses for years while building scale, you know, the customers that focus on current profitability miss what it eventually became. If look what it transformed to. Well, look what really transformed its economics. AWS. In
1:00:36
Alex Chompff: 2025, Amazon Web Services was only 18% of Amazon's revenue, but it generated
1:00:45
Unidentified speaker (speaker_0): forty-five $45.6 point six billion dollars of its eighty $80 billion dollar operating income. That's fifty-seven 57%. percent.
1:00:51
Unidentified speaker (UNKNOWN): And
1:00:51
Unidentified speaker (speaker_0): now Oz is one of the very businesses that's supplying the enormous compute and data center infrastructure that the AI companies need. Amazon says Oz added 3.9 gigawatts of power capacity in 2025, and
1:01:05
Unidentified speaker (UNKNOWN): they
1:01:05
Unidentified speaker (speaker_0): expect to double total power capacity by the end of 2027.
1:01:09
Unidentified speaker (UNKNOWN): It
1:01:09
Unidentified speaker (speaker_0): says the day AI demand is already constrained by available
1:01:12
Alex Chompff: capacity. And so the Amazon strengthens. my concern more than it weakens it. It struggled for years. It's become one of the greatest companies in history. But if you look at what ultimately became its engine, it's Oz. And it didn't just get better at shipping books and toilet paper. It built one of the world's dominant cloud and data center services. And
1:01:32
Alex Chompff: the irony is Oz is now selling the picks and shovels
1:01:37
Unidentified speaker (UNKNOWN): in the
1:01:37
Alex Chompff: world of the AI gold rush. The Amazon may be a better side of this trade. It's getting paid to provide compute, whether Anthropic, OpenAI, or somebody we haven't heard of ultimately wins. I think this is where the numbers are startling. Reuters reported yesterday that Anthropic's confidential IPO prospectus projects at
1:01:57
Alex Chompff: least $518 billion of AI infrastructure spending over 10. years, with roughly 80% of those commitments are non-cancelable or payable regardless of usage. And it filed confidentially for its IPO in June. Axios reported that it needs additional capital soon, whether the IPO or
1:02:17
Alex Chompff: another funding route. So my problem with comparing Anthropic to Amazon is Amazon eventually became the owner of Oz. Anthropic is the customer. that's writing enormous checks to the Oz of the world. And that's two very different places to sit from a capital structure of the AI boom.
1:02:35
Unidentified speaker (UNKNOWN): And
1:02:35
Alex Chompff: if I'm Amazon, I build the data center and Anthropic pays me for compute.
1:02:40
Unidentified speaker (UNKNOWN): If
1:02:41
Alex Chompff: I'm Anthropic, I got to generate enough economic value from Claude to pay hundreds of billions of dollars of infrastructure commitments and then still leave. enough economics for my shareholders. So before I say, well, Amazon lost money too, I think the big question is what in the world is anthropics
1:03:00
Alex Chompff: oz? Where are they going to get that? And coming back to your point on
1:03:05
Unidentified speaker (UNKNOWN): the
1:03:05
Alex Chompff: ninth inning, you know, of AI, I think we could be much later inning of easy money AI investment story. I mean, AI adoption could be innings. inning two or three. Well, today's valuations infrastructure build out already discount innings seven, eight, and nine. And the internet in 1999 wasn't in the ninth inning.
1:03:25
Alex Chompff: It was barely getting started. The internet stocks were the thing that was the ninth inning. There's so many pieces here that are I just create so many headwinds for that. This is where my skepticism really kicks in. You You're telling me this technology is so economically powerful that's going to transform civilization. You're simultaneously telling me it could
1:03:45
Alex Chompff: potentially destroy civilization. You're lobbying Washington heavily over how it's regulated. You're asking governments to facilitate the energy infrastructure necessary to build it.
1:03:56
Unidentified speaker (UNKNOWN): And
1:03:56
Alex Chompff: now you're coming to the public markets needing enormous amounts of additional capital. As an investor, I'm allowed to ask. whether those things are connected.
1:04:05
Kurt Chisholm: Yeah, it's tough. I mean, Barb, what do you think in terms of investing in AI? I've got some more thoughts too, but in terms of investing in AI, I mean, there's different ways you can invest in AI, you can use AI. How should investors consider AI in their process or in their investments?
1:04:22
Unidentified speaker (speaker_0): I loved the parallel that both of you have drawn. between the actual technology because AI is free to use for most people. I would gander the majority of people don't pay for it. You're not going to make money as a company if nobody's paying for your
1:04:42
Unidentified speaker (speaker_0): services. Back to what Doug said about Amazon selling the picks and shovels. I've been watching a program on PBS about the West. And it's in the gold rush days. And in the beginning, there was tons of gold, and people were getting so rich. And then it ran out. But you know who was really
1:05:01
Unidentified speaker (speaker_0): making money? Were the stores selling the picks and shovels, like Doug said. If you're going to invest in AI, and full disclosure, I'm not looking for AI stocks, AI. funds, whatever. But if you're going there, look who's giving them the
1:05:21
Unidentified speaker (speaker_0): underpinnings of what they need to run. And that would be the place that I would think about investing. But I want to flip this a little if this is okay with you guys and talk about actually using AI as an investor. Because from what I'm hearing from the
1:05:41
Unidentified speaker (speaker_0): discussion, none of us are big. Oh, let's dive right into the new anthropic IPO, or let's just find the next biggest AI stock and glob on. Because that's a recipe for disaster, unless you have a crystal ball. But there's a lot of talk
1:06:01
Unidentified speaker (speaker_0): in how AI.
1:06:03
Unidentified speaker (speaker_0): can change the world of investing and make investing smarter, better, more successful. I want to give a little example of what I did and how it blew my mind. I happen to have a, you know, a legitimate portion of my portfolios in
1:06:23
Unidentified speaker (speaker_0): fixed assets. That's bonds, bond funds, short-term short care. term cash, short term bonds, whatever. I'm of that age where, you know, I'm trying to preserve some capital, which allegedly I don't know if bonds is the right way to do it right now. But that said,
1:06:40
Alex Chompff: by percent no risk, Barb. I mean, come on, you're sitting on
1:06:43
Unidentified speaker (UNKNOWN): a
1:06:43
Alex Chompff: gold mine.
1:06:46
Unidentified speaker (speaker_0): I've been asking myself that same question, but what I've been doing is trying to see where I can get some kind of return in this rising interest rate environment on my short-term cash, cash that I want to keep very, very liquid. So I asked AI, you know, do some comparisons, two-year treasuries,
1:07:06
Unidentified speaker (speaker_0): short-term bond funds, money market funds, and put together a nice comparison sheet. Beautiful job.
1:07:13
Unidentified speaker (UNKNOWN): Went
1:07:14
Unidentified speaker (speaker_0): to check the data.
1:07:15
Unidentified speaker (speaker_0): It was all wrong.
1:07:17
Unidentified speaker (UNKNOWN): I
1:07:17
Unidentified speaker (speaker_0): mean, it was literally wrong. So, what did I do? I went in there, I created a spreadsheet. I got the data myself from the websites. What is this money market fund paying now? What is this short-term bond fund paying now? And I did it myself.
1:07:37
Unidentified speaker (speaker_0): I was like
1:07:39
Unidentified speaker (speaker_0): shocked that what I thought was the most simplest question was done wrong by AI. Thoughts?
1:07:50
Kurt Chisholm: You raise a good point, Barb. And I think a lot of people in our industry are interested in using AI for investing. Testing. And I have not met a single person who has found a valuable use case for it. There are advisors who use it, they do it for note-taking and things like that. And they maybe do a little research. But from what I've seen and all the
1:08:09
Kurt Chisholm: people we talked to on the show and otherwise, it does have valuable uses. If you're in the bioscience or biotech space, it's really useful. I think we think of AI in the wrong way. Most people think of AI as this potentially sentient thing that you can talk to and have a conversation.
1:08:27
Kurt Chisholm: AI is basically it's machine learning, first of all, which has been around for 25 years or over 25 years. But machine learning is basically taking big data, so taking a big data set and finding ways to thread the needle and figuring out like how things link together. The whole point, big data was a big deal back
1:08:47
Kurt Chisholm: in the late aughts, I think it was, or early tens. and it was a big trend. People loved it. Like, oh, big data, like you know, and IBM and a few other companies were out there trying to figure out the whole big data trend. Well, machine learning is basically taking the big data. In order to have big data, you need a clean data set. Reddit is not a clean data set,
1:09:07
Kurt Chisholm: right? That's how AI and anthropic and all these these models learn, is on Reddit. I don't think Reddit's a good clean clip.
1:09:13
Alex Chompff: Mm-hmm. advisor, Yelp.
1:09:15
Unidentified speaker (speaker_0): I know it is insane. I mean, not to interrupt, but I have to go in and say, use only these sources. The fact that they would even think to use Reddit and Yelp as a source completely disproves Proves the whole methodology.
1:09:35
Kurt Chisholm: Right. So for instance, take like bioscience or biotech. Some of the data sets they have are the cleanest data sets on earth. I forget which ones, but there's some, I think it was like Cancer. There's certain data sets that are extremely clean. It's how it was designed. And everyone knows it. I just off the top of my head can't remember it. But there's a certain data set that comes to
1:09:55
Kurt Chisholm: health science, and it's really clean. Now, if you're using AI with that, that's probably very good. Except AI is not good at math.
1:10:03
Kurt Chisholm: So just keep that in mind. But the data itself, you're getting clean data. The internet is not clean data. You know how many trolls are out there? Forget about it. It's not clean, but that's where it's getting its stuff. So if you want to understand why you're not getting clean data, well, part of it is, or why you're not getting good answers because not getting clean And data. I mean, let's talk about use
1:10:23
Kurt Chisholm: case for investing. I'm not a big fan of investing in AI or AI. AI-related companies at the moment. As I said, we're in the ninth inning or even extra innings in the bubble. There will be a time to buy it when it's cheap after it's all bombed out and buying Amazon at like two bucks a share back in like oh three, it was like dirt cheap, you know, or Apple
1:10:43
Unidentified speaker (speaker_3): for like
1:10:43
Kurt Chisholm: what it was, like eight bucks or something like that. There'll be a time to do that. I don't think now's the time, but you know, there'll be a time to do that. So there's two parts. One is investing in AI or AI related companies. I would not invest in AI companies. AI related companies, yeah. Cybersecurity is probably decent. Nuclear is going to be really important because you need
1:11:03
Kurt Chisholm: energy. Semiconductors, there's certain companies out there, they're probably a little overpriced, but you know, good companies. There's related companies. I wouldn't invest in the frontier models. I think they're worth zero personally. You know, you have those companies and then you have using AI for investing. So if using AI for investing isn't any good. Well, I was
1:11:23
Kurt Chisholm: working on a project last week and had to do with options. For those of you not familiar with options, I'll just simplify this. I asked it to do some calculations for me and do some back testing. And you know, I was trying to test out this thing and see what it could do. Well, it made some egregious errors. First of all, it assumed that annual returns were
1:11:43
Kurt Chisholm: from November to November. Okay. And it miscalculated the year. So it had 2021 as a really bad year, but 2022 was the bad year. Those are simple things that if you don't know anything and you just take the data, you're going to make egregious errors.
1:12:03
Kurt Chisholm: because it wasn't even close to the performance numbers for those years for like the SP 500. It also made the egregious error of like the options expiration. It didn't know that it's like the third week of the month every single time. It just didn't know that. And the math was wrong in many occasions. And I would not have caught the math
1:12:23
Kurt Chisholm: unless the fact that I was actually doing the math in parallel to see if it was accurate. The big problem with AI is not a computer. It's not going to get the math right every single time, like an Excel spreadsheet. It is like expecting an English major to be an expert at math. They're two different things. When you're thinking about AI, the way
1:12:43
Kurt Chisholm: I think about AI is with investing is I'm not thinking about it, frankly. I'm playing around with it, but I'm really not using it to the degree.
1:12:51
Kurt Chisholm: that we all hope it could be. What AI is going to end up being, we all are afraid that AI is going to take our jobs. I feel like the South Park skit, you know, took our jobs. You know, I need to play that skit here. But it's not going to take our jobs. What it is going to do is it's going to make our jobs better because it's a tool.
1:13:11
Kurt Chisholm: Like software is a tool. We have CRM. CRMs and all these things we have like workday or I'm drawing a blank on the name of the other one, but we've got all these like software tools we use at work. AI is just gonna be another software tool. It's not gonna take your jobs. It's just another software tool you have to use.
1:13:30
Unidentified speaker (UNKNOWN): A
1:13:30
Kurt Chisholm: lot of people think it's gonna destroy all these jobs. It's not. I will put money on this if somebody wants to make a bet. I don't know if they can even do this, probably not allowed to do it, but I will take a bet if anyone wants to take me up with this. Just be a gentleman's bet, because otherwise I might get in trouble. But if you want to bet me, that's fine. It's not going to take anybody's jobs. I know it'll be attributed to that. And they'll say, oh, we lost all these jobs.
1:13:50
Kurt Chisholm: AI. No, they would have fired those people anyway. And they're looking for an excuse. AI is not going to take the jobs. It's going to change your job. You're going gonna to have to learn a different software tool, a different skill set, but it's not going gonna to take your jobs because it doesn't have that. ability. That's the claim I'm making right here. I know people probably think I'm cuckoo, but I will make this claim. It is not going to take your jobs because it's not
1:14:10
Kurt Chisholm: capable of doing that. It is not going to become sentient. It is not going to become a person. It's not going to have personhood or any of that crap that people are afraid of. It's not going to destroy the world. It is just going to be a different operating system that we have to learn. That's my two cents. What do you think, Doug?
1:14:24
Alex Chompff: There's so much in
1:14:25
Unidentified speaker (UNKNOWN): that
1:14:25
Kurt Chisholm: you're welcome.
1:14:27
Alex Chompff: First of all, I'm going to come back to the bad data part of it. The danger with AI right now
1:14:33
Unidentified speaker (UNKNOWN): is
1:14:33
Alex Chompff: hallucinations don't look like hallucinations at all. AI will still confidently make
1:14:42
Alex Chompff: and that's a pretty serious problem when we're talking about somebody's money. There's a lot of people like, hey, I, you know, AI is going gonna to help me to do my own financial planning and my financial analysis. One of the things about what we do for a living is we know enough to know what we're being told doesn't smell right. We've smelled it before. It's like, eh,
1:15:02
Alex Chompff: I gotta watch that. Most people don't understand how to do that. I had a client who was getting into AI who went through and said, Hey, I was running some analysis on it and blah, blah, blah, and some questions I said, I'm trying to understand your fee. because it seems like it's higher than these others. Well, AI did the calculation and basically calculated that
1:15:22
Alex Chompff: the financial advisory fee on his portfolio was about 30 basis points higher than it actually was. That's significant. You know, 0.3%
1:15:33
Alex Chompff: adds up a lot. And it was completely wrong. It couldn't do that basic math in. the information. So look, I'm comfortable allowing it to helping me analyze the data. I'm much less comfortable in letting it supply the data I'm analyzing. Give it a 10K, ask it to dissect a business, great. Ask it from memory
1:15:53
Alex Chompff: what Google's margins were, what Amazon's AWS revenue was, what an ETF owns, what the fee on that ETF, internal expense ratio on that ETF was, what the tax loss is. Now I'm verifying every material number before I make a decision. And I think this hallucinations that look so confident is
1:16:13
Alex Chompff: something that investors need to understand. If Google gives you no result, you know you don't have the answer. AI can give you a company name, a revenue number, a evaluation, valuation, a quotation from the CEO, historical return, and an extra. explanation of why the stock moved, all written beautifully, confidently. Humans do this,
1:16:32
Alex Chompff: right? The most wrong tend to start acting like the most confident in their wrongness. That's the same thing with AI. I mean, it's almost too human in the fact that it is arrogantly confident in its false facts. And because the surrounding analysis sounds intelligent,
1:16:49
Unidentified speaker (UNKNOWN): the
1:16:50
Alex Chompff: false facts. becomes harder to detect. The most dangerous answer isn't I don't know. It's a beautifully reasoned answer, all built on fact that isn't true. I don't want AI to be my Bloomberg terminal. I want it sitting next to my Bloomberg terminal. Primary source data
1:17:09
Alex Chompff: first, AI second. SEC filing, AI can analyze it. Actual portfolio holdings, AI can analyze them. Fund prospectus, AI compares it. Client financial data, AI can stress test it. Actual tax code and source, help interpret it. But not, hey Gemini,
1:17:30
Alex Chompff: what does this company earn? Then invest based on that response. Use it as your analyst, not your source. This explains why.
1:17:39
Alex Chompff: advisors aren't going to disappear. It comes back to conversation around cognition becoming free. Barb, if you're right that AI is effectively free
1:17:48
Unidentified speaker (UNKNOWN): and
1:17:49
Alex Chompff: the information itself becomes incredibly cheap, but cheap information doesn't necessarily equate to reliable information. So the value of a professional in someone in this field can't continue to be: I know things you don't know. AI destroys that mode. Access to information is greater than it's ever been.
1:18:08
Alex Chompff: But the value becomes I know which questions to ask, which sources to trust. When my answer doesn't smell right, what actually to do with that answer? I'm in college plan. You can go out on AI and run all sorts of scenarios about what to ask and what to do with college. I do that all the time, but
1:18:28
Alex Chompff: I also dissect and eliminate the things that I know are completely untrue. I know what I know to do that. Unfortunately, if you don't, it's going to take you down a rabbit hole that will quickly destroy your outcome. This is a world where we say everybody now is this incredible investment,
1:18:48
Alex Chompff: powerful investment analyst for 20 bucks a month. Okay. But everybody also is an analyst capable of manufacturing a completely fictitious fact with absolute confidence. And so I think the primary source has actually become more valuable in the AI world, not less. The SEC filings, the audited financials, the
1:19:08
Alex Chompff: actual company disclosures. Because eventually I need to be able to double-check double check where that fact. originated. So this is where I've landed today. I think Alex is right about cognition becoming dramatically cheaper. I think Barb, you're right about approaching essentially free access and extraordinary analytical capability.
1:19:28
Alex Chompff: And I don't think, Kirk, you're crazy to be skeptical about where we are in the investment cycle. AI can make me a dramatically better investor, it can read these filings. it can challenge my thesis, it can compare companies, it can identify concentration, it can model scenarios, and it can argue the other side better than I can.
1:19:48
Alex Chompff: That's language. But I don't outsource truth to it. Give AI verified information, let me think. But don't let AI invent the information and then marvel at how intelligently it analyzes what it just invented. I'll give you a personal antidote on this with this hallucination. I actually was using it to
1:20:07
Alex Chompff: analyze fantasy football. What should I do? You know, here, what should I do with my team? Well, guess what? Chat analyzed my roster, matchups, consistency, all of it. It recommended that I actually give away the quarterback who had the best statistical results over the last three years. That data is. right there
1:20:27
Alex Chompff: compared to the other two. And when I challenged it with the actual results and the data, it basically was like, Yep, you're right. I was wrong. But well, hey, that's fine in fantasy football. Although I would have been really annoyed for the rest of the season that, you know, I just gave this away and I'm basically dealing with bench players on my team.
1:20:47
Unidentified speaker (UNKNOWN): But
1:20:47
Alex Chompff: it's a very different conversation when that's my ridiculous. retirement portfolio. And what bothered me wasn't that AI made a bad prediction. Humans make bad predictions consistently. It's how easy it was for AI to construct a very convincing argument for something and then construct an equally convincing
1:21:07
Alex Chompff: explanation afterward for why that argument was wrong. And then you got this ridiculous example everybody's seen online, ask an incredibly sophisticated. AI how many R's there are in Strawberry. And the models have famously got it wrong. Go out there and look it up. They've asked it over and over. Think about the contradiction. The thing that can explain quantum
1:21:26
Alex Chompff: mechanics, analyze a financial statement, and write computer code,
1:21:30
Unidentified speaker (UNKNOWN): and
1:21:30
Alex Chompff: it can stumble counting letters in a word. We got a long way to go.
1:21:33
Unidentified speaker (speaker_3): Yeah, I couldn't agree more. All right, we're gonna wrap it up here. Barb, final thoughts from you.
1:21:38
Unidentified speaker (speaker_0): I think. you have to count on your own good judgment. And if you're young or you don't have good judgment, or you don't have a knowledge base with which to evaluate how you are using AI or which AI companies that you are considering
1:21:58
Unidentified speaker (speaker_0): investing in, I would suggest take a step back.
1:22:03
Unidentified speaker (speaker_0): and learn a little bit of information about
1:22:08
Unidentified speaker (speaker_0): what the questions you are asking AI. Don't depend on them. Verify, verify, verify. And in terms of investing in AI, you want to make sure that when you invest in a company, the keys are. do they have
1:22:28
Unidentified speaker (speaker_0): a marketable product with hopefully a big moat? That means it's tough for other companies to get into that industry and compete. And what are their growth projections? Can they keep up growth in the future? And just even Stephen raising those points, it's like I can't think of
1:22:48
Unidentified speaker (speaker_0): a company that is AI. straight up that is going to fit that criteria. So do your homework. Get educated. There's no easy path to money, to investing, to knowledge. You've got to do the work. I'm Barbara Friedberg. You can find me on YouTube and
1:23:08
Unidentified speaker (speaker_0): at Barbara Friedberg Personal Finance.
1:23:11
Unidentified speaker (UNKNOWN): Thanks
1:23:11
Unidentified speaker (speaker_3): for coming on, Barb. Doug, final thoughts from you. And where can people find more about you?
1:23:15
Alex Chompff: So I think where I land after all of this is that I'm incredibly optimistic about AI. I use it every day. It has been an incredibly powerful tool to making almost everything I do easier. It has allowed me to analyze things faster. It's allowed me to
1:23:34
Alex Chompff: construct better language. It's allowed me to design things. that would have taken enormous amounts of time or skill that I may or may not have. And ultimately, it's allowed me to do everything faster, which is incredible for the efficiency. But there's two words that have to go together. We live in a world where if you need
1:23:54
Alex Chompff: help, Silicon Valley no longer has anybody to provide any level of customer service because why? They've become incredibly efficient.
1:24:03
Alex Chompff: But they're not effective. I'm increasingly skeptical of anyone who tells me they know exactly how this is all going to play out. Because while it may be efficient, it is yet to combine with a certainty of effective. It can transform the world and today's AI investments. It can still be overpriced. Google can
1:24:23
Alex Chompff: build an incredible product that potentially disbups disbumps its own business. model. It spent 25 years making money off what people click, and now it's sending people outside of that to Gemini. Amazon can benefit because AWS owns the infrastructure everybody needs. And companies we haven't even heard of yet may ultimately capture enormous amounts of
1:24:43
Alex Chompff: value.
1:24:43
Unidentified speaker (UNKNOWN): And
1:24:44
Alex Chompff: I've learned personally not to blindly trust this machine. You know, I helped it manage my fantasy football team and convinced me to give away. what turned about to be the best producing quarterback. I came back with the results. It said, yep, you're right, I was wrong. And that's funny when it's fantasy football. It's considerably less funny when it's your retirement. So use
1:25:03
Alex Chompff: AI. Learn it. Let it make you better at what you do. Use it to challenge your thinking and analyze information, but do not outsource your judgment to it. AI may make cognition abundant, but it hasn't made truth abundant. And I think that's actually a good analogy for financial planning.
1:25:23
Alex Chompff: Nobody knows exactly what AI does next, what the market does next, what interest rates do next, or which company is going to become the next Amazon. And you shouldn't need to know. And I mean, a good financial plan doesn't require correctly predicting one version of the future.
1:25:39
Unidentified speaker (UNKNOWN): It
1:25:39
Alex Chompff: should allow you to participate in opportunities while still protecting the money that already has a job, your retirement income, liquidity, your kids' education, and the things your family actually wants money to accomplish. And that's what we do at Mergent Advisors. And you can find out more at mergentadvisors.net. Advisors.net. That's
1:25:59
Alex Chompff: M-E-R-G-E-N-T-M-E-R-G-E-N-E-N-E-N. M-E-R-G-E-N-T, M-E-R-G-E-N-T. We help families bring investments, retirement income, taxes, risk, liquidity, and college planning together in one coordinated plan. So your future doesn't depend on getting every prediction right. So if you're approaching retirement, already retired, or simply wondering whether all the
1:26:18
Alex Chompff: pieces of your financial life are working together seamlessly,
1:26:22
Unidentified speaker (UNKNOWN): go
1:26:22
Alex Chompff: to mergent advisors.net, start a conversation with us, participate in. the future, and just don't make your financial plan dependent on predicting it. Cause smart money families don't guess. They plan.
1:26:35
Unidentified speaker (UNKNOWN): Well,
1:26:35
Unidentified speaker (speaker_3): thanks for coming on, Doug. So that's the show for this week. Thanks again for joining us. The Money Tree Investing Podcast. My name's Kirk Chisholm, wealth manager of Innovative Advisor Group. We don't just manage your wealth, we make your life better. You can find more about me at innovativewealth innovativewealth.com dot com. And of course, you can find me every week here on the show. We share a lot of charts in the show. so if you're looking for charts and show notes, check out our show at Moneytree Podcast dot com. Don't forget,
1:26:55
Unidentified speaker (speaker_3): do your own research. This show is for informational use only. We're not telling you what to think, merely how to think, but investing. We're also not selling products and services, so do not consider this advice. If you're seeking financial advice, give us a call. We take clients selectively, but only if we like you. Have a great week ahead, and remember, no one will care about your money like you do. So invest. in
1:27:15
Unidentified speaker (speaker_3): your life.
1:27:17
Unidentified speaker (UNKNOWN): Thank
1:27:17
Unidentified speaker (speaker_0): you for listening to the Money Tree Investing Podcast. Visit us at Moneytree Podcast.com for more free investing resources.
This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3+sortformer-v2.1 and interpreted by Qwen3.6-35B-A3B-UD-IQ3_S.gguf. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.

More from Money Tree Investing

Money Tree Investing

The College Planning Mistakes That Could Cost Your Family Thousands

Episode discussion: College planning expert Brian Safdari argues that higher education is a complex business and families must start planning early to optimize financial aid and avoid debt traps.

Speakers’ claims: College planning expert Brian Safdari argues that the US higher education system is deliberately complicated to maximize revenue, with tuition costs rising steadily since 1980. He advises families to begin strategic planning in seventh or eighth grade to align academic preparation with student interests and financial goals. Safdari claims that colleges use 'strategic enrollment management' to manipulate rankings and that high-income families can leverage institutional aid and tax strategies to reduce costs. The discussion emphasizes that college planning is inseparable from retirement planning, as both draw from the same family balance sheet. Speakers warn against relying solely on guidance counselors or standardized rankings, advocating instead for proactive, holistic financial and academic preparation.

College PlanningFinancial AidHigher Education Costs
1h 13m0 assets22 claims to verify
Money Tree Investing

The Busiest Week Of The Quarter Reveals...

Episode discussion: Midterm election outcomes priced in; tech volatility and recession risks discussed

Speakers’ claims: Kirk Chisholm and Doug Habrin discuss how midterm election results were anticipated by markets, attributing recent volatility to institutional repositioning rather than fundamental shifts. Chisholm highlights sell-offs in AI and chip stocks, warning of recession risks and stagflationary pressures despite resilient labor data. Habrin contrasts current economic resilience with the 1970s, noting U.S. energy independence mitigates oil embargo risks. The speakers also analyze Tesla's Cybertruck production halt, gold market divergences, and big tech's decoupling of revenue from headcount.

Market VolatilityMidterm ElectionsTechnology Sector
53 min4 assets9 claims to verify
All insights