GUY: Good morning... it is Friday, May 15th, 2026, and this is Morning Signal. I’m Guy.

AVA: And I’m Ava... good morning. We have one of those classic market mornings where the headlines look terrifying, the bond market looks annoyed, and stocks are basically saying, “Yeah, yeah, but is the AI capex train still leaving the station?”

GUY: Exactly. The cleanest way to frame today is this... markets are still acting like they can look through geopolitical shocks as long as three things don’t break at the same time: liquidity, energy flows, and the AI narrative. If those stay mostly intact, traders are buying dips, not hiding under the desk.

AVA: And that is either a sign of real resilience... or classic late-cycle complacency. I think that’s the core argument running through almost every podcast in the briefing today.

GUY: Totally. So let’s start with markets and macro, because that’s where the whole thing begins.

GUY: Over on Odd Lots, the Trump-Xi summit was treated by markets as an actual risk-on catalyst, not a ceremonial photo op. The numbers were pretty clear. A50 futures were up about 0.5%, Taiwan opened roughly 1.7% higher, the Nikkei gained about 0.7%, the Golden Dragon index jumped 4%, and KWEB was up 5%. Then you had Alibaba, Tencent, and Baidu all indicated sharply higher. That is not “nice diplomatic vibes.” That is the market pricing lower trade friction and maybe a softer stance on China’s AI supply chain.

AVA: Right... and over on Odd Lots, the presence of Jensen Huang around that summit orbit mattered a lot too. When the CEO of NVIDIA, ticker NVDA, shows up anywhere near U.S.-China diplomacy, the market immediately assumes the conversation is not just soybeans and tariffs... it’s semis, compute, export controls, and who gets to participate in the AI buildout.

GUY: Exactly. Markets basically said, “If this summit merely avoids a train wreck, we rally first and ask harder questions later.” And honestly, that’s pretty rational in the very short term. When positioning is defensive and you remove one layer of policy anxiety, the first move is relief.

AVA: Hold on though... relief from what exactly? Because over on Odd Lots, Mark Cudmore’s point was not that the underlying issues are solved. His point was that if the summit simply doesn’t end in a breakdown, that’s enough for a tactical rally. But unresolved issues can come back fast... especially if the Strait of Hormuz situation keeps pressure on oil and shipping.

GUY: That’s the key distinction. Markets are trading the first derivative, not the final answer. They’re saying, “Less bad than feared.” And in this tape, less bad is often enough.

AVA: Which is wild when you line it up against the bond backdrop. Over on Odd Lots, the panel tied that China optimism to hotter U.S. PPI and a 30-year Treasury yield back above 5%. So equities are shrugging and the long end is basically saying, “I’m not so sure.”

GUY: Yeah, and that split matters. Stocks can party on diplomacy headlines for a day or two. The 30-year doesn’t care about diplomatic body language. The 30-year cares whether inflation gets stickier and whether term premium needs to rise.

GUY: Over on The Compound and Friends, Barry Ritholtz gave the other side of the story... the resilience case. His line was, “It’s not the news, it’s the reaction to the news.” He pointed out that the market has repeatedly sold off on Iran escalation headlines, then reassessed, then recovered... sometimes even finishing green. He’s basically saying investors are getting better at separating scary headlines from economically durable damage.

AVA: And over on The Compound and Friends, Ritholtz backed that up with an actual macro reason, which I think is important. This wasn’t hand-wavy optimism. He said the U.S. is structurally less vulnerable to an oil shock than in the 1970s. We’re much closer to energy independence, we export oil and gas, and household energy spending has fallen from roughly 12% of budgets in the 1970s to around 6% to 7% now.

GUY: That is a huge change. If energy is half the budget share it used to be, the pass-through from oil spikes into consumer pain is just mechanically smaller. Back in the ‘70s, an oil shock was like somebody yanking the fuel line on the whole economy. Now it’s more like a tax... still bad, but not instantly recessionary.

AVA: I buy that... up to a point. But over on The Meb Faber Show, Tom Lee made the subtle version of the same argument. He said $100 oil by itself doesn’t kill the U.S. economy. So far, so good. But he also said crude is “an unknown” because if Strait of Hormuz disruptions persist, the real problem may show up in petroleum products and inventories, not just the front-month crude quote.

GUY: Right. That’s where people get lazy. They think, “Oil at $100 isn’t 1979, therefore who cares.” No. What matters is transmission. And over on Odd Lots, the panel laid that out well: hotter PPI, 30-year above 5%, shipping risk through Hormuz... that chain can reprice inflation expectations and the long end even if growth doesn’t fall off a cliff.

AVA: So the danger isn’t necessarily recession first. It’s duration first.

GUY: Exactly. That’s the whole ballgame. If oil goes up and people don’t stop spending immediately, you can still get nailed through discount rates. Long-duration assets can get hit by higher real yields or higher inflation risk premia before GDP looks terrible.

AVA: And over on The Compound and Friends, Ritholtz was blunt on the Fed piece... if oil spikes materially, near-term rate cuts are “for sure” off the table. That part seems almost obvious, but the market keeps wanting to have its cake and eat it too... risk-on equities and easier policy.

GUY: I don’t buy both at once. If energy inflation reignites, the Fed is not cutting just because the equity market wants a cuddle. And honestly, the bond market has been sending that message louder than stocks have.

GUY: Over on The Meb Faber Show, Tom Lee stayed constructive on equities anyway. He’s got 7,300 as this interim “magic number” on the S&P 500, 7,700 as his year-end 2026 target, and even 7,800 as a plausible overshoot before a major correction. That is still a very bullish roadmap.

AVA: But over on The Meb Faber Show, Lee also admitted stocks were a lot cheaper at the end of March and that now a lot of good news is priced in. So this is not blind cheerleading. It’s more like... medium-term bullish, tactically less excited after the rerating.

GUY: Exactly. He’s bullish, not oblivious. And I think that distinction matters. There’s a difference between saying “the path higher continues” and saying “this is a great entry point right now.” He’s not really saying the second thing.

AVA: And over on Excess Returns, Andy Constan gave almost the exact opposite framing. He said stop asking whether the AI story is real. His point is that a technology can be real and the market can still be in a bubble regime around it. He thinks we’re in the peaking phase right now.

GUY: Yeah... Constan’s framework was probably the sharpest warning in the whole briefing. He doesn’t define bubbles as one crazy valuation print. He defines them as multi-stage regimes. Root conditions, escalation events, then a peaking phase. In his telling, the root condition was AI becoming market-actionable around Microsoft’s January 10th, 2023 OpenAI investment. Then the 2023 regional-bank stress led to easier conditions, which helped escalate the trade. And now you have this peaking phase where investors are extrapolating miraculous growth too far into the future.

AVA: That’s the important nuance. Over on Excess Returns, Constan explicitly said semiconductor order books are “nuts.” He’s not denying the current revenue. He’s saying the market is not just valuing the next couple quarters... it’s capitalizing an almost absurdly long runway of hypergrowth.

GUY: And that’s always how the dangerous part of a bubble works. It’s not that the thing is fake. Fiber was real. The internet was real. Railroads were real. AI is real. The question is whether returns for investors get pulled too far forward. If you price ten years of miracles into today’s multiple, even good news can disappoint.

AVA: Over on RenMac Off-Script, Jeff deGraaf gave the simplest quant version of that warning. His rule is that when an index or sector doubles in two years or less, it enters “bubble zone.” He said the SOX had crossed that threshold about two weeks earlier, and the KOSPI had too.

GUY: That rule is elegant because it’s empirical, not ideological. He’s not saying “AI bad.” He’s saying rapid doubling changes the distribution of outcomes. Historically, realized volatility goes up, drawdown risk rises, and the next 6 to 12 months get trickier even if the melt-up continues for a while.

AVA: And over on RenMac Off-Script, the KOSPI point was sneaky important. Samsung and SK Hynix now make up 43% of the KOSPI, versus roughly 18% to 20% around 2000. So if people think they’re buying “Korea,” they’re often just buying a highly concentrated memory and AI hardware trade.

GUY: That concentration is a neon sign. It tells you leadership has gotten very narrow. Narrow leadership can persist... but it is fragile. It’s like a bridge held up by fewer and fewer cables.

AVA: Wait really... 43% in two names is insane.

GUY: It’s huge. And over on RenMac Off-Script, deGraaf’s operational takeaway was not “short it.” It was “size down.” That’s such an important distinction. People hear bubble and think they’re supposed to be heroic. No. Bubble regimes can keep going far longer than your short book can survive.

AVA: Which is where over on Odds on Open, George Livadas really reinforced the risk-management angle. He runs about 35% beta-adjusted net exposure with around 150% gross. He’s very deliberate about anti-crowding. And his short sizing is designed to survive 50% overnight squeezes... fraud and fad shorts are often only 50 to 70 basis points each.

GUY: That’s professional scar tissue talking. He’s saying, “I don’t need to be right tomorrow. I need to still be alive next month.” In late-cycle, narrative-driven markets, survivability beats swagger.

AVA: And over on Odds on Open, Livadas said when the Russell 2000 is up 3%, he’s often losing money, and when it’s down 3%, he’s often making money. That’s because he’s set up to monetize stress rather than chase momentum. There’s a philosophical point there... you don’t have to mirror the tape to make money.

GUY: Exactly. It reminds people that there’s more than one way to play a market regime. You can admit the bubble-ish stuff is hard to time and still build a book that benefits from the eventual sloppiness.

GUY: Over on The Meb Faber Show, Tom Lee made another point I liked... he said high-yield spreads are a better early warning signal than equities at turning points. One of his favorite bottoming tells is when stocks are still weak but high yield stops widening.

AVA: That makes sense. Credit markets are usually less theatrical than equities. If spreads stop widening, the system may be stabilizing even while the stock market is still panicking.

GUY: Right. Equities are emotional. Credit is plumbing. When plumbing starts to unclog, that matters. And over on The Meb Faber Show, Lee also pushed back on broad private-credit doom narratives. He expects losses, especially where software underwriting got sloppy, but he does not see a systemic event.

AVA: And over on The Meb Faber Show, the Oracle example was useful there. He said Oracle CDS got attention, but Oracle, ticker ORCL, ranked only about 246th in the S&P 500 by default risk. So headline spread widening by itself can be misleading. Relative rank matters more.

GUY: Exactly. People love turning every widening spread into “the next crisis.” Sometimes it’s just repricing. Not every cough is pneumonia.

AVA: Before we move on, I want to hit recession odds because over on The Compound and Friends, Ritholtz had a very nuanced take. He said no recession “today, tomorrow, next month”... but by Q4 of 2026, he’d put the odds around 50-50 if policy errors keep piling up.

GUY: That framing is dead-on. He’s not looking for one asteroid. He’s looking for cumulative damage. Tariffs, war risk, softening labor data, higher yields, policy whiplash... enough straws and the camel notices.

AVA: And over on The Compound and Friends, he suggested prediction markets may still be underpricing recession risk even after odds went from around 20% to 35%, roughly in the neighborhood of Ed Yardeni’s estimates. The key idea, borrowing from Lakshman Achuthan, is that strong economies survive shocks, weak economies tip from smaller incremental hits.

GUY: That’s macro in one sentence. The shock is only half the story. The underlying health matters more. A fit boxer can take a punch. A concussed boxer can’t. The U.S. economy right now looks more like the first than the second... but it’s not invincible.

AVA: So if I sum up the markets and macro section... over on Odd Lots, The Compound and Friends, The Meb Faber Show, Excess Returns, RenMac Off-Script, and Odds on Open... the consensus is weirdly narrow. No immediate collapse, probably more resilience than bears admit, but way more vulnerability to rates, concentration, and position sizing than the broad index level suggests.

GUY: That’s exactly it. The market is not saying “nothing matters.” It’s saying “show me actual damage.” But if that damage comes through the bond market or through overextended AI valuations, the turn can be sharper than people think.

AVA: All right... let’s shift into tech and AI, because this is where the story gets more physical than people realize.

GUY: Take it away.

AVA: Over on The a16z Podcast, in the episode with Ben Horowitz called “Your ONLY job is Right Product, Right Time,” Horowitz said the real AI bottlenecks now are not just model quality. They’re electricity, chips, cooling, tokens, and especially transformers. He said a16z actually backed a transformer company because grid hardware scarcity is directly constraining new power plants and AI capacity.

GUY: That’s one of the most important points in the whole briefing. Everybody talks about AI like it’s a cloud of software magic. It’s not. It’s a giant industrial project wearing a hoodie.

AVA: Exactly. And over on The a16z Podcast, Horowitz was basically saying the next constraint is in atoms, not just bits. If you can’t get the transformer, you can’t energize the substation. If you can’t energize the substation, you don’t get the data center. If you don’t get the data center, your AI roadmap is just a slide deck.

GUY: That shifts the investable universe. Maybe the next winners aren’t just NVIDIA, Microsoft, AMD, Broadcom... maybe it’s the boring industrial suppliers, thermal management firms, grid-enablement companies, and power equipment names that normal tech tourists ignore.

AVA: Right. And over on Excess Returns, Constan’s point fits right into that. He said he’s a long-run believer in AI as a 40-year technological arc. He is not anti-AI. His warning is that adoption timing and investor pricing are two different things. Society can benefit massively while capital markets still overpay in the near term.

GUY: That’s the Cisco lesson. Great company, bad stock for a long stretch because the starting price was too insane. When people say “but AI is real,” I always want to say, “Sure... so was the internet in 1999.”

AVA: Over on RenMac Off-Script, deGraaf reinforced that specifically for semiconductors. If the SOX is already in bubble zone, that doesn’t mean the AI buildout stops tomorrow. It means the expected return from here gets worse and the volatility profile gets nastier.

GUY: Right. Late-stage winners can still go higher, but position sizing has to change. Owning a crowded winner after it doubles is not the same thing as owning it before the crowd found religion.

AVA: Over on Odd Lots, Mark Cudmore added a really interesting twist... he argued that Greater China tech may now be the more sustainable AI equity trade relative to U.S. megacap AI because the valuations are less extreme. That does not mean China suddenly has better core chip IP than the U.S. It means the market may have overdiscounted China’s participation in the AI ecosystem.

GUY: I think that’s worth taking seriously. U.S. megacap AI names have become everybody’s default answer. At some point, the better trade is not the best story... it’s the less crowded expression of the same story.

AVA: And over on Odd Lots, the summit reaction mattered because it suggested traders are willing to imagine a version of the future where China still captures a meaningful chunk of the AI stack... maybe not at the bleeding edge of frontier training chips, but in applications, domestic cloud, inference, data services, industrial deployment, all of that.

GUY: Yeah. You don’t need China to “win AI” for China tech to rerate. You just need the market to move from “permanently excluded” to “less excluded than feared.”

AVA: Now let’s talk about a company-level case that’s fascinating... over on Business Breakdowns, in the episode “How Kaz Nejatian Is Rebuilding Opendoor,” Kaz Nejatian said Opendoor, ticker OPEN, was EBITDA positive as of April 1st, 2026 and is on track to be adjusted net income positive by year-end 2026.

GUY: That’s a huge shift in the story. For a long time, Opendoor was basically discussed like a financing accident waiting to happen. If they’re genuinely moving toward sustained profitability, the debate changes.

AVA: Exactly. And over on Business Breakdowns, Nejatian’s key reframing was that Opendoor is a market maker, not a prop desk. I thought that was a really smart way to describe it. The business should be judged on spread discipline, inventory velocity, data advantage, attach rates, and cost of capital... not on swinging for maximum gross profit per house by taking directional housing risk.

GUY: That market-maker framing also explains why greed kills you in that model. Over on Business Breakdowns, Nejatian used that example where if the seller thinks a house is worth $400,000 and Opendoor offers $300,000, only the sellers with adverse information accept. So if you widen spreads too much, you select for lemons.

AVA: Right. It’s adverse selection in plain English. If your bid is too low, the only people who hit it are the people who know something you don’t. That is exactly how you torch a market-making model.

GUY: And over on Business Breakdowns, the profit pool around the transaction is bigger than people think. He laid it out pretty specifically... 6% to 7% in core transaction costs, 1% to 2% in title and escrow, 300 to 400 basis points in mortgage margin, and 100 to 200 basis points in insurance margin. So the goal is not one fat spread on the home... it’s becoming the thin waist of the whole housing transaction.

AVA: That “thin waist” concept is so important. Over on Business Breakdowns, Nejatian basically wants Opendoor to be the checkout layer for housing. Control title and escrow first, because that’s the choke point, then attach mortgage, warranty, insurance, and everything else at lower customer acquisition cost.

GUY: It’s a very platform-like ambition in a market people still think of as local and fragmented. If he pulls it off, the value is not just in buying houses. It’s in orchestrating the whole workflow.

AVA: And over on Business Breakdowns, he also said Opendoor has fewer than 70 engineers and expects even non-engineers to use Claude, ChatGPT, Codex, and Grok for tasks like writing SQL. He basically said people who can’t do that should not be working at a tech company. That’s a strong cultural filter.

GUY: I actually think more companies are going to move that way. AI use becomes part of baseline literacy, not a specialist tool. The interesting bit is that it can widen productivity gaps without requiring huge headcount growth.

AVA: Which brings us to hidden infrastructure. Over on Lex Fridman Podcast number 496, the FFmpeg conversation was incredible. VLC has been downloaded more than 6 billion times. FFmpeg underlies YouTube, Netflix, Chrome, Discord... and the guests argued it may be among the biggest CPU consumers in the world.

GUY: That’s one of those “the internet runs on weird, underloved software” stories. Everybody obsesses over the flashy app layer, but the substrate is doing absurd amounts of work.

AVA: Over on Lex Fridman Podcast number 496, they also made a point that matters for AI video and edge computing... up to 45% of files may not be GPU-decodable, so you need software fallback after checking codec and device support. That means if your system assumes hardware acceleration everywhere, you’re going to break on a huge long tail of real-world content.

GUY: That matters more than people realize. AI-generated video, multimodal systems, content moderation, consumer playback... all of that depends on moving media around reliably. If half the weird real-world files need software decode, then CPU and optimization still matter a lot.

AVA: And over on Lex Fridman Podcast number 496, they discussed compression as an industrial-scale optimization problem. Audio often gets compressed around 10x, video around 100x to 200x, with ambitions toward 1000x in some contexts. That’s not just a cool technical detail. Every improvement lowers bandwidth, storage, and compute costs across the whole digital stack.

GUY: So again... the AI story becomes physical. Better compression means less bandwidth, less storage, less inference-adjacent cost, less data-center stress. The glamorous frontier model narrative is resting on this giant pile of engineering thrift.

AVA: Over on No Priors, in the episode “Fixing the Internet with Jaron Lanier,” Lanier gave the darker frame for all of this. He warned that low-friction digital networks don’t democratize value by default... they centralize it. One node gets influential, then becomes famous for being famous, and dominance compounds.

GUY: I think he’s basically right. Networks tend to concentrate value at the orchestration layer. It doesn’t matter if the open-source foundation is broad and collaborative... the profits usually pool where distribution, compute, customer ownership, or payment control sits.

AVA: Which is why Opendoor is interesting in that context too. If a company becomes the transaction layer, it can centralize bargaining power even in a market that looks fragmented on the surface.

GUY: Okay... before we go fully into philosophy, let’s move into geopolitics, because this is where the market, the AI buildout, and physical bottlenecks all collide.

AVA: Over on Odd Lots, the Trump-Xi summit was explicitly treated as an economic event, not just a statecraft event. That’s a really important shift. Diplomacy is now being priced through trade policy, AI restrictions, semiconductor access, and supply-chain normalization almost instantly.

GUY: Right. We used to think geopolitics was this separate box over in the corner. Now geopolitics is just another line item in the AI model. If Trump and Xi smile at each other, KWEB rallies. That’s the world we’re in.

AVA: And over on Odd Lots, Jensen Huang’s presence amplified that interpretation because he is almost a walking symbol for the AI capex cycle. If he’s anywhere near the delegation, traders assume the summit has semiconductor implications, even if nobody says so directly.

GUY: The market hears “Jensen” and thinks export controls, China revenue exposure, inference demand, and maybe a less hostile environment for the ecosystem. It’s Pavlovian at this point.

AVA: Hold on though... that relief rally could be fragile. Over on Odd Lots, Cudmore was pretty clear that if the summit avoids a breakdown, markets rally first, but if Hormuz tensions persist for the next month, that could overwhelm the diplomacy bounce by pushing oil, inflation, and yields higher.

GUY: Exactly. One shock gives, another takes. Markets can celebrate one de-escalation and then get hit by a different transmission channel entirely. That’s why I keep coming back to the bond market. It’s less sentimental.

AVA: Over on The Meb Faber Show, Tom Lee basically agreed that current oil pricing may not reflect the “acute shortage” developing in petroleum products if disrupted traffic through the strait continues. He was careful not to slap a giant target on crude because resolution could still happen. But the message was... don’t get lazy just because front-month crude hasn’t gone vertical yet.

GUY: Right. Crude is the headline price. Products are where pain can show up first. Diesel, jet fuel, feedstocks... the real economy runs on those, not on CNBC chyron symbols.

AVA: And over on In Good Company, Aliko Dangote gave the most concrete evidence of that. He said urea went from around $400 per ton in February to $850 after the Middle East crisis. Polypropylene in the UK jumped from about $900 to roughly $3,000. And aviation fuel was oversold into mid-July even while his company was producing 20 million liters per day.

GUY: That’s not theory. That’s actual scarcity showing up in actual product markets. When fertilizer doubles and petrochemicals triple, you can’t wave that away as “headline noise.” That’s industrial inflation with teeth.

AVA: Exactly. And it also shows that geopolitical stress redistributes profits. Over on In Good Company, Dangote’s businesses benefited because local refining and logistics control became more valuable in a tight market. So war risk doesn’t just create fear... it creates winners with physical capacity.

GUY: That’s a massively underrated point. In every geopolitical scare, people ask “what does it hurt?” They should also ask “who has the scarce thing?” Refiners, shippers with access, fertilizer producers, petrochemical players... sometimes those become the accidental geopolitical winners.

AVA: Over on The Compound and Friends, the hosts made another fascinating argument... that markets themselves may be the main real-time check on U.S. policy extremism now. With Congress less constraining, they suggested stocks falling and bond yields rising can force policy retreats. They even used that “TACO” shorthand... “Trump always chickens out.”

GUY: It’s a goofy phrase, but there’s a serious idea underneath it. If policymakers are willing to push until asset prices scream, then asset prices become part of the policy process. Markets are no longer just observers. They’re referees... or at least fire alarms.

AVA: Do you buy that?

GUY: Partly. I think markets can discipline at the margin, especially when both stocks and bonds move the wrong way at once. But I don’t think you want to base your portfolio on the assumption that policymakers will always blink in time. Sometimes they don’t. Sometimes they blink late. Sometimes they create the mess and then “fix” only half of it.

AVA: Right. And over on Excess Returns, Constan’s version of that reflexivity is more about central banks than politicians. He argued the Fed repeatedly extends speculative regimes by easing after local stress events... LTCM in the late ‘90s, regional-bank stress in 2023. His point is that policy reactions can prolong bubble behavior even with inflation still above target.

GUY: That’s why markets keep treating drawdowns as buying opportunities. They’ve been trained. If enough pain shows up somewhere, the adults come in with softer conditions. That conditioning may be rational... until it isn’t.

AVA: And over on The Meb Faber Show, Tom Lee added another policy wrinkle... the transition to a new Fed chair. He said 11 of the last 13 new Fed chairs saw at least a 10% drawdown in their first year. That’s a remarkable stat.

GUY: It is. A new Fed chair is like a new driver taking over on the highway. Even if they’re competent, the market tests them. It tries to figure out the reaction function... how they think about inflation, employment, financial conditions, all of it. That uncertainty alone can create turbulence.

AVA: So geopolitics right now is not just “war bad, summit good.” It’s really about how diplomacy, energy chokepoints, and policy reaction functions feed into inflation, yields, and AI supply chains.

GUY: Exactly. This is one system now.

AVA: Which takes us nicely into the cross-currents, because the most interesting part of today isn’t any single headline. It’s how the pieces fit together.

GUY: Let’s do it.

GUY: Over on The Compound and Friends, Barry Ritholtz’s interpretation of this market behavior is that it reflects real macro resilience. Over on Excess Returns, Andy Constan’s interpretation of very similar behavior is that it reflects bubble psychology. That is the biggest intellectual disagreement in the whole briefing.

AVA: Right. Same facts, different lens. Repeatedly buying geopolitical dips can mean investors have correctly learned that the U.S. economy is less oil-sensitive than before. Or it can mean investors are trapped inside a dominant narrative and reflexively rationalize every interruption.

GUY: And the honest answer is probably... both. The U.S. really is more resilient than in the 1970s. Household energy burden is lower. Domestic production is higher. But that doesn’t mean every buy-the-dip is wise. Sometimes a stronger economy just gives a bubble more room to inflate.

AVA: I like that. Resilience can actually extend speculative excess. Strong foundations don’t prevent overpricing... they can support it.

GUY: Exactly. Over on Odd Lots and The Compound and Friends and Excess Returns, the more subtle point is about transmission. Higher oil does not necessarily cause an immediate recession, especially in the U.S. But it can still matter enormously for equities through the long end of the curve. If inflation looks stickier, the 30-year yield above 5% becomes a problem for the exact sectors that have led the market.

AVA: So the mistake would be saying, “Oil won’t crash GDP, therefore oil doesn’t matter.” No. Over on Odd Lots, the danger path is through higher discount rates. And over on Excess Returns, that’s especially relevant if semis and AI are being valued on aggressive forward assumptions.

GUY: Right. If your valuation depends on years of future growth discounted at a friendly rate, a higher long end is poison. It attacks the denominator, not the numerator.

AVA: Over on The a16z Podcast, The Meb Faber Show, In Good Company, and Lex Fridman Podcast number 496, another big synthesis is that the AI boom is becoming physical. Horowitz talked about transformers, power, cooling, chips. Tom Lee talked about petroleum-product tightening. Dangote talked about fertilizer, polypropylene, and jet fuel. FFmpeg’s creators talked about the hidden computational scale of media delivery.

GUY: That convergence is huge. Digital abundance sits on top of scarce atoms... power equipment, semiconductors, cooling, fuels, network hardware, CPU cycles. If you keep treating AI as just a software multiple story, you’re missing where the bottlenecks are moving.

AVA: And those bottlenecks may also be where some of the next returns migrate. Maybe the obvious platform names keep winning, but maybe more of the economic value starts showing up in industrials, utilities, equipment suppliers, and less glamorous infrastructure.

GUY: Yeah. The next great AI trade might look less like buying another cloud platform at 30 times sales and more like finding the company that can actually deliver the transformer in nine months instead of eighteen.

AVA: Over on Odd Lots, Excess Returns, and RenMac Off-Script, there’s also a potential regional rotation story hiding in plain sight. If U.S. semis and the KOSPI are entering bubble-ish territory, and Greater China tech is cheaper, then leadership in the AI trade may broaden geographically even if U.S. firms remain the technological leaders.

GUY: I think that’s one of the smartest non-consensus ideas in the brief. Not “sell America, buy China” as some dramatic slogan. More like... if the secular AI thesis stays alive, the incremental dollar may chase less crowded, less expensive expressions of that theme.

AVA: And over on Odd Lots, Cudmore’s point was specifically that China AI might now be more sustainable from a valuation standpoint. That doesn’t mean shorting NVDA tomorrow. It means the next phase of the trade could be less concentrated and more price-sensitive.

GUY: Right. Leadership broadens late in cycles sometimes. The first inning rewards the obvious monopolists. Later innings reward the suppliers, the laggards, the cheaper regionals, the picks and shovels nobody looked at when euphoria was highest.

AVA: Over on Business Breakdowns and No Priors, there’s another connection I really liked. Nejatian wants Opendoor to be the “thin waist” of housing transactions. Lanier warns that digital networks tend to concentrate value at the central node. Put those together and you get a broader lesson... AI-enabled platforms don’t just increase efficiency, they often centralize bargaining power.

GUY: Exactly. The control point gets paid. If you own the checkout layer, or the orchestration layer, or the distribution layer, you get to tax the flow. That’s bullish for platform businesses and less great for fragmented intermediaries.

AVA: But over on Lex Fridman Podcast number 496, FFmpeg and VLC complicate that story in a useful way. Some of the most essential pieces of the digital stack are open source and create enormous social value without capturing much economic value directly. So importance and market cap are not the same thing.

GUY: That is such an important reminder. Value creation and value capture diverge all the time. Open source can make the ecosystem possible while the profits accrue somewhere else... usually to whoever owns customers, compute, or monetization.

AVA: Over on The Compound and Friends, Excess Returns, and The Meb Faber Show, there’s also a feedback loop running through policy. Markets may discipline policymakers. Central banks may extend speculative regimes. New Fed chairs often trigger drawdowns. Put together, that’s a very unstable equilibrium.

GUY: Yeah. Investors buy risk because they assume policymakers will cushion downside. That expectation itself encourages more risk-taking. Then valuations get richer, concentration gets worse, and when something finally breaks, the downside is nastier. It’s the classic moral hazard loop, just with AI and geopolitics layered on top.

AVA: Over on RenMac Off-Script, Odds on Open, and Business Breakdowns, the risk-management advice across totally different domains converged in a way I thought was kind of beautiful. DeGraaf says the bubble signal is for sizing, not shorting. Livadas sizes shorts so 50% squeezes don’t kill him. Nejatian says widen spreads too much and adverse selection kills the model. Different fields... same lesson.

GUY: Which is... don’t be greedy in nonlinear systems. That’s the real meta takeaway. When feedback loops are strong, the smart move is usually to build for survivability, not perfection.

AVA: Exactly. The edge comes less from nailing the exact top or bottom and more from constructing something that can survive variance.

GUY: And the biggest cross-current of all... over on basically every source in this briefing, macro, geopolitics, and tech are no longer separable buckets. China diplomacy is an AI trade. Middle East conflict is a duration trade. Open-source video infrastructure is part of the AI compute stack. Opendoor is a housing company that’s really a data, capital-markets, and AI-operations story.

AVA: That’s the modern investing challenge. The old departmental map... geopolitics over here, semis over there, rates somewhere else... it’s obsolete. The action is in the links.

GUY: All right... let’s land this with what we’re watching, because the next few weeks actually matter a lot.

AVA: Over on Odd Lots, the first thing to watch is the Trump-Xi summit outcome over the next 24 to 72 hours. Markets have already priced in some de-escalation. If the communiqué is benign on trade and doesn’t hint at tougher AI restrictions, Hong Kong tech, China ADRs, and names like BABA, TCEHY, and BIDU could extend the relief rally. But if expectations got ahead of reality, that reversal could be fast.

GUY: Yep. When KWEB jumps 5% on the hope, the burden shifts to the follow-through. Relief rallies built on diplomacy are great until the footnotes come out.

AVA: Over on Odd Lots, The Compound and Friends, and The Meb Faber Show, the second thing is Strait of Hormuz traffic and petroleum-product flows over the next several weeks. This is the hinge. If disruption stays limited, Ritholtz’s resilience thesis probably wins. If shipping impairment persists into June, then Cudmore and Tom Lee’s inflation-and-shortage warnings become much more serious.

GUY: Watch not just crude... watch products. Jet fuel, diesel, petrochemicals. Over on In Good Company, Dangote’s comments suggest downstream tightness may already be more severe than headline oil prices imply.

AVA: Over on Odd Lots and The Compound and Friends, the third thing is U.S. inflation and the long end over the next one to two months. We’ve already got hotter PPI and a 30-year Treasury yield above 5%. The June and July CPI and PPI cycle will matter a lot. If energy stress starts bleeding through and the long end stays elevated, duration-sensitive equities could feel it even without an obvious recession scare.

GUY: Right. If the 30-year keeps grinding higher while the S&P pretends everything is fine, eventually one of them has to blink. My bias is... the stock market has been a little too relaxed.

AVA: Over on RenMac Off-Script, the fourth thing is the SOX and the KOSPI over the next 6 to 12 months. Both have already entered what deGraaf calls bubble zone. Historically, that is not a great setup for forward returns. It doesn’t mean they dump tomorrow. It means any continuation higher should probably be handled with tighter risk discipline.

GUY: Especially the KOSPI with Samsung and SK Hynix at 43% combined. That is a concentration stat worth memorizing.

AVA: Over on The Meb Faber Show, the fifth thing is the Fed reaction function in the second half of 2026 as leadership changes. If Lee’s stat holds and the new Fed chair’s first year brings a 10% drawdown, markets will be hypersensitive to any hints that the new chair interprets inflation differently or tolerates tighter conditions longer.

GUY: Translation... don’t just watch the dots, watch the philosophy. A new chair can change the market’s mental model of what data matters.

AVA: Over on In Good Company, the sixth thing is aviation fuel tightness into mid-July 2026. Dangote said production was effectively oversold through mid-July even at 20 million liters per day. If that remains true, it validates the thesis that the oil complex is underestimating product-level scarcity.

GUY: And if aviation fuel stays tight into peak summer travel, that’s not a niche issue. That touches airlines, freight, inflation optics, all of it.

AVA: Over on Business Breakdowns, the seventh thing is Opendoor’s profitability milestone into year-end 2026. Management says EBITDA positive as of April 1st and adjusted net income positive by year-end. If they hit that, the market has to stop valuing it like a permanent casualty and start asking harder questions about attach rates, funding edge, and whether the “market maker, not prop desk” framing is actually working.

GUY: I’m not saying it becomes the next great compounder automatically... but the conversation changes a lot if OPEN proves the unit economics are durable.

AVA: Over on The Compound and Friends, the eighth thing is the accumulation of recession risk into Q4 2026. This is less a dated catalyst and more a scoreboard. Watch labor softness, tariff swings, energy inflation, and high-yield spreads. If enough of those deteriorate together, Ritholtz’s 50-50 recession odds by Q4 stop sounding conservative.

GUY: And over on The Meb Faber Show, watch those high-yield spreads carefully. If equities wobble but credit stays contained, that’s one signal. If both start to crack, that’s a different regime.

AVA: So the short version of our watchlist is... diplomacy headlines, Hormuz shipping, June and July inflation data, the 30-year yield, semiconductor leadership, Fed messaging, aviation fuel into mid-July, and whether companies like Opendoor can actually deliver on operational turnarounds in a higher-rate world.

GUY: Right. And underlying all of it is one big question... can the market keep looking through shocks, or are we getting closer to the point where the bond market, the physical world, and valuation gravity all start speaking at once?

AVA: That’s the question. Because if they do, this goes from “impressive resilience” to “very fragile equilibrium” pretty quickly.

GUY: And if they don’t... the melt-up can keep going longer than the skeptics want to admit.

AVA: Which, annoyingly, is often how these markets work.

GUY: It really is. All right... that’s where we’ll leave it for this morning.

AVA: Thanks for starting your day with us. We’ll be back tomorrow to see whether diplomacy held, yields behaved, and the market is still pretending every problem is somebody else’s problem.

GUY: Have a great Friday... stay sharp, and we’ll see you next time on Morning Signal.