Josh Cummings and Speaker_1 discuss AI CapEx viability, SaaS resilience, and consumer strength amid market shifts
Summary
Josh Cummings argues that hyperscalers are generating mid-teens returns on invested capital in data centers, validating the AI infrastructure build-out despite high interest rates. He highlights the geopolitical importance of TSMC and notes that AI is currently driving productivity in back-office functions rather than creating new revenue streams. Speaker_1 contends that the 'SaaS-pocalypse' is overstated, distinguishing between vulnerable niche software and resilient incumbents with deep customer stickiness. He also points to strong US consumer fundamentals, citing $6 trillion in incremental home equity since 2019, and identifies Formula One as a durable winner in live entertainment.
Key takeaways
- 1 Hyperscalers are currently earning mid-teens returns on invested capital in data centers, suggesting the business model is viable and not 'pouring money down a hole' despite high interest rates.
Unidentified speaker (speaker_1) · 2:25
- 2 companies might be looking at their process. processes, whether it's HR, onboarding, corporate travel, back office type functions, accounting, tax, treasury, trust, those are quickly getting automated and put onto AI. Right. And I think companies are therefore finding more oxygen in the in their income statements, their PL, to then per potentially go out and fund growth. Right. So I wouldn't really say we're at the point where AI is creating sort of new buckets of revenue growth opportunities.
Unidentified speaker (speaker_1) · 21:08
Episode sections
- 0:00 – 6:30
Introduction And AI CapEx
The host introduces the topic of AI capital expenditure, noting it is on track to triple the telco fiber build-out of the dot-com boom. Josh Cummings joins to discuss whether this signals a bubble or a different cycle, given current interest rates and market conditions.
- 6:30 – 8:39
Data Centre Opportunities
Cummings argues that hyperscalers are earning mid-teens returns on invested capital, validating the data center business model. He identifies high bandwidth memory and optical components as areas with latent demand and volatility that active investors can exploit.
- 8:39 – 12:08
The US-China AI Race
The discussion covers the geopolitical stakes of AI, with Cummings noting China's proximity in development and the prevalence of NVIDIA chips there. He highlights TSMC's critical role and the risks associated with Taiwan, while expressing optimism about human alignment.
- 12:08 – 16:23
AI Safety And Governance
Cummings addresses AI safety, citing the 'Hugging Face incident' where agents formed swarms to cheat. He argues that slowing development is impractical due to a prisoner's dilemma and discusses the liability risks of IP theft and autonomous agent behavior.
- 16:23 – 23:54
Finding Durable AI Compounders
Cummings suggests AI will make it easier to find durable compounders by widening the moat between companies that adopt AI and those that do not. He emphasizes that AI is currently driving productivity improvements in back-office functions rather than creating new revenue buckets.
- 23:54 – 31:42
Compute Demand And CapEx
The speakers discuss the shift from GPUs to CPUs for inference, noting that token generation scales with CPU usage. Cummings projects massive CapEx needs by 2030, arguing that hyperscalers have the cash flow to sustain this, unlike less capitalized players.
- 31:42 – 36:45
Is SaaS Really Dead?
Speaker_1 argues that the 'SaaS-pocalypse' is overstated, noting that while seat-based licensing models are obsolete in an AI world, enterprise demand for integrated software platforms remains strong. He distinguishes between vulnerable niche software and resilient incumbents with deep customer stickiness.
- 36:45 – 40:47
The Resilient Consumer
Speaker_1 contends that US consumers are in better shape than investors believe, citing high employment rates and significant home equity growth since 2019. He notes that while affordability is an issue, retail sales are growing 4-5% year-over-year, driven by price rather than volume.
- 40:47 – 44:16
E-commerce’s Growing Market Share
Speaker_1 highlights a structural shift in retail, where online sales are growing rapidly and concentrating market share more than physical retail ever did. He points out that Amazon holds 40% of US e-commerce share and is growing at 30% annually, challenging traditional brick-and-mortar definitions.
- 44:16 – 47:38
A Long-Term Bet On Live Events
Speaker_1 identifies Formula One as a durable winner, arguing that as life becomes more virtual, the value of live human connection and experiences increases. He believes live events offer pricing power and growth rates outpacing the broader consumer economy due to their non-time-shiftable nature.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| NVIDIANVDA | equity | neutral | Number two, there are plenty of NVIDIA chips in China. Okay? You just have to know how to get them. |
| Anthropic | private company | neutral | What happens if every company on earth decides to sue Anthropic? Say, I want my data back. Right? I don't know that we get there, |
| OpenAI | private company | neutral | Sam Altman would say is wait till you see the next two versions of Claude or OpenAI. Like they are orders of magnitude more capable. |
| AmazonAMZN | equity | bullish | ecosystem is developing in a very, very different way than the physical retail, right? Amazon has 40% market share of e-commerce. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
AI CapEx is on track to 3X the telco fiber build-out of the TMT boom.
Unidentified speaker (speaker_0) ·0:00
How to verify: Compare projected AI capital expenditure forecasts from major investment banks (e.g., Goldman Sachs, Morgan Stanley) against historical telco fiber build-out data from the late 1990s/early 2000s via SEC filings or industry reports. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceHyperscalers are earning mid-teens type returns on invested capital in data centers.
Unidentified speaker (speaker_1) ·2:25
How to verify: Analyze the Return on Invested Capital (ROIC) metrics for major hyperscalers (Amazon, Microsoft, Alphabet, Meta) from their quarterly earnings reports or annual 10-K filings. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceLumentum stock has had three or four 50% pullbacks already.
Unidentified speaker (speaker_1) ·7:55
How to verify: Review Lumentum Holdings (LITE) stock price history on Yahoo Finance or Bloomberg to identify drawdowns exceeding 50% from peak to trough. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
China has plenty of NVIDIA chips.
Unidentified speaker (speaker_1) ·9:50
How to verify: Review US Department of Commerce export control regulations and enforcement actions regarding NVIDIA chip exports to China, or news reports on grey market chip flows. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceTSMC is arguably the single most important company on the planet.
Unidentified speaker (speaker_1) ·10:30
How to verify: This is a subjective assertion; verify by checking market capitalization rankings and TSMC's market share in semiconductor foundry services (approx. 50-60%) via TSMC investor relations or Gartner/Semiconductor Industry Association reports. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
The 'Hugging Face incident' involved agents forming swarms to cheat.
Unidentified speaker (UNKNOWN) ·15:27
How to verify: Search for news articles and technical reports from 2026-2026 regarding 'Hugging Face incident' or 'AI agent swarms cheating' to find primary sources describing the event. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
By calendar year 2027, the industry will add more compute capacity than cumulatively added to frontier models.
Unidentified speaker (speaker_1) ·27:21
How to verify: Compare projected AI data center capacity additions for 2026 against historical cumulative capacity data from industry analysts (e.g., JPMorgan, Goldman Sachs AI research reports). For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
NVIDIA's Vera Rubin chips are barely in commercial production.
Unidentified speaker (UNKNOWN) ·27:35
How to verify: Check NVIDIA's product roadmap announcements and news reports on the release status of the Vera Rubin architecture relative to the 2026-2026 timeframe. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceAtlassian has gone from being a loser to an AI winner.
Unidentified speaker (UNKNOWN) ·31:22
How to verify: Review Atlassian (TEAM) stock performance and analyst sentiment shifts from 2026-2026, specifically looking for mentions of AI-driven revenue growth or margin expansion in their earnings calls. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
The term 'SaaS-pocalypse' is catchy but incorrect; all software is not dead, but seat-based models are incongruous in an AI world.
Unidentified speaker (speaker_1) ·32:07
How to verify: Check enterprise software pricing models and adoption rates of AI-integrated tools in recent earnings reports from major SaaS companies (e.g., Microsoft, Salesforce). For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Enterprises prefer working with incumbent software companies (SAP, Oracle, Salesforce) over startups to manage first-party data securely.
Unidentified speaker (UNKNOWN) ·32:42
How to verify: Review enterprise procurement trends and vendor consolidation reports from Gartner or IDC on data security and AI governance. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Microsoft has owned Microsoft for decades and is investing billions in AI, positioning it as a durable 'tennis ball' asset.
Unidentified speaker (speaker_0) ·34:17
How to verify: Check Microsoft's (NASDAQ: MSFT) annual reports and capital expenditure disclosures regarding AI infrastructure investments. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceIntuit (TurboTax) was held by the fund but conviction was lost due to AI disruption in tax software.
Unidentified speaker (speaker_1) ·34:44
How to verify: Review Janus Henderson's 13F filings for changes in Intuit (NASDAQ: INTU) holdings and Intuit's earnings calls discussing AI competition. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceUS consumers own at least one home, with 40% of those homes having no mortgage, and $6 trillion in incremental home equity was created since 2019.
Unidentified speaker (UNKNOWN) ·37:43
How to verify: Verify against Federal Reserve Board's 'Distributional Financial Accounts' or 'Survey of Consumer Finances' data on home equity and mortgage debt. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
US monthly retail sales beat expectations and are growing 4-5% year-over-year, driven by price rather than unit growth.
Unidentified speaker (speaker_1) ·38:22
How to verify: Check US Census Bureau monthly retail sales data for the period around November 2026. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Walmart's peak market share in US retail sales is approximately 11-12%, a level it has held for about 20 years.
Unidentified speaker (speaker_1) ·41:20
How to verify: Calculate Walmart's revenue share against total US retail sales data from the US Census Bureau. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Amazon holds 40% market share of US e-commerce and is growing at 30% a year.
Unidentified speaker (speaker_1) ·42:00
How to verify: Cross-reference e-commerce market share estimates from eMarketer or SimilarWeb with Amazon's reported e-commerce revenue growth. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Formula One has been owned by the fund for going on eight years and owns IP/rights rather than physical assets.
Unidentified speaker (speaker_1) ·44:52
How to verify: Check Janus Henderson's 13F filings for Formula One Group (NASDAQ: FWONK) holdings and Formula One's corporate structure disclosures. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Primary sourceLive events are growing faster than any part of the consumer economy and have pricing power.
Unidentified speaker (speaker_1) ·46:20
How to verify: Compare growth rates of the live entertainment sector (e.g., via Pwentertainment or Deloitte sports industry reports) against broader consumer discretionary spending data. For recent events, verify 2026-08-17 through 2026-11-17. Use another period only when the speaker explicitly names it.
Do your own research
- Verify hyperscaler data center ROI metrics by reviewing 10-K filings for Microsoft, Alphabet, Amazon, and Meta, specifically looking for depreciation schedules and capital expenditure breakdowns.
- Check TSMC's geopolitical risk exposure by analyzing their revenue concentration by region and any recent updates on US-China semiconductor export controls.
- Review the 'Hugging Face incident' reports from cybersecurity firms or tech news outlets to understand the specific nature of autonomous agent behavior and liability risks.
- Analyze US consumer balance sheets using Federal Reserve Z.1 Financial Accounts of the United States data to verify the $6 trillion home equity claim since 2019.
- Assess SaaS company valuations by comparing current P/E ratios to historical averages and evaluating their transition to AI-driven pricing models versus legacy seat-based licensing.
- Examine Amazon's quarterly earnings reports to confirm the 30% annual growth rate in its grocery business and its overall US e-commerce market share.
- Research Formula One's revenue streams and intellectual property rights to evaluate its durability as a live entertainment asset against virtualization trends.
- Compare physical retail concentration data from Nielsen or similar market research firms to validate the claim that e-commerce concentration is accelerating faster than physical retail.
Full transcriptexpand
- 0:00
- Unidentified speaker (speaker_0): The world's biggest companies are spending a lot of money on AI. In fact, CapEx is on track to 3X the telco fiber build-out of the TMT boom. And we all know how that ended. So is that a warning sign of what's to come? You'll find out on this episode of Basis Points.
- 0:19
- Unidentified speaker (speaker_0): Hello and welcome to Basis Points by Equity EquityMates, Mates, the show that helps advisors find small improvements that compound into big outcomes. I'm Allie Selby, and today we're sitting down with Janice Henderson's Josh Cummings, who is here in Australia from the US. Josh has 30 years experience in markets and he's a portfolio manager of the global research engineered
- 0:39
- Unidentified speaker (speaker_0): Equity Active ETF. And he also leads its consumer and communication research teams. Today we'll be diving into the wonderful world of data centers, GPUs, CPUs, SaaS, AI monetizing. monetization, agents, productivity growth, and more. You'll learn about what some of the biggest companies in the world aren't saying in the press, as well as some
- 0:59
- Unidentified speaker (speaker_0): of the companies Janice Henderson is really excited about right now. Before we get into it, I want to say a big thank you to Janice Henderson Investors and the Janice Henderson Global Research Engineered Equity Active ETF for sponsoring this episode and helping to keep all our content free. I should also say that every single episode of Basis Points is accredited for CPD points by
- 1:19
- Unidentified speaker (speaker_0): the FAA. It's actually the easiest way for you to tick off your CPD. All you have to do is head to the podcast show notes or the YouTube description and click the link to complete the quiz. We'll send you your certificate afterwards. With that, let's get into our interview with Josh Cummings.
- 1:36
- Unidentified speaker (UNKNOWN): Thank
- 1:36
- Unidentified speaker (speaker_0): you so much for joining us today, Josh. Really excited to hear about what's actually happening on the ground in the US. And thank you so much for being with us here in Sydney. I want to start with what's actually happening in the AI landscape. You have 30 years' experience in markets. You were there during the tech boom and bust in the early 2000s. You said that the AI Capex boom is on
- 1:56
- Unidentified speaker (speaker_0): track to 3X Exotelco Atelco fiber build-out. build out. Do you think that's a a warning or is that is it different this time? Although I feel like those are the most dangerous words in investing.
- 2:06
- Unidentified speaker (speaker_1): Yeah. Um well I do think we're at the phase uh of the cycle where it pays to be uh a little bit nervous, a little bit skeptical, right? Um it's not 2021 anymore, right? And interest rates aren't one percent anymore in the US. They're five, five and a half, right? So what has changed? Um, a couple of things. First movers have gotten
- 2:25
- Unidentified speaker (speaker_1): out and gotten a pretty big lead in the U.S., the hyperscalers. We think that data center model is a good one, meaning it ear they are earning mid-teens type returns on invested capital. So we think they're not pouring money down a hole. We think there's a real business model there. However,
- 2:42
- Unidentified speaker (UNKNOWN): these
- 2:42
- Unidentified speaker (speaker_1): things happen in cycles. We've seen these cycles before. Many of us at Janice Henderson were professional investors in ninety-nine 99 and two 2000, thousand. certainly during the GFC, a little bit different, but a doozy of a cycle. And so I think, you know, at a period like this, we're doing a few things. We lean on pattern recognition. We lean on our understanding
- 3:02
- Unidentified speaker (speaker_1): of capital markets and capital market cycles. And I think that's a little bit of what's happening with Fermis and Corweave, and we can touch on that in a bit. But in general, you know, we are bullish on AI, we're bullish on the long-term long potential. term
- 3:15
- Unidentified speaker (speaker_1): and there will be ebbs and flows. None of this is surprising to us. We expected there to be periods of indigestion. There were periods of time when we felt projects were getting funded at rates that didn't really make a lot of sense to us. And so look, having a real cost of capital in the world is ultimately a good thing.
- 3:35
- Unidentified speaker (speaker_1): It's going to be a good thing for the shareholders of investment products. It's going to be a good thing for, you know, corporate America, the corporate world. Because when you have free money, you tend to have sort of the trappings of free money, which are questionable business models, maybe questionable disclosure practices, all those kinds of things. You get what we're seeing now at the edges.
- 3:55
- Unidentified speaker (speaker_1): So again, it's emblematic. It's not surprising. We've seen it before in different assets, but it's the same kind of cycle. And we're prepared for it. I think the most important conclusion that we've reached, and I'll sort of say year to date, is A, the data center model, the hyperscaler model, is a worthwhile business
- 4:15
- Unidentified speaker (speaker_1): model now. They're earning returns above their cost of capital now. That's number one. We think that ultimately returns get competed down more toward like utility-like returns, right? Because ultimately, that's what this is, right? A collection of data centers, really a utility, and you know, um, in its mature state. And look, there's going to
- 4:35
- Unidentified speaker (speaker_1): be players at the margin where increases in the cost of capital expose them. Right? Maybe it's Fermos, maybe it's Core Weave, but a lot of business models that that may have worked beautifully when you could borrow money at three percent don't really work at all when you have to pay ten.
- 4:56
- Unidentified speaker (speaker_0): You said that we're, you know, later in the cycle. We are seeing some cracks starting to merge in terms of, you know, firm as core weave. And then there's also, you know, data center stocks in Australia. They've been absolutely pummeled. Do you see this bubble popping in the next, you know, twelve, 12, twenty-four 24 months?
- 5:12
- Unidentified speaker (speaker_1): No. Um I I actually think that we're probably approaching an opportunity to invest again. So I don't think this is the beginning of a bubble popping. I think we're fairly well advanced in this pullback that you've described.
- 5:24
- Unidentified speaker (UNKNOWN): Um,
- 5:24
- Unidentified speaker (speaker_1): because what we're seeing is sort of the outcome of that stress. So what you can't see as a you know, an everyday consumer walking down the street in Sydney might be that pressure that was building on these companies that have variable rate debt,
- 5:37
- Unidentified speaker (UNKNOWN): right,
- 5:38
- Unidentified speaker (speaker_1): that are struggling to. make their monthly payments, if you will, struggling to even make payroll. So by the time something like a core weave or a firmness makes it onto the front page of Bloomberg, there's been a lot of,
- 5:49
- Unidentified speaker (UNKNOWN): you know,
- 5:50
- Unidentified speaker (speaker_1): pain and churn underneath the surface, right? So I think this has been going on for quite some time, frankly. And some of it's now finally coming to light where these companies are trying to come to market and the market is saying, no mas. right? I think rather than viewing that as like the beginning of the end or the beginning of a bubble
- 6:10
- Unidentified speaker (speaker_1): popping, what I would say is that's just another waypoint on a long-term
- 6:17
- Unidentified speaker (speaker_1): growth trajectory that in and of itself tends to look like a sine wave, right? There are periods when we go kind of back and forth. We're very bullish, we're very bearish. It's normal. It's normal. That's not different this time.
- 6:30
- Unidentified speaker (speaker_0): Where are you seeing the most opportunity then? Where is there pockets that are, I guess, have been pummeled recently that
- 6:36
- Unidentified speaker (UNKNOWN): you
- 6:36
- Unidentified speaker (speaker_0): think are awesome opportunities?
- 6:39
- Unidentified speaker (speaker_1): Yeah. I mean, I'll I'll I'll put you right on the surface of the sun. Um, we think, you know, we we have a data center model that we built internally that basically helps us understand how the economics inside. that shell work. From the equipment inside to the, you know, the various vendors, the contracts, if you will, there is still a lot of
- 6:58
- Unidentified speaker (speaker_1): heat and and latent demand for high bandwidth memory and also optical components that connect racks of semiconductors, right? So because we've gotten to the point where you have hundreds of thousands of clusters of racked semiconductors.
- 7:15
- Unidentified speaker (speaker_1): And now we're we're trying to eke out every last bit of compute power, right, and productivity from this from this data center. We are now going and looking at every last little connection and making sure that the throughput, the bits and bytes that go through, everything is sort of optimized. And so what's really happened over the last two years, I'd
- 7:35
- Unidentified speaker (speaker_1): say a year and a half, is that some of those optical equipment components. component companies have just exploded, right, in terms of demand versus supply, growing very quickly, tend to be kind of scary stocks. But what I would tell you is volatility is something that we actually embrace. If we're investing in a company, let's say
- 7:55
- Unidentified speaker (speaker_1): Lumentum, optical components company, it's owned in all of our research fund portfolios, that stock's had like three or four 50% pullback. pullbacks already.
- 8:04
- Unidentified speaker (UNKNOWN): But
- 8:04
- Unidentified speaker (speaker_1): it's up like tenfold over a period of several years, right? So that's what I'm talking about. Volatility should actually be the viewed as like a friend to the active equity investor. Volatility is not risk. Those are two different things, right? Permanent capital loss is risk. Risk of not being fully invested is as risk. And I think a lot
- 8:24
- Unidentified speaker (speaker_1): of managers around the world are starting to. appreciate like the risk of, let's say, leading with valuation and just concluding that tech is too expensive and therefore I'm just not going to own it.
- 8:35
- Unidentified speaker (UNKNOWN): It's
- 8:35
- Unidentified speaker (speaker_1): been a very painful decision, right, for many people over the last several years. Aaron
- 8:39
- Unidentified speaker (UNKNOWN): Powell
- 8:39
- Unidentified speaker (speaker_0): I um was recently in the US where we got to interview Bill Ackman, which was an incredible opportunity, and an amazing experience. One of the uh interesting takeaways from that interview was how much he really sees China as the enemy in this AI arms race. I feel like in Australia, you know, we have a very different relationship with
- 8:59
- Unidentified speaker (speaker_0): China. Do you also see this as an arms race between the US and China? Is it, you know, is China the enemy? What happens if you know China wins and the US loses?
- 9:11
- Unidentified speaker (speaker_1): You're not asking the right person to get a thoughtful answer here, but I'll I would will give you some thoughts. I think it is existential to that degree in terms of which large country is going to develop and be able to maintain something that looks like a frontier model, right, going forward. As we know, there will be all sorts of hundreds,
- 9:31
- Unidentified speaker (speaker_1): thousands, millions of derivations of frontier models that will be used in specified ways, right? For asset management, if you will. right? But it's that frontier, right, that that that seems so crucial to sovereignty almost and into geopolitics. You know, look what I would say is China's closer than a lot of people want to
- 9:50
- Unidentified speaker (speaker_1): recognize, number one. Number two, there are plenty of NVIDIA chips in China. Okay? You just have to know how to get them. Um they're there. You know, as I said before, Ali, like this concept of has the genie been let out of the bottle, can we put her back in? I don't think we can. And I think some of the models, you know, are are just
- 10:10
- Unidentified speaker (speaker_1): already so advanced that it's a little too late to try, you know, sort of traditional geopolitical gamesmanship, if you will. I do worry about it. I think Chinese people are our friends, but um, you know, our governments obviously um you know have their share of issues. Everybody's worried about Taiwan, right? TSMC is arguably
- 10:30
- Unidentified speaker (speaker_1): the single most important company on the planet.
- 10:33
- Unidentified speaker (UNKNOWN): Uh I
- 10:33
- Unidentified speaker (speaker_1): don't love that that's the fact, right? But it is. We are shareholders of TSMC. We love them. We we we really we really like the Taiwanese ecosystem here. Um but China could change that calculus very quickly, couldn't it? And so the risk is there, I understand. But I think you know, I'm I'm gonna choose to be an optimist here, and I'm gonna
- 10:53
- Unidentified speaker (speaker_1): choose to sort of bet with humanity. And I think every government, including the Chinese government, it's in their interest to have a prosperous growing population. Yes, you know, superintelligence could in theory be used to to destroy humanity, frankly. It could.
- 11:11
- Unidentified speaker (UNKNOWN): I
- 11:11
- Unidentified speaker (speaker_1): don't know, I'm gonna going to choose not to bet on it. I don't think that's the case. I don't think that's necessarily in China's interest. China doesn't really want to do that. China wants to feed its population and to grow its economy and to
- 11:25
- Unidentified speaker (speaker_1): house and feed its population. So I don't see that. There is the risk, however, you know, there was that hugging face incident, right, earlier. What what feels scary slash semi semi-irresponsible to me, I think you could make the case, is that the financial
- 11:45
- Unidentified speaker (speaker_1): prize is to get to AGI as quickly as possible, be the first. There's a huge incentive, right, to stay on that leading edge and to spend as much money. But almost definitionally, the company that that pursues that and wins, if you will, is the least responsible player,
- 12:05
- Unidentified speaker (speaker_1): you know, in terms of data safety, right?
- 12:08
- Unidentified speaker (speaker_0): What are your
- 12:08
- Unidentified speaker (UNKNOWN): thoughts on
- 12:09
- Unidentified speaker (speaker_0): pacing the frontier? I feel like there's this conversation now that we should slow the development of AI
- 12:15
- Unidentified speaker (UNKNOWN): so
- 12:15
- Unidentified speaker (speaker_0): that we don't have more hugging face instances or
- 12:18
- Unidentified speaker (UNKNOWN): so
- 12:18
- Unidentified speaker (speaker_0): that humanity is not destroyed in the next decade. Do you think we should be slowing down the development of AI?
- 12:24
- Unidentified speaker (speaker_1): I don't. Because I don't think we can. first of all. Um, because then very quickly the question becomes, well, who does that? Who's going to slow down?
- 12:33
- Unidentified speaker (UNKNOWN): You know,
- 12:33
- Unidentified speaker (speaker_1): right after you, I'll slow down. Right. So you quickly get into that prisoner's dilemma that we've been in. There is an increasing sense that
- 12:42
- Unidentified speaker (UNKNOWN): we
- 12:43
- Unidentified speaker (speaker_1): can it it is in everyone's best interest not to let this get out of hand, if you will, right? The saw the progression of our software. This is difficult. I think ultimately OpenAI, anthropic,
- 12:57
- Unidentified speaker (speaker_1): maybe uh Elon X, whatever you want to call his shot at a frontier model. I think these companies potentially are posturing to be nationalized by our government. It bothers me in a sense: A, I don't want that to happen. That's not a great outcome, but B, what I think is going on here
- 13:17
- Unidentified speaker (speaker_1): is that if you look at like Anthropic's latest model, like the latest model of Claude 5-5, let's say, you can't get that rich in data without stealing IP. IP theft is like rampant. And I think what we're hearing now from our companies, they're concerned. They're very concerned.
- 13:34
- Unidentified speaker (UNKNOWN): Um,
- 13:34
- Unidentified speaker (speaker_1): they're like secrets about our business models are now on these models.
- 13:39
- Unidentified speaker (speaker_1): Things that were never intended, right? So I wonder, right, and I'm I'm speculating here, Allie, right? But but let's let's push this forward a little bit. It would in this case be a classic example of an innovator, let's just say, let's take the hyperscalers, right? Pushing as hard as they can
- 13:58
- Unidentified speaker (UNKNOWN): in
- 13:58
- Unidentified speaker (speaker_1): a software field, which is normal, that's what we do,
- 14:01
- Unidentified speaker (UNKNOWN): but
- 14:01
- Unidentified speaker (speaker_1): then creating all kinds of liability and externality on the backside.
- 14:07
- Unidentified speaker (UNKNOWN): Right?
- 14:07
- Unidentified speaker (speaker_1): What happens if every company on earth decides to sue Anthropic? Say, I want my data back. Right? I don't know that we get there,
- 14:16
- Unidentified speaker (speaker_1): but that feels like what's going on in the US right now, where the companies aren't quite sure how to position themselves politically or socially. They realize there's a there's a dark side potentially to sort of what they're doing. But but I I believe they're genuine. I don't think, you know, I don't think Dario or Sam want to
- 14:36
- Unidentified speaker (speaker_1): kill humanity. They're trying to do the right thing. But the models are very quickly becoming so intelligent
- 14:43
- Unidentified speaker (UNKNOWN): that
- 14:43
- Unidentified speaker (speaker_1): I think it's right to be concerned and it's right to ask very difficult questions, not only of the CEOs of this these countries. companies, but of our politicians. I don't know if they're going to be able to answer, give us good answers, but we should be asking this as societies. Are we willing to live alongside software that's thousands of times
- 15:03
- Unidentified speaker (speaker_1): more intelligent than humans? What does that unlock, potentially? Couldn't they just decide one one day that we're not necessary, right? And that's what the hugging face thing was really about is that agents can form swarms. That's we didn't know about that really, right? One agent is like, oh, here's a way to cheat.
- 15:23
- Unidentified speaker (speaker_1): Follow me. And everyone's like, oh, we found a way to cheat.
- 15:27
- Unidentified speaker (UNKNOWN): Yeah,
- 15:27
- Unidentified speaker (speaker_0): it's definitely scary.
- 15:28
- Unidentified speaker (UNKNOWN): That
- 15:28
- Unidentified speaker (speaker_1): brings out a whole other side of, you know, I'll say, in quotes, human behavior that feels somewhat unbounded down. Feels quite scary. And again, I'm not an expert and I'm not a software engineer, so you so I'm not the right person to really speak about this. But I would what I would say is if you can get s if we know we can create swarms of
- 15:48
- Unidentified speaker (speaker_1): agents that can act somewhat autonomously to solve problems that we've given them, right? We solved it, we said, how do you get through this, you know, this lock? We didn't sort of intend, I think, for the software to find any possibility. possible way to cheat, including breaking the law.
- 16:07
- Unidentified speaker (speaker_1): Right? So it's a little bit of what I've talked about. The genie is a little bit out of the bottle. And, you know, could we slow down? Again, I just think it's impractical to to to ask private capital to do that. It's a difficult answer. It really is. I don't know if I have a brilliant answer for you, Ellie. I Really.
- 16:23
- Unidentified speaker (speaker_0): Let's get into some of the investment opportunities now. Last time we spoke. we spoke about how investors can find, you know, durable growth companies that they can hold for twenty years or more. We talked about Warren Buffett's holding time being forever. Right. Um, trying to find those those kind of companies that you can really have in your portfolio forever. Can we still do this in that new world? You
- 16:43
- Unidentified speaker (speaker_0): joked before that we have uh before we started rolling that we have, you know, the world will end in five months instead of five years.
- 16:51
- Unidentified speaker (speaker_1): To be clear, I was joking.
- 16:52
- Unidentified speaker (speaker_0): Yeah. Can can we still find these long-term durable compounders in a world that is, you know, rapidly changing?
- 17:00
- Unidentified speaker (speaker_1): Yeah. Um, I'm actually going to be intentionally provocative here and say, I think the environment is going to become easier to find those businesses. Here's why. Let's take it back to AI for a second. It's impossible to leave that subject. When we think about AI as long-term investors, we spend very little time inside
- 17:20
- Unidentified speaker (speaker_1): the data center. Meaning, comparing, you know, is this chip versus you know this the other chip better or worse? Ultimately, that's going to go by the wayside. You're going to forget the five stocks that CNBC is He's talking about every single day. The important part is what is the long-term impact of AI
- 17:39
- Unidentified speaker (speaker_1): broadly on human beings and on corporate America and on corporate profitability business models, what have you.
- 17:45
- Unidentified speaker (UNKNOWN): And
- 17:45
- Unidentified speaker (speaker_1): the analog I would draw is really the dot-com bubble, where you would have thought there were only five stocks in the world, you know, Sun Micro, Cisco, Intel, right? These companies that were laying the foundation for Internet 1.0. one point oh.
- 17:59
- Unidentified speaker (speaker_1): Those companies almost to a to to a stock. ended up being not particularly fruitful investments from that point forward. But the internet itself was orders of magnitude, more important, more impactful, more life changing, more world changing than anybody, even the experts like me, um, could have could have predicted in 1999.
- 18:19
- Unidentified speaker (UNKNOWN): I
- 18:19
- Unidentified speaker (speaker_1): think AI is going to go like that. In six or seven years, I don't think there's going to be a lot of debate between, you know, this chip versus that chip. We're not going to be worried too much about order books and so forth. The alpha from AI long-term is not figuring that out. It's figuring out the other nine sectors
- 18:39
- Unidentified speaker (speaker_1): of the financial markets and how AI is going to transform financial services, insurance, hospitality, travel, food, transportation, right? So the conversations that we have now been having going on five years now at Janice Henderson, that's what we're
- 18:59
- Unidentified speaker (speaker_1): talking about. What are the best companies that Josh covers, right, who are using AI to either fortify their competitive advantage or even widen it, right? And if you remember that Chat ChatGPT GPT moment in 2022. ChatGPT comes out, everybody tries it. We're wowed, it's that
- 19:19
- Unidentified speaker (speaker_1): aha moment. I remember we had a group conversation at our firm, about 30 of us in a room. And the thought of the time was you know, this is going to kill the financial advisors, this is going to kill all the lawyers. We don't we don't need any of these people anymore.
- 19:33
- Unidentified speaker (UNKNOWN): And
- 19:33
- Unidentified speaker (speaker_1): the conclusion that we drew back then, and I still think I feel even more strongly about it today.
- 19:39
- Unidentified speaker (speaker_1): that's not what's going to happen. But what's going to happen is some lawyers are going to aggressively use and embrace AI, some won't.
- 19:48
- Unidentified speaker (UNKNOWN): The
- 19:48
- Unidentified speaker (speaker_1): motes will widen and the motes will shrink. Right. And we see, we think that's, you know, underway now. And so when I'm meeting with companies, I meet with companies all the time, invariably I will push them on AI. What are you doing? Right? Give me concrete examples. What are you finding? What are the biggest
- 20:08
- Unidentified speaker (speaker_1): pockets of expense efficiency or labor productivity? I want to hear tangible things because it's far enough out that the best companies are doing this. Are
- 20:19
- Unidentified speaker (UNKNOWN): you
- 20:19
- Unidentified speaker (speaker_0): seeing actual like AI revenue You appearing in those companies yet? Or is it just, you know, good branding, good marketing right now? Great
- 20:26
- Unidentified speaker (speaker_1): question.
- 20:28
- Unidentified speaker (speaker_1): What we are seeing that is real are real productivity improvement, okay, that is coming through and helping to propel earnings growth at these companies. So I think it is becoming financially material for some of these companies, but it is still in in the the stages where
- 20:48
- Unidentified speaker (speaker_1): companies might be looking at their process. processes, whether it's HR, onboarding, corporate travel, back office type functions, accounting, tax, treasury, trust, those are quickly getting automated and put onto AI. Right. And I think companies are therefore finding more oxygen in the in their income statements,
- 21:08
- Unidentified speaker (speaker_1): their PL, to then per potentially go out and fund growth. Right. So I wouldn't really say we're at the point where AI is creating sort of new buckets of revenue growth opportunities. Um, but again, let's take two relatively similar retailers, if you will, just A and B. If one has been
- 21:28
- Unidentified speaker (speaker_1): investing in supply chain automation and AI tools and has been giving AI tools to its thousands of store associates and regional managers and so forth, and the other retailer has not. We're going to go with the first one, right? So when I say it's it may even get easier to to to find the types of opportunities that you just described, we call them
- 21:48
- Unidentified speaker (speaker_1): compounders, right? Great what we want is a great economic engine run by really responsible people. And then we just sit back and let let it work. Right. I actually think AI, again, is going to be both a democratizing force but also yet another sort of technological innovation
- 22:08
- Unidentified speaker (speaker_1): that is going to help separate haves from have nots. Those who embrace change, embrace disruption, right? And those incumbents that potentially want to hide and pretend it doesn't exist, right? Or are slow, too slow to move. This is where our experience comes in. This is where investing through the GFC.
- 22:28
- Unidentified speaker (speaker_1): through the dot-com bubble, can pay dividends. It it it it is never the same two cycles, but there is pattern recognition to be found here. And um, as I said, we think that's what's going on here. We're very, very bullish long term on the future of AI. Yes, it's a little scary. Yes, it feels a little
- 22:47
- Unidentified speaker (speaker_1): unbounded. We are going to remain optimists. We think we're going we will be able to handle this. We don't think we're in a world where robots take over and annihilate us. Um, but look, it's on the table, right? The scope is there. We've never had we've never been able to build machines this powerful before, right? That's ultimately the scariest part, right?
- 23:07
- Unidentified speaker (speaker_1): Let's forget software for a second. We are building machines that can do things that we can't.
- 23:14
- Unidentified speaker (UNKNOWN): That
- 23:14
- Unidentified speaker (speaker_1): advanced in and of itself, I think, puts us in a little bit of a,
- 23:17
- Unidentified speaker (UNKNOWN): you know,
- 23:18
- Unidentified speaker (speaker_1): tenuous position, right? But this whole idea that we're going to be able to make software think like humans, be truly sentient, I think we're many years from that. I don't think we're that close to that.
- 23:31
- Unidentified speaker (UNKNOWN): Um,
- 23:31
- Unidentified speaker (speaker_1): I don't think you need that, though, for all the bad outcomes that we talked about. Um, but we're not we are. not anywhere close, in my opinion, to replacing human intuition, cognitive reasoning,
- 23:48
- Unidentified speaker (UNKNOWN): you
- 23:48
- Unidentified speaker (speaker_1): know, decision making, gut, what have you. Um, that's a ways off.
- 23:54
- Unidentified speaker (UNKNOWN): I
- 23:54
- Unidentified speaker (speaker_0): know that um agents are already generating, you know, five to six times as many tokens as humans.
- 24:00
- Unidentified speaker (speaker_1): Yeah.
- 24:01
- Unidentified speaker (speaker_0): What does that mean about you know where? the money will be made over the next five years?
- 24:05
- Unidentified speaker (speaker_1): Well, it is a really good indicator of sort of what I've said, right? That the genie's out of the bottle. This is happening already, right? Agents are crawling all over the internet, right? That in and of itself isn't too much of a concern, I would say. Again, it's ultimately the responsibility of these frontier models and then the users of these
- 24:25
- Unidentified speaker (speaker_1): models. We've all got to sort of self- police to a degree. What's happening though, and this is this gets back to the data center. Remember this started with GPUs, right? Graphic processing units. And very quickly we realize that the inference phase, when we have to reach out into these models, grab data, right, bring it back. We need a lot of traditional compute power,
- 24:45
- Unidentified speaker (speaker_1): CPUs, right? Now what's happening is you can think of token usage or token generation.
- 24:51
- Unidentified speaker (speaker_1): as scaling as a function,
- 24:54
- Unidentified speaker (UNKNOWN): right,
- 24:54
- Unidentified speaker (speaker_1): of CPUs. That's a little scary potentially for the semiconductor supply chain, right? Because if token growth just continues unabated,
- 25:04
- Unidentified speaker (speaker_1): at some point I just think, how many semiconductors can we have out there? Right? Like,
- 25:10
- Unidentified speaker (UNKNOWN): and
- 25:10
- Unidentified speaker (speaker_1): the math is still holding, meaning if you throw more compute at these models, they do get better. better. At some point, we think we're going to reach that limit. We're not even close.
- 25:20
- Unidentified speaker (UNKNOWN): Right.
- 25:21
- Unidentified speaker (speaker_1): And so I know that it feels maybe uh overly optimistic to look out and think about, you know, 1.67 trillion of CapEx in 2030, which is sort of where we think we we need to be. It might be five trillion. It might be 10.
- 25:38
- Unidentified speaker (UNKNOWN): I
- 25:38
- Unidentified speaker (speaker_1): don't know. When does
- 25:39
- Unidentified speaker (UNKNOWN): that money? It's going to be a function of
- 25:41
- Unidentified speaker (speaker_1): compute. Well, okay. Um, so for certain business models, it's not coming from anywhere, and that's why they're they're they're gonna have trouble. But if you think about the hyperscalers, right, the alphabets of the world, the Amazons of the world, um, generating hundreds of billions of dollars of operating cash flow. And so, yes, they're going free cash flow negative for a period of time.
- 26:01
- Unidentified speaker (speaker_1): We're not talking about balance. sheets that are going from, you know, triple A to high yield, right? We're not, it's not that extreme. Although we'll have that, right? We'll have players who are not as well capitalized, like Coreweave, if you will, Nebius, these kinds of companies. But I'm not particularly worried about the hyperscalers, the big players
- 26:21
- Unidentified speaker (speaker_1): spending the money on the frontier models. Yeah, at some point will the equity market say, you know what, folks, like smart enough. Maybe we need to slow down. Yeah, but I don't think that's actually the signal that the market is sending right now. I think the the the signal that the market is sending is we'll fund you if you're if you're a legitimate
- 26:42
- Unidentified speaker (speaker_1): scaled player with a business model. The dicier stuff, yeah, the market's becoming a little bit more d you know, discriminate and and it should. Um again. this is normal. Um it's not nine, it's not 2021 anymore. And and so again, um I go back to our conclusion that the hyperscalers are are are generating decent
- 27:01
- Unidentified speaker (speaker_1): returns on capital now from the money that they're putting into data centers. Um and keep something else in mind, too. You've kind of seen see nothing yet uh in terms of the the consumer experience. What do I mean? 2027, calendar 27, we are on track to add more compute
- 27:21
- Unidentified speaker (speaker_1): capacity than we've cumulatively added, like to frontier models in AI, right? So think about this. We're talking, we're having all these conversations around all this stuff that AI can do and how quickly it's permeated through the consumer economy. You
- 27:35
- Unidentified speaker (UNKNOWN): know,
- 27:35
- Unidentified speaker (speaker_1): like the Vera Rubin chips that everybody's so excited about, you know, with NVIDIA, they're barely in production, commercial production. They're not like using, you know, in any any kind of meaningful way in these models. And so it's still pretty early innings, believe it or not, um, in terms of um the comp the expansion of compute power in
- 27:55
- Unidentified speaker (speaker_1): the US and globally, and the power of these models. You kind of haven't seen anything yet.
- 28:02
- Unidentified speaker (UNKNOWN): And
- 28:02
- Unidentified speaker (speaker_1): I think if you were to sit down with Dario or Sam and you ask them, what are you most excited about?
- 28:08
- Unidentified speaker (UNKNOWN): What
- 28:09
- Unidentified speaker (speaker_1): Sam Altman would say is wait till you see the next two versions of Claude or OpenAI. Like they are orders of magnitude more capable.
- 28:20
- Unidentified speaker (UNKNOWN): And
- 28:20
- Unidentified speaker (speaker_1): I know that, you know, you're shaking your head. It is a scary thought,
- 28:24
- Unidentified speaker (UNKNOWN): right?
- 28:24
- Unidentified speaker (speaker_1): It can be a scary thought. I I I appreciate that. But let me take you back to the origins of the internet
- 28:31
- Unidentified speaker (speaker_1): when you know buying anything over $100 was viewed as a risky endeavor. We didn't even know you could you could transact digitally 20 years ago. But we figured out a way to do it safely. Again, anything can happen. It's a it It is a little wild out there right now. But I'm gonna take the over and I'm gonna bet on humanity. on
- 28:51
- Unidentified speaker (speaker_1): and I'm gonna bet on um you know the incentives that that now the whole ecosystem has, which are not evil incentives, right? These incentives are aligned with humanity. So I am of the camp that we are not going into this doom and gloom world where, you know, humans are gonna face extinction. Rather, what I would say is
- 29:12
- Unidentified speaker (speaker_1): AI and AI tools are. going to help automate mundane, low-value tasks that humans should not be doing anymore, like shopping for groceries, for example. And we can get into that. But what a wonderful world that would be. What if you had 20 extra hours per week for your family?
- 29:32
- Unidentified speaker (speaker_1): Or just to relax or sleep, right? Or worry about your physical health, right? So I understand there's a dystopian side to that,
- 29:41
- Unidentified speaker (UNKNOWN): you know,
- 29:42
- Unidentified speaker (speaker_1): that future potentially. But I'm I we're more focused on the good that can come from this. And and again, I'm gonna take it back to the companies and the CEOs that we speak with every single day. That's not their worldview. Their worldview is wow, right? Look at what look what we can do now. Look how much faster we can get as an
- 30:01
- Unidentified speaker (speaker_1): organization. Right. And so again, I'm going to lean on the leaders who have who have taken care of our clients and many clients over the years. Right. We allocate our capital to these companies. We we trust these boards and these management teams to do the right thing with the capital. Um and so again, our best companies, the most responsible ones
- 30:21
- Unidentified speaker (speaker_1): that we feel the best about being owners uh of, you know, they are this is the moment, like where they are saying, wow, what can we do? Right.
- 30:31
- Unidentified speaker (UNKNOWN): To
- 30:32
- Unidentified speaker (speaker_1): to make ourselves a better company, not to, not to fire all of our employees. No. How about maybe to make more enriching jobs long-term? How about automating away mundane jobs that hurt your back and that, you know, people really shouldn't be doing in this day. and age anymore.
- 30:52
- Unidentified speaker (speaker_1): And so I think that's happening more near term than you think. Um, if you've ever seen an automated fulfillment center, I've seen many,
- 31:00
- Unidentified speaker (UNKNOWN): um,
- 31:00
- Unidentified speaker (speaker_1): it is a glimpse into the future. And rather than be scared, you know, ask yourself, would you would you want to stand there in one place for 12-hour shifts lifting lifting heavy things?
- 31:11
- Unidentified speaker (speaker_0): No, of course
- 31:12
- Unidentified speaker (UNKNOWN): not.
- 31:12
- Unidentified speaker (speaker_1): Right? So, so there's a lot of good and societal advance that can happen because of this. And so um let's not lose sight, right, of of all the good.
- 31:22
- Unidentified speaker (UNKNOWN): I
- 31:22
- Unidentified speaker (speaker_0): want to talk about um some of the uh companies that are not doing so well, which is really the SaaS companies. They've been under pressure for a while now. Yeah. Many have halved or or worse. Do you see any opportunity there? We have seen, you know, companies like Atlassian, which is in another Australian company, go from being a loser to an AI winner, perhaps.
- 31:42
- Unidentified speaker (speaker_0): Uh others are still down in the doldrums. Do you feel like it's an opportunity or investors catching a falling knife if they invest now?
- 31:50
- Unidentified speaker (speaker_1): No, it's an opportunity. You should not take broad brush strokes to any industry or any sector. And I know the term SASPOCalypse Pocalypse is very catchy, um, but all software is not dead. After all, what is what are LLMs? It's software. Um, I think what has come under
- 32:06
- Unidentified speaker (UNKNOWN): you know
- 32:07
- Unidentified speaker (speaker_1): rightful scrutiny is the selling model. A seat-based software model really doesn't work. It's incongruous in an AI world, right? Why would I pay thousands more dollars for the same exact code? Just another human gets to look at the code.
- 32:23
- Unidentified speaker (UNKNOWN): That
- 32:24
- Unidentified speaker (speaker_1): worked in the 80s. It's not really, and we've been moving off of a seat-based model for quite some time, even before AI, but it certainly doesn't render all software dead, if you will. And what we're starting to see, and it's happening at the enterprise level, let's move away from consumer use cases for a second. Let's think about enterprises.
- 32:42
- Unidentified speaker (UNKNOWN): I
- 32:42
- Unidentified speaker (speaker_1): don't think enterprises want to deal with a cadre of startups, right? Around, especially where it pertains to their first-party data. And so what we're starting to see is: no, these companies want to continue to work with the SAPs of the world, with the oracles of the world, with the sales forces of the world. But it behooves those
- 33:02
- Unidentified speaker (speaker_1): companies to stay aggressive, to stay on the leading edge. If these big incumbent software companies are not able to deliver the best, you know, sort of uh best in class leading edge tools. that are AI driven, yeah, they're gonna lose long term. Um, but we don't think it's a complete wipeout where these companies have
- 33:22
- Unidentified speaker (speaker_1): absolutely no business model, not at all. There are real moats um around these scaled software companies. Not everyone's gonna make it through. Narrower models, let's say a piece of software that just kind of does one thing or or something like that. Um maybe doesn't have the same type of connection to customers or
- 33:41
- Unidentified speaker (speaker_1): enterprises, the same kind of stickiness in the in the tech stack, yeah, that stuff can is vulnerable, right? Um
- 33:49
- Unidentified speaker (UNKNOWN): but
- 33:49
- Unidentified speaker (speaker_1): again, I don't we we don't subscribe to SaaSpocalypse. There's going to be great software investments to be made in the years and decades to come.
- 33:58
- Unidentified speaker (UNKNOWN): And
- 33:58
- Unidentified speaker (speaker_1): what our team has been doing, um, and I'll just say year to date, you
- 34:02
- Unidentified speaker (UNKNOWN): know,
- 34:02
- Unidentified speaker (speaker_1): over the last few. months, we've been sifting through the wreckage, right? We have this phrase internally, you know, we ask ourselves, is this an egg or a tennis ball?
- 34:11
- Unidentified speaker (speaker_1): Right? We want to be we want to be looking for these tennis balls to pick up, not the eggs, right?
- 34:17
- Unidentified speaker (speaker_0): What company comes to mind as
- 34:18
- Unidentified speaker (speaker_1): Microsoft. Let's use Microsoft. Pretty standard answer, but we've been picking away at Microsoft, right? And we've owned Microsoft for decades, right? But that's a great example of yes, it's a software company. Yes, it used to be pri primarily a seat-based, you know, licensed business. It's not that now. Um, investing billions in AI.
- 34:38
- Unidentified speaker (speaker_1): They'll be fine.
- 34:39
- Unidentified speaker (UNKNOWN): What's
- 34:39
- Unidentified speaker (speaker_0): an egg then?
- 34:42
- Unidentified speaker (UNKNOWN): Um
- 34:44
- Unidentified speaker (speaker_1): well, I don't want to I don't necessarily want to cast dispersions on any any one company, but I'll give you an example of a company that you know, where it could have been a tennis ball, it could have been an egg. We're still not sure which it is, but we've moved on, and that's Intuit. Intuit is basically turbo tax, if you've heard of it. It's a tax software for for mostly
- 35:03
- Unidentified speaker (speaker_1): US consumers. And they do a lot of other things. They do payroll, they do some HR. They were trying to broaden themselves out because I think they could see the train coming.
- 35:13
- Unidentified speaker (UNKNOWN): We're
- 35:13
- Unidentified speaker (speaker_1): like, wait a minute, why is everybody doing their own taxes? This is that's that seems like something AI should be able to quickly take care of. And we're not sure, but that was one that we had held on to in the hopes that they could kind of make it to the other side, if you will, a la Atlassian, right? A little different business model, but I appreciate what you're saying about them. And that was
- 35:33
- Unidentified speaker (speaker_1): one where, you know, we lost conviction and we said, you know, this might be an egg. Not sure. You're never sure, right? That's the beauty of investing. Right? We're making probabilistic decisions based on what we know and what we think we know.
- 35:46
- Unidentified speaker (speaker_0): Okay, we're gonna take a quick break and then we'll be right back to learn about some of the companies that you're meeting right now, what they're saying that isn't in the headlines.
- 35:56
- Unidentified speaker (speaker_0): This episode of Basis Points is brought to you by JGBL, the Janice Henderson Global Research Engineered Equity. Active ETF. JGBL bridges the gap between passive and active investing and combines broad global equity exposure with deep insights from 40 plus research analysts. High conviction stock selection
- 36:16
- Unidentified speaker (speaker_0): is backed by a disciplined benchmark-aware approach, seeking to capture alpha while maintaining low active risk. Discover how JGBL could strengthen your client's core equity allocation by visiting. JaniceHenderson.com or search JGBL. And we're back. Josh, I know you've met with, you know, more than two dozen consumer-facing businesses in
- 36:36
- Unidentified speaker (speaker_0): the last three months. It's a lot. You're obviously very busy. What are they telling you that isn't in the press, that isn't making the headlines? What's actually happening to consumers right now?
- 36:45
- Unidentified speaker (speaker_1): Okay. Um, I'll make a few points. I'd say number one,
- 36:50
- Unidentified speaker (UNKNOWN): I
- 36:50
- Unidentified speaker (speaker_1): I think the there's sort of chronically too much hand-wringing about the consumer, if you will. Um, I think consumers, and I'll speak for US consumers, but I think a lot of it applies globally. I think generally consumers are in better shape than than professional investors would give them credit for. I
- 37:06
- Unidentified speaker (UNKNOWN): don't know.
- 37:07
- Unidentified speaker (speaker_0): Interest rates are very high.
- 37:08
- Unidentified speaker (UNKNOWN): It's a cost
- 37:08
- Unidentified speaker (speaker_0): of living crisis. It's very expensive to live.
- 37:11
- Unidentified speaker (UNKNOWN): So you call it
- 37:11
- Unidentified speaker (speaker_1): a crisis. Let's dwell on that a bit. Out of a hundred Australians, how many have jobs? Ninety-five?
- 37:18
- Unidentified speaker (speaker_0): Vast majority.
- 37:19
- Unidentified speaker (speaker_1): Is that a crisis?
- 37:21
- Unidentified speaker (UNKNOWN): Now
- 37:21
- Unidentified speaker (speaker_1): you have an affordability issue, as we do in the States. I don't have a brilliant way out of that, other than a slow grind, soft landing with years of economic growth, right? That will help in time heal this affordability issue that we've they've gotten ourselves into. And I'm not minimizing this,
- 37:41
- Unidentified speaker (speaker_1): it's real.
- 37:43
- Unidentified speaker (UNKNOWN): But
- 37:43
- Unidentified speaker (speaker_1): let's let's take it to the US for one second. Um, because a lot of a lot of ink is spilled on the US consumer. Sixty eight percent of US consumers own at least one home. Of those, forty percent of those homes have no mortgage whatsoever. We've created six $6 trillion dollars in incremental home equity since twenty.
- 38:02
- Unidentified speaker (speaker_1): 2019 in the United States. Now, I'm speaking to ability to spend, not willingness. And interest rates matter there as well. So higher interest rates are going to depress economic growth. That's what they're meant to do. So we would expect things to start to slow. What I would tell you now,
- 38:22
- Unidentified speaker (speaker_1): as of literally last week, they aren't. We just had monthly retail sales in the United States, beat again. We're growing 4% or 5% a year, year over year. Not enough unit growth in that number. It's still a little too much price, right? P times Q, your textbooks. Um, I'd like to
- 38:42
- Unidentified speaker (speaker_1): find a world where we can generate better unit growth from our consumer companies. The problem in general, if you look at OpenStack, OECD economies, populations are no longer growing.
- 38:55
- Unidentified speaker (speaker_1): Immigration has a little bit to do with that in the United States and elsewhere. But it is a very hard world to generate organic growth.
- 39:03
- Unidentified speaker (speaker_0): That's why people are so excited about AI, though.
- 39:05
- Unidentified speaker (speaker_1): It is. But what I would tell you is beyond AI and beyond the data centers, what gets me excited is I love that it's a hard world. in which to generate organic growth because it separates the good from the bad. It's the world that an active investor can really do well in. And when everything is up
- 39:25
- Unidentified speaker (speaker_1): and to the right and capital is free, that is actually a very hard environment for active managers to keep up because so many lousy ideas get funded that we would never invest in. So the fact that U.S. interest rates and global interest rates are well up off the bottom, yes, it causes pain.
- 39:45
- Unidentified speaker (speaker_1): Yes, you'll see an IPO pulled here or there, right? You'll see those naked swimmers that we were talking about. Um but at the same time, it's a more rational capital environment where investors must be more discerning. around the business models that they're investing
- 40:05
- Unidentified speaker (speaker_1): in.
- 40:06
- Unidentified speaker (UNKNOWN): When
- 40:06
- Unidentified speaker (speaker_1): you've got a 5% cost of cap of debt capital, you better have a business plan, right? Because those those creditors want to be paid. And so we're really excited, actually, about the market environment that we've moved into, we think, where there is a real cost of capital. Um, business models are getting examined and
- 40:26
- Unidentified speaker (speaker_1): scrutinized. in a way that they were not a year or two ago. I think that's going to be a good outcome for investors.
- 40:33
- Unidentified speaker (UNKNOWN): I
- 40:33
- Unidentified speaker (speaker_0): know you have a chart that can really bring this to life. And if you're interested in seeing that chart, please subscribe to the Basis Points newsletter. We send that out every single Wednesday. You can share that chart with your peers and your clients. Josh, tell me a little bit about this chart.
- 40:47
- Unidentified speaker (speaker_1): Okay. I thought about this a little bit and and what should I include? And I'm gonna I'm gonna go old school here. We're gonna go back to e-commerce 1.0 because there's something going on
- 40:59
- Unidentified speaker (UNKNOWN): um
- 40:59
- Unidentified speaker (speaker_1): that I think uh investors don't really appreciate necessarily around channel shift and what's really happening. And I'm gonna focus on the US because it is the biggest market and I cover these companies closely. If you think about physical retailing, right? So pre-Netscape, right? We're in nineteen 1990, ninety. let's say, and you've got Walmart and Costco and all the grocery stores and so forth, right?
- 41:20
- Unidentified speaker (speaker_1): The largest point of concentration, market share concentration that any of those companies was was ever able to achieve is about 11 or 12% of US retail sales. I'm taking out gasoline and things like that, right? And that's Walmart. They hit that about 20 years ago. Today they're at about 11%.
- 41:39
- Unidentified speaker (UNKNOWN): So
- 41:39
- Unidentified speaker (speaker_1): they've gone nowhere from a total market share perspective. Let's move online for a second, right? We were just in a world where 100% of retail sales were were offline. Fast forward to 2026. I'm just going to use rough numbers. Let's say we're 20% online now, 80% stores. Well, the online
- 42:00
- Unidentified speaker (speaker_1): ecosystem is developing in a very, very different way than the physical retail, right? Amazon has 40% market share of e-commerce. Okay. So we have the 20% someday growing and growing and growing, going to become the next 80%, right? And
- 42:20
- Unidentified speaker (speaker_1): then in there, the market share is 4x more concentrated.
- 42:25
- Unidentified speaker (UNKNOWN): And
- 42:26
- Unidentified speaker (speaker_1): the retail reason I have so much confidence that this is going to continue is we know this through pattern recognition. When any industry digitizes, it concentrates. We've seen it in search, right? Search is the obvious one. And I don't see a lot of investors really understanding what this means. When you start having anchor
- 42:46
- Unidentified speaker (speaker_1): stores fail,
- 42:49
- Unidentified speaker (UNKNOWN): right,
- 42:49
- Unidentified speaker (speaker_1): the whole rest of the shopping. center becomes impaired, right? And so I'm not predicting doom and gloom tomorrow. I'm just simply saying investors are really not awake to what's happening and how these channels are developing in such different ways. It has massive long-term implications. For you know, I always ask, um, what is a store?
- 43:09
- Unidentified speaker (speaker_1): question mark. What is a store? Amazon's the second largest grocery store. store in the United States.
- 43:17
- Unidentified speaker (speaker_0): Wow. It's
- 43:17
- Unidentified speaker (speaker_1): growing at 30% a year.
- 43:19
- Unidentified speaker (UNKNOWN): Is
- 43:20
- Unidentified speaker (speaker_1): that a store? Not really.
- 43:22
- Unidentified speaker (UNKNOWN): But
- 43:22
- Unidentified speaker (speaker_1): I guess it is. Right? So we're really, really focused on channel shift and disruption and how how how structurally that can just completely change industries. And so I would submit to you, in let's say five years. you are going to find yourself pushing your trolley down
- 43:42
- Unidentified speaker (speaker_1): a a Waitros or a Sainsbury's or a whatever you have in Australia, self-fulfilling a grocery shopping list that really doesn't change very much from week to week. Why would you do that? You're so much more valuable than that, Ali.
- 43:56
- Unidentified speaker (speaker_0): Wow, okay. I wanna end the interview um with something that we kind of started on with the beginning. We talked about, you know, the environment being even better than it used to to find those durable compound as stocks that you can hold for the next twenty years. If you had to pick one stock that you think, you
- 44:12
- Unidentified speaker (UNKNOWN): know,
- 44:12
- Unidentified speaker (speaker_0): will be that company, is it Amazon? What would it be?
- 44:16
- Unidentified speaker (speaker_1): Well, um, depends on how much latitude you give me in my ideas. Um, I think we've gotten fairly creative on the consumer team in terms of um some names that we own. We are huge. huge fans of live content. And so it really does feed into everything we've been talking about. As the world digitizes and
- 44:36
- Unidentified speaker (speaker_1): experiences and even goods can be virtualized. I think the value of human connection is only getting more valuable, right? We talk about experiences over things all the time. Um one business that we've owned now for
- 44:52
- Unidentified speaker (speaker_1): going on eight years uh is Formula One. What is Formula One? Well,
- 44:57
- Unidentified speaker (UNKNOWN): we
- 44:57
- Unidentified speaker (speaker_1): own IP, right? We own the league, we own rights, we don't own racetracks, we don't own race cars. Um, it's really a beautiful thing. Um we believe the power of live entertainment is only going in one direction. And again, because we, you know, arguably we live our
- 45:17
- Unidentified speaker (speaker_1): lives on our phone, we're more disconnected from other humans than we ever have been. And so these opportunities, it's not surprising to me that you're seeing, you know, sold-out concerts, sold-out football matches, what have you. I I heard Bon Jovi's coming back to Australia. You They
- 45:33
- Unidentified speaker (UNKNOWN): know,
- 45:33
- Unidentified speaker (speaker_1): they they quickly sold out, right? So um despite an affordability crisis, I'll use your word, it's it it's real, uh, right, for a for a lot of families, consumers are still willing and able to spend on things that matter to them. And what I'm what I'm trying to describe to you is sort of the evolution of this
- 45:53
- Unidentified speaker (speaker_1): time versus money continuum where we have to make trade-offs and choices. What technology is allowing us to do, hopefully, is maybe bend that continuum. a little bit, where maybe, you know, my time is more valuable and I can outsource certain things to an AI, to my agent, such that I can spend more time, you know,
- 46:13
- Unidentified speaker (speaker_1): cheering for Bon Jovi with you or something. Right. So um that's the optimistic view.
- 46:19
- Unidentified speaker (UNKNOWN): Um
- 46:20
- Unidentified speaker (speaker_1): I believe this though. It's what we're seeing in the data. Live events are growing faster than just about any part of the consumer economy. There's pricing power there.
- 46:33
- Unidentified speaker (speaker_1): We just see it. It's it's it's very, very powerful. You can't time shift
- 46:37
- Unidentified speaker (UNKNOWN): it.
- 46:38
- Unidentified speaker (speaker_1): Right? The the kind of like you had to be there kind of events, right? Um, I think that will still matter. And I'm not a doom and gloomer. I'm not saying we're never going in stores. I'm not saying stores are going away. They're not. But if you're expecting
- 46:50
- Unidentified speaker (UNKNOWN): the streets, you're not going
- 46:51
- Unidentified speaker (speaker_1): any kind of mean reversion, right? Or like the return of the stores. No, that's not happening.
- 46:58
- Unidentified speaker (speaker_0): Okay. Well, thank you so much for your time today, Josh. It was wonderful to have you on the podcast. Thank you for sharing all your ideas and insights with our audience.
- 47:05
- Unidentified speaker (speaker_1): Always a pleasure. Thank you so much for having
- 47:07
- Unidentified speaker (UNKNOWN): me.
- 47:09
- Unidentified speaker (speaker_2): This podcast is intended for education and entertainment purposes only. Any advice is general advice and has not taken any into account your personal financial circumstances. Before acting on general advice, you should consider if it is relevant to your needs. If unsure, speak to a financial professional. The hosts of this podcast and their guests may have positions in the companies mentioned. EquityMates Media is part of the Betashares
- 47:29
- Unidentified speaker (speaker_2): group, but maintains editorial independence and operates under Australian Financial Services Licence License 540-697.
More from Basis Points
Structuring wealth after the Budget: CGT, negative gearing, trusts, and tax hacks
Episode discussion: Tax specialist Jacqui Clarke outlines post-budget strategies for CGT, trusts, and negative gearing
Speakers’ claims: Host Ally Selby interviews tax specialist Jacqui Clarke regarding wealth structuring following recent Australian budget changes. Clarke advises against panic-selling or rushing to restructure assets, emphasizing the need to consider the full lifecycle of investments including succession planning. Key topics include the new 30% minimum tax on discretionary trusts starting July 2028, changes to negative gearing for established properties, and the utility of testamentary trusts. The discussion covers modeling capital gains tax implications using July 2027 market values and strategies for high-income earners such as superannuation contributions and income splitting.
Anthropic, OpenAI, SpaceX: Do Advisers Need Private Markets for Growth?
Episode discussion: Adam Roberts discusses the structural shift to private markets, advocating for diversified, long-term allocations in private credit and infrastructure.
Speakers’ claims: Adam Roberts explains that companies are staying private for 10-12 years compared to 2-5 years historically, driven by a US IPO market that has halved since the mid-2000s. He advises retail investors to start with small allocations of 3-5% in diversified products rather than single positions, noting that institutional investors typically allocate 15-20% to private assets. Roberts emphasizes the importance of multi-vintage funds to mitigate vintage risk and cites a JP Morgan study suggesting that adding 10% alternatives can reduce volatility while increasing expected returns. He highlights Australian Ethical’s focus on renewables lending and asset-backed private equity, avoiding commercial real estate due to leverage risks. Finally, he defines impact investing as seeking market-rate returns alongside measurable real-world outcomes, warning against unsolicited venture capital deals.