Super-Spiked Podcast

Refining margins and power demand growth driven by geopolitics and infrastructure, not just AI

19 min Listen to the original Episode page
RefiningPower DemandGeopoliticsEnergy InfrastructureAI ImpactOil DemandUS EconomyRegulatory Risk

Summary

Arjun argues that elevated crack spreads are structural due to geopolitical disruptions and lack of new refinery construction, rather than temporary anomalies. He highlights a strategic shift in US refining competitiveness against a backdrop of global demand plateaus. The episode further posits that US power demand will grow structurally at 2-3% annually, driven by grid aging, electrification, and industrial reshoring, with AI being a secondary factor. Management quality in energy is praised for prioritizing balance sheets over volume, while AI tools are noted as aids for junior analysts.

Key takeaways

  1. 1
    The speaker’s firm estimates structural US power demand growth at 2-3% annually, rejecting higher street forecasts as overly optimistic, analogous to their 1 million barrels per day structural oil demand growth estimate.

    Arjun · 11:58

Episode sections

  1. 0:38 – 2:27

    Refining Macro and Geopolitical Disruptions

    Arjun discusses how crack spreads are elevated due to Russian refinery disruptions and Middle Eastern export shutdowns, arguing these spreads should not be normalized. He notes that new refinery construction in developing nations has slowed significantly, even absent geopolitical shocks, due to a 'net zero' mindset affecting investment decisions.

  2. 2:27 – 7:11

    US Refining Strength and Developing World Dynamics

    The speaker highlights a strategic shift in US refining, particularly on the Gulf Coast and in the Midwest, moving from weakness to strength due to reduced global competition. He advises that countries like India must balance expanding EV capacity with maintaining robust refining sectors to avoid energy dependency.

  3. 7:11 – 10:11

    Power Demand Growth Drivers Beyond AI

    Arjun asserts that US power demand will grow at a structural rate of 2-3% annually, driven by aging grids, electrification, industrial reshoring, and data centers. He distinguishes between consumer AI hype and the more impactful 'physical AI' (robotics, autonomous driving) as key long-term power drivers.

  4. 10:11 – 14:10

    Regulatory Scrutiny and Interest Rate Sensitivity

    The speaker argues that congressional scrutiny of tech executives is unlikely to alter the fundamental growth trajectory of data centers, citing historical precedents in banking. He notes that power utilities are highly sensitive to interest rates, suggesting that a cooling in tech spending might ease Treasury yields and benefit regulated utility valuations.

  5. 14:10 – 19:04

    Management Quality and AI Usage in Research

    Arjun praises current energy management teams for prioritizing profitability and fortress balance sheets over volume growth. He concludes by explaining his use of AI tools like Claude for visualizations and data processing, emphasizing that junior analysts use AI skeptically to enhance, not replace, their own judgment.

Claims worth verifying

Crack spreads are way above historic averages, driven by Ukrainian drone strikes on Russian refineries and Middle Eastern export shutdowns due to Strait of Hormuz turmoil.

Arjun ·0:48

How to verify: Check ICE Brent Crack Spread data or Platts crack spreads for 2026-2026 and cross-reference with news reports on Russian refinery damage and Middle East shipping disruptions. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

New refinery additions in the developing world (Middle East, India, China) are significantly lower than historical norms, with China's transportation fuel demand on track to plateau.

Arjun ·1:52

How to verify: Review IEA World Energy Outlook or OPEC Annual Statistical Bulletin for projected refining capacity additions by region and China's domestic fuel consumption forecasts. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

Major agencies and street analysts forecast oil demand to peak or plateau in the 2030s, with some scenarios extending to 2040.

Arjun ·3:07

How to verify: Search for 'peak oil demand' forecasts in IEA, OPEC, and EIA reports published between 2026 and 2026. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

US refining, particularly on the Gulf Coast and Midwest, has shifted from a weakness to a strength due to reduced global competition and strategic brownfield expansions.

Arjun ·3:35

How to verify: Analyze EIA refining capacity data and earnings reports from major US refiners (e.g., Marathon, Valero, Phillips 66) for capital expenditure trends in brownfield projects. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

US power demand will grow at a structural rate of 2-3% annually.

Arjun ·11:55

How to verify: Compare this claim against EIA Short-Term Energy Outlook (STEO) or AEO2026 forecasts for US electricity consumption growth rates. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

The US is number one in data centers in the world by a good margin.

Unidentified speaker (UNKNOWN) ·8:17

How to verify: Review data from Synergy Research Group or similar industry analysts on global data center capacity by country. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

Congressional scrutiny of tech executives did not alter the fate of the banking sector post-GFC, as credit flows shifted to private credit rather than stopping.

Unidentified speaker (UNKNOWN) ·10:49

How to verify: Examine Federal Reserve Financial Accounts of the United States (Z.1) for trends in bank lending vs. private credit growth from 2026-2026. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

Current energy management teams prioritize profitability, advantaged growth, and fortress balance sheets over volume growth ('love of the barrel').

Arjun ·14:48

How to verify: Analyze recent earnings call transcripts and 10-K filings of major integrated energy companies for management commentary on capital allocation priorities. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

Primary source

The speaker uses Claude for Excel and PowerPoint to create visualizations like waterfall graphs, but retains full control over investment opinions and argument structure.

Arjun ·16:34

How to verify: This is a personal claim about the speaker's workflow; verification requires direct confirmation from the speaker or review of internal process documentation if available. For recent events, verify 2026-08-14 through 2026-11-14. Use another period only when the speaker explicitly names it.

Do your own research

  • Verify current crack spreads and refinery utilization rates via EIA (Energy Information Administration) data.
  • Check global refinery construction pipelines and capex announcements from major refiners (e.g., Exxon, Shell, Reliance) in SEC filings or press releases.
  • Review IEA (International Energy Agency) and OPEC oil demand forecasts for the 2030s to assess the 'peak oil' consensus.
  • Analyze US power demand growth projections from EIA and utility company earnings calls, distinguishing between AI-driven and structural growth.
  • Assess India's energy policy and EV adoption rates versus refining capacity expansion plans via government publications or industry reports.
  • Evaluate US data center power consumption estimates from sources like the EIA or industry groups like the Data Center Energy Association.
  • Check interest rate sensitivity of regulated utility valuations using historical beta data against Treasury yields.
  • Review congressional hearing transcripts or regulatory filings regarding tech executive scrutiny to assess potential impact on data center growth.
Full transcriptexpand
0:07
Unidentified speaker (UNKNOWN): Hi,
0:08
Arjun: it's Arjun. We are going to continue with the theme from last week of electoral supervolatility and go through a series of macro and strategy questions that are an outgrowth of the geopolitical Political supervol Super Bowl world in which we live. We're going to focus this week on two of the topics that have come up in a
0:28
Arjun: lot of our recent meetings and events we've been part of in terms of macro perspectives that we have around refining and the power sector. Let's get into it. Then the first thing we want to address is how are we thinking about the refining macro? We clearly have crack spreads that are way above historic averages that
0:48
Arjun: everyone, I think, instinctively says these are. not normal spreads, clearly driven by ongoing uh refinery disruptions in Russia from the Ukrainian drone strikes, and of course, the uh shutoff of product exports from many of the Middle Eastern countries due to the Strait of Hormuz and the uh ongoing war
1:07
Arjun: turmoil with Iran. On That turmoil, even if you think it's likely to continue, whether it's a stalemate or what have you, that it can persist through 2026, 2027, maybe even longer than that, it's going to be very hard for anyone to say that somehow is quote normal, and that we should normalize these types of crack
1:27
Arjun: spreads and make investments accordingly. And neither we nor anyone are arguing for
1:32
Unidentified speaker (UNKNOWN): that. What
1:33
Arjun: I will observe though is even prior to this kind of more recent period of Strait of Hormuz destruction. disruptions, ongoing Ukrainian drone strikes in Russia, even prior to that, the quantity of new refineries, especially in the developing world, historically, it's been the Middle East, India, and China that have seen the overwhelming majority of new
1:52
Arjun: refinery additions. That quantity of expected new meal was nowhere near where it's been throughout much of my career. There's the obviously the massive plant in Nigeria, the Dungate refinery. There are some expansions in India, not so much in China anymore, perhaps because uh especially their transportation fuels are on track to plateau um in the
2:12
Arjun: near term. But the the the the the again, the volume of new refineries just not what it once was, even if we strip out uh any uncertainty on And how long the geopolitical conflict portion of high crack spreads uh would would rise. It does seem
2:26
Unidentified speaker (UNKNOWN): to be a good idea.
2:27
Arjun: that refining is one of those sectors where that net zero mindset, the idea that oil demand could uh peak and plateau in coming years or into decades type timeframe, that seems to be playing a huge role, uh, we suspect, in especially the developing world and not having the degree
2:47
Arjun: of new refinery additions that we've had historically. Even for the discrediting of those sharply declining net zero scenarios, almost everybody, certainly all the large companies that forecast these things publicly, all the major agencies, I think just about all the street analysts, all have some version of a peak or plateauing
3:07
Arjun: oil demand sometime in the 2030s, 30s in the worst case by 2040. And with that type of backdrop, one of the outgrowths seems to
3:15
Unidentified speaker (UNKNOWN): be
3:16
Arjun: a lack of interest, excitement, desire, whatever the right phrase is, for new refinery editions, especially in the developing world. And that is our focus. I think this bodes very well for the US players that are in the US Gulf Coast and the Midwest. This has gone to an area of
3:35
Arjun: weakness for the US energy industry 25, 30 years. ago to a real area of strength. It's the strength for the sector and is the strength, frankly, for the United States. One of the real kind of areas where we have manufacturing dominance is in US refining, especially in the Gulf Coast and the US-Midwest. Huge competitive advantage
3:55
Arjun: to not have anywhere near the degree of competition, either domestically and especially globally, that again has existed for most of. my 25 or 30 year career. I think there are some interesting moves made during earnings season to bring back a previously shut in refinery that's going to take uh, I think about a year to restart. There's been some other moves to
4:15
Arjun: trim some um,
4:16
Unidentified speaker (UNKNOWN): you know,
4:17
Arjun: I guess non core refineries are on the edges and reinvest in existing plants. plans, So kind of brownfield expansions from many and I think essentially all of our largest U.S. refiners. Very interesting to see those strategic moves. And I think it bodes well for the outlook and the strength of, again, especially US Gulf Coast and Midwest refining outlook over
4:37
Arjun: the very long term. I think if you're a developing country, and I participated in a discussion earlier this week about India, I think one of the debates would be when you're a crude short country, as say India is, how do you think about investing? in new electric vehicle plants, L and G trucks, versus, say, oil refining? And I think you all might
4:57
Arjun: by this time know our view on this. We're not picking between these alternatives in a country like India. Our opinion would be they do have to think about continuing to build on the strength of their refining sector while also So meeting the challenge of significantly expanding and growing their electric vehicle capacity, such. that they don't become dependent on China for the
5:17
Arjun: EVs and that they have alternatives to having to otherwise import increasingly substantial quantities of oil. You're trying to do both.
5:25
Unidentified speaker (UNKNOWN): And
5:25
Arjun: I do think it is that still the legacy of that net zero mindset and this notion that everyone seems to be so confident It in, despite five plus billion people using a fraction of the energy the rest of us. uh get to take for granted, that somehow oil demand is going to just magically plateau in the 2030s, and with it is coming a
5:45
Arjun: dearth of new refining capacity. So make no mistake, we have covered refining for way too long to not realize we are going to go through some 12-month period. It could be next year, it could be the year after that. Refining is going to feel terrible in the short term. It is the nature of it, it is deeply cyclical, boom. bust on shorter term
6:04
Arjun: timeframes. It was as recently as I think the second half of 2024, first half of 2020
6:10
Arjun: coming off that COVID recovery and initial start to Russia, Ukraine, where margins were good and then they were weak for a kind of a full. We're going to have those kind of conditions again. Our focus right now is on that longer term outlook. Are we going to need new refineries in the developing world? Absolutely. we are. Are we getting them in the quantity we once did? No,
6:30
Arjun: we're not. Are people too pessimistic on a plateauing of oil demand in the 2030s? I think people are too confident in that view. It is so ubiquitous. And again, I don't see anyone solving for how is everyone on Earth some decade, some year, someday, becoming energy rich? And how is that not going to happen with still growing quantities?
6:51
Arjun: of oil? I've not seen anyone actually do that math. And this bodes well. Our point today is on especially US Gulf Coast and Midrash refining. It is a really interesting, I think, strategically important area that I think as Americans, we are blessed to have this type of industry in our country.
7:11
Arjun: Second key topic area I want to talk about is power inducing. and data centers, of course, post the election news and the blue tsunami that we had. And what are our views on power demand growth? That is coming up in a lot of meetings. And so first, make no mistake, power demand is going to grow in the
7:31
Arjun: United States, including the great state of Texas. taxes, And we are absolutely confident it will. The growth drivers are multifaceted. While AI and tech and data centers get a lot of attention, it is just one of a number of demand drivers. We do have aging grids, especially in the developed world. That's true in the United States, to
7:51
Arjun: varying degrees true in uh parts of Europe. We've had a big surge in various intermittent resources. How do you firm that up? How do you deal with the intermittency? How do you deal with deck duck curves? Uh, all those kinds of things.
8:03
Arjun: Um, we have
8:05
Unidentified speaker (UNKNOWN): uh
8:06
Arjun: electrification, we have hopefully have industrial reshoring, we have many different drivers of power demand. This is not just about AI and data centers, though make no mistake.
8:17
Unidentified speaker (UNKNOWN): The
8:17
Arjun: US is number one in data centers in the world by a good margin, and we are absolutely going to and should remain number one. If this is an area where we somehow see cede our advantage to China, it would be an extremely disappointing kind of thing. There is no way as a country we should let China or anyone else catch up
8:37
Arjun: on data centers. So we are in a moment where uh American voters feel a lot of angst. They've taken it out on tech and AI data centers. We can say that's justified, we can say it's not justified, it is, but are we not gonna have going to these things grow? They are absolutely going to grow, uh, even if we're going through a little bit of sort of backlash as
8:57
Arjun: people feel right now. A lot of attention gets oriented towards kind of consumer AI, Chat GPT, uh, the equivalent from some of the other the other companies. This is really, at the end of the day, going to be about business productivity and what we might collectively call the physical AI space. Drones, autonomous driving. robotics space, the
9:17
Arjun: list is long. This is our application of new tech. This is going to be the driver of power demand within this component of all the factors we cited for power demand growth. And as those technologies get developed, with it, we're going to need data centers and we're going to need power to power those data centers. Uh And it
9:37
Arjun: we have an optimistic view. on the progress being made in a lot of those those new technologies. It's never a straight line up, there'll be setbacks. But when from what we see firsthand in the various businesses we're involved with and
9:54
Unidentified speaker (UNKNOWN): um
9:54
Arjun: have as clients and so forth, I think that business productivity, physical AI space is going to be the driver here. And I think that's not the something we feel very, very good about. Will congressional scrutiny impact data centers and tech?
10:11
Arjun: We shall see. If you're in the oil industry and you've been hauled before Congress, it is not a pleasant thing, I am sure.
10:18
Unidentified speaker (UNKNOWN): But
10:19
Arjun: did it actually change the fate of anything Uh related to the energy sector? Not that we can see. After the great financial crisis or during it, I guess, leading bank CEOs, investment bank CEOs as well were hauled in front of Congress. Did that help anything? Not that we could see. Who are those? Who are those four? What American watches
10:38
Unidentified speaker (UNKNOWN): any
10:39
Arjun: of these executives being hauled in front of Congress? They, oh yeah, wow, I really feel better about my congressperson as a result of that. Anyway, it might happen here to the tech executives. I mean, go back to the banking crisis.
10:49
Unidentified speaker (UNKNOWN): Um,
10:49
Arjun: it is true that. a lot of regulation came into place, that those I can't remember what the acronym was, globally systemic systemic and uh important banks, whatever the phrase was, they face a lot of regulation. And now we've got a massive private credit industry. So the flow of credit to markets didn't change at all. In fact, it's been as healthy and as
11:09
Arjun: high as ever. Just who provided it did change. So we're not tech analysts. We can't predict how the structure of the technology industry may or may not change based on what Congress does. But the actual technologies, people are going to want, they're going to want it, they're going to use them. And with that, and the point of this segment is we will see power
11:29
Arjun: growth as a result of it.
11:31
Unidentified speaker (UNKNOWN): We
11:31
Arjun: don't have, I think, at Veriton officially published forecasts, but in a similar manner that we have continued to express confidence. of around 1 million barrels a day per year of structural oil demand growth, our equivalent power number for the United States would be on the order of 2 to 3%. So there were some of
11:52
Arjun: annual
11:55
Arjun: growth in power demand.
11:58
Arjun: There are forecasts that were much higher than that, that you can see from street analysts. and consultants and others out there. We never really believed in that. Maybe I haven't publicly talked about that enough. I'm going to say the equivalent of sort of around a million barrels a day of structural oil demand growth, the equivalent on the power side is about a 2%, 2-3%
12:18
Arjun: USK ground power power demand. And that is something we we feel uh very, very good about.
12:23
Unidentified speaker (UNKNOWN): If
12:23
Arjun: we're in a moment in time
12:25
Unidentified speaker (UNKNOWN): where
12:25
Arjun: some of the most aggressive species spending by the technology companies is going to slow down for one reason or another. And if there were some concerns on the credit of some of the, it's become a very levered type of uh bet that some of the technology sector companies have made. Um,
12:41
Unidentified speaker (UNKNOWN): to
12:41
Arjun: the extent that growth in debt was starting to compete with U.S. treasuries, to the point that US Treasury rates were rising. Well, power is one of those highly interest rate interest rate sensors. sensitive sectors.
12:52
Unidentified speaker (UNKNOWN): And
12:52
Arjun: if some of the significant upward pressure we saw on rates on US Treasury yields, if that eases a bit. And again, we're not treasury market analysts either, per se. But if that did ease from any easing in tech spending, might actually give a little bit of a life to especially some of the regulated utilities whose valuations are very sensitive
13:12
Arjun: to interest rates and who have uh the valuations of have suffered as the the treasury yields have gone. So we don't uh that is never, as you know, um an investment call. Just observing power is a very rates sex to rate-sensitive kind of sector. Uh perhaps some of the cooling of the most extreme optimism around tech and tech spending might help that point a little bit. So, our
13:32
Arjun: bottom line, we continue to expect somewhere around a positive two 2% to three 3% percent Kager in US power demand. Um We long ago left the hopes and dreams phase of massive excitement about how fast power could grow. And we transitioned to the execution phase of the cycle, which was always going to be more challenging. Maybe
13:52
Arjun: tech's taking some of the steam off the most enthusiastic tech kind of forecasts out there They're in spending might not be the worst thing in the world for the execution portion of the sector. There are absolutely going to be. winners and losers. And, you know, that is something everyone involved should be very
14:08
Arjun: cognizant of. Just a final point we want to address this week is just kind of an observation coming through the bulk of the most recent earnings season and to really give some favorable marks to a wide swath of energy companies that is both in the traditional oil and gas. space, it's across the power
14:28
Arjun: value chain, uh, and sort of everything in between of how they are navigating what is really a very challenging geopolitical and general macro environment. The quality and the nature of the current management teams, and again, I'm gonna speak broadly and generally here, Year, uh, it's as good as I can think of in the
14:48
Arjun: context of my 30-year career. Um, there's a focus on profitability, there's a focus on um advantaged growth, and there's sort of um just a running a real business to these things. The the line used to be, and this is more for the oil and gas names, is well, let's just let's not just focus on barrels for the sake of barrels. To
15:08
Arjun: some degree, this is what's going on in the tech sector, tech for the sake of take tech. Um
15:15
Unidentified speaker (UNKNOWN): I
15:15
Arjun: don't find the management teams as in love of the barrel or the MCF or the electron uh than that they were in decades past. It's moved from that type of mindset, the love of the barrel, the love of the MCF, the love of the electron, to a love of profitability, advanced growth, fortress balance
15:35
Arjun: sheet, all of these kind of terms. I do wonder when. tech's going to go through the equivalent
15:43
Arjun: reckoning. Maybe that started now. I have no idea. I'm not a tech analyst. The point of this session was to kind of give kudos to the sector, generally speaking. There's always going to be exceptions of just being better managed and better run. And it does make us generally feel um good about how the sector is going to navigate all the significant turmoil, the volatility, the election
16:03
Arjun: results that have come our way.
16:06
Unidentified speaker (UNKNOWN): So
16:06
Arjun: I'll end this video on a personal note. And I do want to just spend a moment, I don't think I've done this yet, of how do we use AI in creating Super Spiked?
16:14
Arjun: I'm going to say to me, this is a tool just as Excel or Word or port PowerPoint are is a tool. I'm speaking to as an actual person. This is not some uh virtual image of myself. This is my actual voice. It's not even pre-recorded for better or for worse. You know, uh, I love um Claude for Excel and Claude for
16:34
Arjun: PowerPoint. What I primarily use them for is to make the graphs seem prettier. I can never remember how to do waterfall graphs um or you know, organically. Uh so I actually do add Claude Excel. Hey, like a waterfall graph for this. Or here's some data that I'm showing. What are different options that I and how to visualize it in terms of the actual creation of the
16:54
Arjun: content itself, whether it's the specific opinions expressed, um, how the arguments are structured. Again, for better or for worse, I'm doing those on my own and myself. And I don't quite know how or why one would outsource that. That to me is how you're going to differentiate kind of the median view, the consensus view from folks that hopefully can have
17:14
Arjun: different. differentiated perspectives, which is at least what we strive to have. And so our perspective is this is a great tool. I don't view it as a replacement for anyone or anything we do. You know, I have three kids who are just out of college, one who's now a junior, and one of the debates is well, won't junior people now not be needed at uh at at all these firms? And
17:34
Arjun: I'll I'll speak for the financial services sector. You know, Veridin, Verdin, we
17:38
Unidentified speaker (UNKNOWN): have
17:38
Arjun: I think two people who have that analyst, which is generally that sort of introductory type role at a financial services firm. And their ability to use AI is actually very impressive. And what I appreciate about how they use AI is they are inherently skeptical of it, but embrace the technology at the same time. And it's it's it's
17:58
Arjun: to me an awesome combination. By skeptical I mean it's not just about trusting whatever output comes. It's about understanding what you're feeding into the system, whether what you fed in makes sense, how does the output compare to that? They apply their own judgment and they ask myself, Argent, does this make sense? They'll ask other senior people at the firm,
18:18
Arjun: does this make sense? And
18:19
Unidentified speaker (UNKNOWN): um
18:20
Arjun: why would we not have junior people? It doesn't actually make any sense to me. It is great that they're able to use these tools and do a lot of work that uh, you know, 10 years ago, 20 years ago, one year ago would have taken many, many months or certainly many, many weeks of effort that they're now able to do much faster as a result of AI.
18:41
Arjun: Point being, it is a great and awesome tool. Uh and I uh but I think it is just that. So uh just some perspectives on how we're using AI. Hopefully that makes sense. And uh we'll see you next week. Take
18:53
Unidentified speaker (UNKNOWN): care.
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.
All insights