GUY: Morning Ava, it's May 3rd and... man, what a week to be doing this job, right?

AVA: Tell me about it, Guy. I feel like we're watching multiple tectonic shifts happening simultaneously. The energy markets, AI infrastructure, geopolitics... it's all interconnected in ways that are honestly a little scary.

GUY: Scary is the right word. And let's just dive right into the biggest story because it's not just about oil prices anymore. So I was listening to Macro Voices this week, and Daniel Lacalle made a point that really stuck with me. He's saying we're witnessing the actual collapse of the OPEC system that's managed global oil markets since the 1970s.

AVA: Right, and it's not just supply disruption. The UAE exiting OPEC is... that's structural.

GUY: Exactly. Look at the numbers here - WTI crude has spiked from 79 dollars to over 110 in just two weeks. That's a 40% gain. But here's what's really wild - over on Macro Voices, they're tracking vessel transits through the Strait of Hormuz, and we've gone from 130 daily transits to fewer than 500 total over 40-plus days of this naval blockade.

AVA: Wait, so you're saying fewer than 500 total, not per day?

GUY: Total. Over 40 days. The shipping lanes are essentially shut down. But here's where Lacalle's analysis gets really interesting - the UAE's exit from OPEC isn't just about this crisis. They're signaling they want to produce at maximum capacity rather than honor quotas ever again.

AVA: And that eliminates what he calls "spare capacity" as a market management tool.

GUY: Exactly. The forward curves are now permanently discounting oil 15 dollars plus above January levels through December 2027. That's not crisis pricing anymore - that's a fundamental recalibration of global inflation expectations.

AVA: Which is forcing central banks into hawkish postures regardless of domestic economic conditions. Guy, over on Forward Guidance this week, they were talking about how the Fed just delivered four dissents at their latest meeting. Four dissents! That's the most since 1992.

GUY: That's wild. What were they dissenting on?

AVA: Three members opposed the "easing bias" language while supporting current rates. But here's the kicker - markets have now completely priced out rate cuts by December 2026 and are expecting hikes by March 2027. That's a dramatic shift from where we were just a few months ago.

GUY: And it makes sense when you think about it. If oil is embedding a permanent 15-dollar risk premium, the Fed can't cut rates without adding fuel to the inflation fire. But Ava, there's another layer here that I caught on RenMac. Iran is facing critical storage capacity constraints within two weeks.

AVA: Hold on, within two weeks? What happens then?

GUY: They potentially have to shut down production on wells that date back to the 1950s and 60s. And here's the thing - these old wells, once you shut them down, they may never restart. We're talking about an existential threat to Iran's entire oil industry infrastructure.

AVA: So this isn't just about current supply disruption. This could permanently reduce global oil production capacity.

GUY: Exactly. And then you layer in the fact that US crude storage is projected to hit record lows because we've been supplying global markets during this crisis, while rig counts aren't increasing meaningfully despite higher prices. We're looking at supply constraints that could persist for years.

AVA: OK but Guy, here's where this gets really interesting from a markets perspective. Over on Forward Guidance, they're talking about how the dollar is evolving into what they call a "petrocurrency." Because America went from being the largest oil importer to the largest producer, the dollar now rises with oil prices instead of falling.

GUY: Right, that breaks all the traditional crisis correlations.

AVA: Completely. Gold actually declined while oil surged, which forced central banks that had accumulated gold from 2023 to 2025 to sell positions to offset currency depreciation. The whole playbook is different now.

GUY: And speaking of different playbooks, we need to talk about what Forward Guidance calls the "yield smile" replacing the traditional dollar smile. In this fiscally-dominant world, yields rise both when the economy runs too hot and when equity returns turn negative. It's creating structural headwinds for fixed income regardless of conditions.

AVA: The fiscal dominance angle is crucial here. Over on Invest Like the Best, Paul Tudor Jones was pointing out that US stock market capitalization is now 252% of GDP versus 65% in 1929 and 170% in 2000. And here's the scary part - 10% of federal tax revenues now come from capital gains taxes.

GUY: So the government literally can't afford to let markets correct significantly.

AVA: Right. A 30 to 35% correction would eliminate 80 to 90% of GDP in reverse wealth effects, while those capital gains tax revenues would vanish. It's a fiscal trap.

GUY: And meanwhile, Paul Tudor Jones noted the fastest money supply growth since 2021 globally, led by China but including substantial US expansion. That explains why asset prices have been resilient despite these genuine economic threats.

AVA: But Guy, there's another critical pressure point building, and this is where the tech story intersects with energy. We need to talk about AI infrastructure demands because they're creating what I'm calling a triple constraint.

GUY: Walk me through that.

AVA: So over on Cheeky Pint, they did a deep dive on Siemens Energy, and here's the issue - electricity demand growth is accelerating from 0% annually over the past 20 years to an expected 4% over the next 20 years, driven entirely by AI data centers. But turbine manufacturers like Siemens are remaining conservative about capacity expansion while demand is exploding.

GUY: So we have an energy supply crunch at the exact moment AI is demanding unprecedented power.

AVA: Exactly. And Forward Guidance mentioned that states are starting to impose grid usage restrictions, which is forcing data center operators toward independent natural gas turbine solutions. They're calling it "21st century gas stations for data centers."

GUY: That's actually a brilliant analogy. But let's dig into the AI economics here because I was listening to No Priors, and Base10 is reporting some pretty remarkable numbers.

AVA: Oh yeah, they've grown 30x to over a billion dollar revenue run rate, with 95% of tokens being custom model inference rather than vanilla open source.

GUY: Right, and here's what's interesting about the supply chain - the GPU supply crunch requires 3 to 5 year contracts with 20 to 30% of total contract value prepaid. That creates capital allocation advantages for public companies over private ones.

AVA: But there's a really interesting competitive dynamic emerging. Chinese models are offering comparable performance at about 20% the cost of Anthropic models, and Base10 says they're seeing no security concerns. It's essentially the Chinese government subsidizing US enterprise AI adoption through their model development investments.

GUY: That's wild. So we have economic warfare over energy flows at the same time China is subsidizing our AI adoption?

AVA: Exactly. And over on RenMac, they're tracking how Treasury sanctions are intensifying on Chinese "teapot refineries" that had become Iran's primary export destination. The conflict has devolved into a proxy economic war between US and Chinese staying power.

GUY: But let's talk about the reality of enterprise AI adoption because I think the hype is getting ahead of the infrastructure. Over on a16z, they identified what they call the "integration wall."

AVA: Right, any enterprise with 1000 plus people or 10 plus years old represents this "mass of stuff waiting to be integrated" that AI doesn't actually solve. The integration challenges remain.

GUY: And they're seeing board mandates fail consistently. The pattern is CEO gets consultant, centralized project that nobody understands, and it fails. Meanwhile, they're observing actual code quality degradation when people code with AI.

AVA: Wait, really? I thought AI was supposed to make coding better.

GUY: That's what you'd think, but they're saying when coding with AI, your code gets worse over time, pretty materially. You might get 2 to 3x productivity gains, but they're constrained by security review, code review, and deployment pipeline bottlenecks.

AVA: So the human processes become the constraint, not the AI capability.

GUY: Exactly. And this ties back to the infrastructure concentration risks we're building. Over on Business Breakdowns, they did a deep dive on Cloudflare, which now controls over 20% of global web traffic and processes 2.5 million cyber attacks per second.

AVA: And they had that global outage recently that demonstrated both their essential role and the concentration risk.

GUY: Right. As AI increases digital infrastructure dependence, companies like Cloudflare become more valuable but also create more systemic risk. It's a classic case of concentration creating both opportunity and vulnerability.

AVA: Guy, let's shift to the geopolitical dimensions because I think this Iran crisis is revealing some fundamental shifts in global power dynamics. Over on Macro Voices, they're describing how this conflict has evolved from kinetic to economic warfare.

GUY: Right, with Treasury sanctions intensifying on those Chinese teapot refineries. The strategy is economic pressure rather than military escalation, though Trump apparently considers "both potentially."

AVA: But here's what's interesting - Iran has begun rationing fuel domestically, signaling resource constraints, but JP Morgan estimates several more weeks before forced production shut-ins. That contradicts the Trump administration's claims of imminent capitulation.

GUY: And most vessel transits are happening through diplomatic agreements rather than direct IRGC payments, which suggests the blockade isn't as airtight as the administration claims.

AVA: Exactly. But the European vulnerability is really exposed here. Consumer sentiment is at the lowest levels since the pandemic, the EU lacks strategic petroleum reserves, and they're facing what Macro Voices described as "several weeks of jet fuel remaining" with shortages only resolved by "paying five times normal prices."

GUY: Meanwhile, China benefits from their strategic partnership with Russia - 10 million barrel per day production, 4.5 million exports - plus they have the largest commodity stockpiles globally. It's really become a proxy economic war between US and Chinese staying power, with Europe suffering disproportionately.

AVA: And that's creating some interesting policy responses. European governments prefer windfall profit taxes over price controls, though the populist left is advocating expropriations and direct price intervention.

GUY: At least most policymakers understand that price controls would worsen the supply disruptions.

AVA: For now. But Guy, there's a technology dimension to China's position that I find fascinating. Over on Cheeky Pint, they noted that arbitrary policy decisions create persistent uncertainty that maintains valuation discounts across Chinese equities, despite strong businesses and talent.

GUY: And technology entrepreneurs are increasingly relocating to Singapore for more predictable operating environments.

AVA: Right. So you have this weird dynamic where China is leading in AI model development and subsidizing US adoption, but their own tech ecosystem is constrained by policy uncertainty. It's almost like they're exporting their AI advantage while hampering domestic implementation.

GUY: That seems unsustainable long-term. But let's talk about the cross-currents here because I think we're seeing a fundamental breakdown in traditional market relationships that extends far beyond oil prices into the architecture of global finance and technology deployment.

AVA: Absolutely. The transmission mechanism operates through electricity demand growth, but it intersects with AI infrastructure in ways that amplify both the energy shortage and technology deployment challenges.

GUY: Right, and over on Dwarkesh, they did a fascinating technical deep dive into memory bandwidth bottlenecks. H100 racks can evacuate and replace HBM in about 15 milliseconds, with optimal batch size being 300 times the sparsity ratio.

AVA: OK, break that down for me.

GUY: So for DeepSeek V3, 32 out of 256 experts are active, which gives you an 8 sparsity ratio. So optimal batch is around 2,400 tokens, translating to about 128,000 tokens per second throughput. The point is, there are fundamental hardware constraints that create natural batching cycles.

AVA: So even if we solve the power problem, there are still computational bottlenecks.

GUY: Exactly. And this creates what I think is a critical paradox - the AI sector requires massive, stable power supplies precisely as global energy markets face their most significant disruption since the 1970s oil shocks.

AVA: And Base10's billion-plus revenue run rate requires those 3 to 5 year GPU contracts with 20 to 30% prepaid, but power infrastructure bottlenecks may constrain deployment regardless of chip availability.

GUY: Which means infrastructure companies like Bloom Energy and Siemens Energy potentially become more valuable than AI software companies if power constraints bind technological progress.

AVA: That's a really important insight. The bottleneck shifts from software to infrastructure.

GUY: Right. And then you layer in the currency market dynamics. Over on Forward Guidance, they're watching USD/JPY approaching the critical 160 level with potentially explosive implications.

AVA: Because Japan imports 60% of their food and 90% of oil and gas with a 230% debt-to-GDP ratio.

GUY: It's a lose-lose scenario. If 160 goes, Japan yields rise, US yields rise. If they defend it, they weaken the dollar, NASDAQ sells off, yields rise anyway. Any significant move above 160 would likely trigger carry trade unwinding, causing NASDAQ selloffs precisely when AI companies face power infrastructure constraints.

AVA: So we have energy supply restrictions, currency market instability, and technology deployment bottlenecks all happening simultaneously. That's your triple threat.

GUY: Exactly. And the Chinese factor adds another layer of complexity. Chinese labs like DeepSeek and Moonshot are leading open-source model releases while Chinese teapot refineries face intensifying Treasury sanctions. It's economic warfare and technological collaboration happening at the same time.

AVA: But Base10's data point about 95% of their tokens coming from custom models rather than vanilla open source suggests enterprise differentiation is happening through specialized rather than general AI capabilities.

GUY: Which makes sense given the integration challenges we talked about. Companies aren't just plugging in GPT-4 and calling it done.

AVA: Right. And this is where market concentration amplifies these cross-domain risks. That 252% of GDP stock market capitalization creates unprecedented wealth effects, with 10% of federal tax revenues coming from capital gains that would vanish in significant corrections.

GUY: Plus private equity allocation rising from 7% to 16% of institutional portfolios reduces liquidity precisely when energy and technology shocks require portfolio rebalancing.

AVA: And Paul Tudor Jones's identification of the fastest money supply growth since 2021 globally explains asset price resilience despite genuine threats, but creates additional inflationary pressure as energy costs embed permanently higher levels.

GUY: The enterprise AI adoption pattern also reveals why infrastructure providers may capture disproportionate value. That integration wall means successful deployments require workflow integration that favors established platforms rather than pure AI capabilities.

AVA: Which explains why companies like Cloudflare become more valuable as AI increases digital infrastructure dependence, despite the concentration risks.

GUY: OK, so what are the investment implications across multiple asset classes and time horizons? Erik Townsend's tactical positioning - doubling S&P hedges while maintaining crude oil exposure - reflects short-term conflict hedging.

AVA: But over on The Compound and Friends, Matt Ankram's research showing only 4% of companies from 1926 to 2018 created net wealth above T-bills reinforces the importance of long-term quality focus.

GUY: The "hundred bagger" framework identifying companies with sustainable 15% plus returns on tangible assets becomes more relevant as market concentration creates winner-take-all dynamics across energy, technology, and infrastructure sectors.

AVA: But Ben Felix over on The Diary of a CEO provides a crucial counterpoint to these tactical considerations. His million-simulation bootstrap analysis across 39 countries from 1890 onwards suggests 100% equity allocation with one-third domestic, two-thirds international stocks remains optimal for long-term wealth building.

GUY: Even despite current volatility?

AVA: That's his argument. The 7% long-term stock return assumption faces testing from permanent energy price elevation and AI productivity claims, but Felix emphasizes behavioral factors. He says "the more people look at their investments, the less risk they take, and the lower returns they earn."

GUY: That behavioral insight might be more valuable than tactical positioning during regime transitions.

AVA: Exactly. But Guy, let's talk about what we're watching because there are some really specific catalysts and dates coming up that could determine how all this plays out.

GUY: Yeah, March 2027 is now when markets expect Federal Reserve rate hikes after completely pricing out cuts by December 2026. Energy price elevation is forcing monetary policy recalibration regardless of domestic conditions.

AVA: But much more immediately, we're watching those next two weeks for Iran storage capacity constraints potentially forcing production shutdowns of those 1950s and 60s era wells that may never restart.

GUY: That's an existential threat to their oil industry infrastructure. And then we have the USD/JPY 160 level as a critical technical and policy threshold.

AVA: Where Japan faces that lose-lose scenario we discussed - defend and weaken the dollar causing NASDAQ selloffs, or allow the break triggering global yield increases.

GUY: For Q2 2026, we're watching GPU supply crunch resolution based on those 3 to 5 year contracts with 20 to 30% prepayment requirements. That determines whether AI infrastructure deployment can meet demand projections.

AVA: And December 2027 is when we'll know if that forward curve pricing for permanent 15-dollar plus oil risk premium above January 2026 levels becomes a permanent regime change rather than just crisis pricing.

GUY: The 2026 midterm elections are interesting too. Congressional redistricting impact appears limited by timing, but population migration from Democrat to Republican-led states affects 2030 redistricting with long-term political implications.

AVA: Paul Tudor Jones's warning about lack of democratic input on AI development pace becomes critical as tail risks potentially affect hundreds of millions without public oversight mechanisms. That's an AI safety regulatory framework question that's getting more urgent.

GUY: And then Siemens Energy's capacity expansion decision represents German industrial conservatism versus Silicon Valley AI demand growth creating that fundamental supply-demand imbalance in turbine manufacturing with 20-year infrastructure implications.

AVA: So we're really watching multiple regime changes happening simultaneously - energy market architecture, monetary policy, AI infrastructure deployment, geopolitical alignment. Any one of these would be significant. All of them together is...

GUY: Historic. And potentially very dangerous if they interact in ways we're not anticipating.

AVA: The interconnections are what worry me most. Like that example you gave earlier - energy constraints limiting AI deployment right when AI is supposed to drive productivity growth that justifies current market valuations.

GUY: Right. If AI productivity gains don't materialize because of power constraints, then we have this massive capex spending on infrastructure that doesn't generate expected returns, while oil prices stay permanently elevated, forcing central banks to stay hawkish.

AVA: Which means yields stay high, equity valuations compress, those capital gains tax revenues disappear, fiscal positions deteriorate, and we get a feedback loop.

GUY: Exactly. And all of this is happening while China benefits from energy partnerships with Russia and subsidizes US AI adoption, but faces domestic policy uncertainty that constrains their own tech ecosystem.

AVA: It's almost like we're in a period where traditional competitive advantages and alliance structures are all being reshuffled simultaneously.

GUY: And the speed of change is accelerating. Ben Felix's point about behavioral factors becomes really important here. The temptation is to try to tactically trade all these regime changes, but the complexity might make long-term quality focus more valuable than ever.

AVA: Though Paul Tudor Jones would probably argue that when you see tail risks of this magnitude, you can't just ignore them and hope your long-term allocation works out.

GUY: Fair point. I think the key is distinguishing between what you can predict and position for versus what you just have to accept as fundamental uncertainty.

AVA: Energy price elevation forcing monetary tightening seems pretty predictable. AI infrastructure power constraints seem predictable. USD/JPY breaking 160 and triggering carry trade unwinding...

GUY: That one's more binary. Either it happens or it doesn't, but if it does, the consequences are pretty severe.

AVA: And Iran's storage capacity constraints forcing well shutdowns - that seems like a near-term catalyst with long-term consequences.

GUY: Exactly. So we can position for some of these while accepting that others are just going to create volatility we have to navigate.

AVA: The other thing that strikes me is how much of this comes down to infrastructure and physical constraints rather than just financial or policy decisions. You can't just print more electricity or conjure up oil production capacity.

GUY: Right, and you can't quickly build turbine manufacturing capacity or retrofit data centers for independent power generation. These are multi-year, capital-intensive processes happening while demand is accelerating.

AVA: Which brings us back to why companies like Siemens Energy, Bloom Energy, even Cloudflare might be more important to this cycle than pure AI plays.

GUY: The infrastructure providers, the picks and shovels of the AI revolution, but also the energy transition. Though I'd argue Cloudflare is more like the railroad of the digital economy.

AVA: Good analogy. And just like railroads in the 19th century, essential infrastructure that creates both enormous value and concentration risk.

GUY: Alright Ava, I think we've covered the major themes and catalysts. This feels like one of those moments where we're going to look back and say this was when everything changed.

AVA: Whether for better or worse remains to be seen. But definitely changing. Thanks for diving deep with me today, Guy.

GUY: Always a pleasure. We'll be watching all those dates and levels we discussed. Talk to everyone Monday.

AVA: See you then.