There's Going To Be One Hell Of A Hangover When The Market Party Ends | Louis Gave

AI capex spending accounted for more than three-quarters of the increase in US GDP growth this year, meaning without AI capex the economy would be showing minimal growth—this creates a dependency where continued capex becomes essential to avoid recession.

“You look at AI capex, it basically accounted for more than 3/4 of the increase in GDP growth.”

China's economy is divergent: exports are booming but consumer/domestic demand is crushed by real estate bust, government interference in lending, and destroyed business/consumer confidence. Recovery requires real estate stabilization in major cities (Shanghai, Hong Kong, Shenzhen showing signs) but remains problematic in smaller cities.

“consumer confidence and business confidence have been absolutely crushed. absolutely crushed by the real estate bust by uh government interference on on what the banks could do and where the money could go...you have this sort of very divergent economy where some parts are doing extremely well. Uh and but everything linked to consumption remains uh deeply problematic.”

Japan is the template for where the Western world is heading: negative yields, QE, zero rates, and now positive yields with 300 basis points of term premium, all while maintaining 3.5% inflation. This represents the 'endgame' where governments use inflation to reduce debt burdens and bonds stop being an effective portfolio hedge.

“I've lived uh my entire career I'm 52 years old. I've lived my entire career with Westerners making fun of Japan, looking at Japan and saying these guys are, you know, when Japan first adopted zero interest rates, everybody thought these guys are nuts. Uh and then within a few years, everybody was doing zero interest rates. Then they did then they did QE and then everybody thought, "Oh, these guys are nuts." Uh then they did negative interest rates. Same story. And now what are we seeing in Japan? We're seeing short rates at 1% and inflation at 3 and a half. uh and a yield curve with 300 basis points between the 1 to 30 years. And I think this is exactly what's going to happen in most of the western world.”

The private markets industry grew from $1 trillion to $15 trillion over 20 years partly on promises of high double-digit returns with low volatility, but these promises have largely not been met by the large majority of the $15 trillion raised, partly because scaling these strategies to enormous asset bases typically results in lower returns.

“the private markets grew from 1 trillion to 15 trillion in the space of essentially 20 years. Um, and that was done on the promise of uh high double-digit returns and very low volatility uh which obviously we're finding out 15 years later has not always been met. It's been met by some, but not by the large majority of uh of that 15 trillion that was raised. Partly because it's one thing to deliver high returns and low volatility on 1 trillion. Doing it on 15 trillion is much more challenging. It's the whole history of capitalism. The more money flows into something usually the lower lower the returns.”

Most Americans cannot easily buy gold despite theoretical interest because regulatory and structural barriers (401k plan restrictions, lack of gold options in typical plans) limit access, and even when options exist, most pension funds refuse to allocate to precious metals.

“but buying gold isn't that easy in America, right? Well, and a lot of 401ks won't let you buy GLD. They'll they'll give you six options and that's it of what you can invest in. Yeah. and and and if you're a pension fund, if you're I don't know the the pension fund of the state of Texas or the state of Mississippi, etc., do you have 10% in gold to 20% in gold? No way. No way.”

Japanese financials have crushed it in 2026 due to a steep steepening of the yield curve, pickup in capital spending, and fiscal stimulus; Canadian financials have also performed extremely well despite expected weakness from real estate troubles; US financials have been flat despite strong economic data—this divergence is a key indicator of where systemic risks are not showing up.

“if you look at Japan financials have absolutely crushed it. Uh they've monster performance by financials in Japan which makes sense to to the extent that you've had a big big steepening of the yield curve over there. uh that you are seeing a pick up in capital spending across the board...Canadian financials have crushed it. uh Canadian financials are having a monster year...meanwhile go south of the border and you look at the US where the economy has been quite strong where on paper things should be you know should be going gang busters uh according to most of the economic data and the banks really haven't done very much.”

Gavk's analytical framework divides all economic scenarios into four quadrants: (1) deflationary boom (capitalism's natural state—entrepreneurs always try to do more with less), (2) inflationary boom (democracy's natural state—politicians always try to promise more without paying for it), (3) inflationary bust (energy price spikes), and (4) deflationary bust (when banks overlever and credit collapses). Asset allocation should be based on which scenario you believe is least likely, then eliminate exposure to that scenario.

“our framework, we start off with the idea that economic activity is driven by changes in sorry that asset prices asset prices are driven by changes in economic activity and inflation. And that this this essentially gives you four scenarios. uh you have the deflationary boom. Now the deflationary boom is the natural state of capitalism. Every entrepreneur, every CEO wakes up every morning thinking how can I produce more with less?...the natural state of democracies is the inflationary boom because every politician or almost every politician wakes up every morning thinking how can I promise more to voters without having to pay for it.”

Any investor evaluating the AI trade should assess it through four critical prisms: (1) fundamentals (stretched today), (2) momentum (undeniably strong), (3) investor positioning (crowded), and (4) valuations (make no sense). When all four are either stretched or crowded, the risk-reward becomes unattractive regardless of the underlying business quality.

“when you look at an investment you have to look at it through four prisms. You have to look at the fundamentals and today there's no doubt that the fundamentals on AI are exciting but the numbers look very stretched. You have to look at it through momentum and here the momentum is undeniably very strong. You have to look at the investor positioning and here the investor positioning looks crowded and then finally you look at valuations and and this is the big challenge today is the valuations make no sense.”

Most people and businesses don't actually need the maximum performance AI (Ferrari); they just need 'good enough' AI to get their work done (Toyota). If the market shifts to this preference, capex demand plummets because 'whether you have a Ferrari or Toyota, you're stuck in the same traffic jam.'

“And I think the car is a good analogy because yeah, maybe corporate America there are some things where you're really going to need a Ferrari, but most people they just need to get to work on time, right? Just to get to work on time. And the reality is if all you need to do is get to work on time, whether you have a Ferrari or Toyota, you're getting stuck in the same traffic jam and you're getting stopped at the same traffic light.”

Yen appreciates roughly 10 big figures (10% or more) in a week roughly every 7-10 years, creating significant ramifications when it happens.

“every 7 to 10 years that the yen moves 10 big figures in like a week every now and then right and then when that happens it creates all sorts of of ramifications.”

Deepseek's competitive LLMs suggest China can compete on AI/ML via software optimization rather than pure hardware/compute, meaning US hyperscalers may not need the capex they're planning if 'good enough' models (95% of US performance at 5% of cost) satisfy most use cases. This threatens the AI capex boom.

“the latest models from Deepseek are quite competitive with a lot of the cutting edge LLMs here in America...could actually be again a threat to the AI sector here because if the capital spending decides to cool off by 20 or 30%”

Latin America is experiencing pronounced political rightward shifts (Salvador, Argentina, Bolivia, Ecuador, Chile, Peru, Colombia) with market-supportive policies and corporate governance changes, plus growth of local pension funds creating demand for local assets. These deeply undervalued local assets represent significant opportunity versus crowded global AI trade.

“I've argued with you in the past and um and I've written a lot of pieces in the past couple of years highlighting that there's a lot you know very exciting changes occurring all across Latin America political change corporate governance changes uh growth of local pension funds all of which argue for a rerating of local assets that are deeply deeply undervalued. Uh today we today we have the results of the Colombian election. It's yet another election after Salvador, after Argentina, after uh Bolivia, after Ecuador, after Chile, after Peru. Uh now we have Colombia uh with a with a pretty pronounced rightward shift. Um and and policies that are likely to be much more market supportive.”

China sets the price of oil: it buys as much as possible at $65/barrel (the floor) and stops buying at $100/barrel (the ceiling), effectively capping the global oil price trading band, making $65-100 the range within which most countries can function, with $70 approximately the sweet spot for the U.S. and countries above $100 experiencing real economic pain.

“you're in that trading band. So if you think okay at at 100 I know that China backs off from the market for me this is one of the big lessons of the swarm moose crisis is China is the price setter of oil today. Uh this is what I think we have to take away from this moose crisis. Forget what happens in the Middle East. forget what happens in in US policies etc. The bottom line is China buys as much oil as it can at 65 and stops buying at 100 and that's enough to regulate the market.”

AI and AI-related stocks now comprise 45% of the S&P 500 (not the NASDAQ), representing an extreme concentration where a single thematic trade dominates the entire broad market.

“AI and AI related stocks now comprise 45% of the S&P, not the NASDAQ, but the S&P, right?”

Chinese government bonds have outperformed (up ~5%) while US Treasuries and German bonds have underperformed, making Chinese bonds an attractive investment relative to Western bonds.

“every time I spoke about China it was to say that you wanted to buy Chinese bonds. Uh that the best investment out there were where where Chinese government bonds uh which by the way are up about 5% this year when US treasuries are down and um and German boons are down etc.”

To hedge the 'Godzilla' (Japanese repatriation) tail risk, investors can buy out-of-the-money calls on the yen at attractive prices because FX volatility is extremely low despite significant equity/bond market moves and geopolitical uncertainty. Yen calls are cheap insurance analogous to fire insurance on a house.

“the Godzilla uh scenario to hedge is pretty easy. uh today you can buy some out- of- the- money calls on the yen for fairly little money because the the FX one of the interesting thing I think in 2026 has been how the volatility on FX has been so low given all the geopolitical uncertainties...today today the V in FX is is very cheap so you can you can hedge yourself against the Godzilla effect by buying calls on the yen.”

Japan's Prime Minister Takahashi is nationalistic and focused on re-industrializing and building defense spending domestically, similar to Trump's stated goals. However, Takahashi may strategically wait for the 2026 midterm elections and Trump's potential political weakness before implementing repatriation policies, avoiding immediate US political blowback.

“you have a prime minister in Japan that is very nationalistic. Um that is very keen on spending a lot of money domestically on re-industrializing Japan on building a new defense industry in Japan. Not that different from everything President Trump talks about.”

The combination of AI capex correction + Japanese capital repatriation (Godzilla vs. Giant Robot) could create a two-shoe drop: AI correction causes investors to lose confidence and withdraw from US assets, which then amplifies the repatriation flows, creating a potential spiral of Western asset selling.

“so we we could have a one-two punch here where the AI complex starts to go through a correction and these Asian countries start really taking their assets out...And the reality is today in the US, the exciting story is the AI story. uh if if that one starts to break, it gives you it encourages investors to pull their money out anyway. Uh because which actually then slows the robot down.”

US hyperscalers are currently producing frontier AI models at below cost (like selling Ferraris for $2,000) to hook users, with the aim of extracting full price later after market lock-in is achieved.

“so that's one vision and So in that vision, what people are essentially doing now is is saying we need to produce the best Ferrari and in the process of doing it, we're going to sell Ferraris for $2,000 to people. And so all of us right now are getting, you know, anthropic or uh or CLA, you know, for for nuts for a lot less money than it should be given how much it costs to produce. And we're like, wow, this is awesome. Um, and I guess the aim is to get us hooked and then down the road to say actually you got to pay the full price for the Ferrari.”

The key question in AI investment is not whether you'll win but whether you can articulate what edge you have that the market doesn't know—if you can't answer that for AI capex, you're just playing momentum and need to ask what hedge you have for that momentum. Honest answer: there isn't a good hedge.

“the big challenge I think right now for any investor is to look at this boom and decide okay this AI thing may be very much for real. Um but where are we in this capex cycles? Do the numbers still make sense?...the big IPOs to the to the you know every front um every headline in the newspaper being uh being about AI. The question is you know what what edge do I bring to the table? Like what do I know that the market doesn't know? Um, and this isn't to say that this thing can't go keep going on, but what you're essentially playing is a momentum game. And then, you know, the question has to be what hedge do I have [in this momentum game]?...I'm not personally I'm not sure I have any.”

The traditional 60/40 portfolio is dead; the new recommended allocation is 60% equity, 20% energy, 20% metals, with the specific choices within each category depending on which economic scenario unfolds.

“The new portfolio, the new portfolio, you know, my 2021 book, I hate to plug my book, but I'll do it anyway. Uh, my 2021 book, Avoiding the Punch, the the whole book, the the premise was look, the 6040 is dead. what you now need is 60 equity, 20% energy, 20% uh 20% metals.”

Japan has an enormous net foreign investment position (80% positive, roughly $3.5 trillion in US assets alone), and as yields rise in Japan, Japanese investors may start repatriating capital, selling US Treasuries and other foreign assets to buy Japanese government bonds.

“Japan has an 80% uh negative sorry positive position uh relative to the rest of the world. And so you have to wonder at some point as yields go up when do the Japanese start repatriating capital. Um and for me that will be when they finally start doing that'll be the big the next big leg down on bond markets everywhere because what they'll do is they'll be selling treasuries. They'll be selling French OATS they'll be selling French boons to buy back their own JGB markets.”

China is the price setter of oil in global markets; this was a major lesson of the recent Houthi crisis—forget Middle East politics and US policies, what matters is Chinese demand.

“this is one of the big lessons of the swarm moose crisis is China is the price setter of oil today. Uh this is what I think we have to take away from this moose crisis. Forget what happens in the Middle East. forget what happens in in US policies etc. The bottom line is China buys as much oil as it can at 65 and stops buying at 100 and that's enough to regulate the market.”

The inflationary boom is the most likely scenario today because: (1) oil prices are capped near $100 (China stops buying above this), (2) US deficits remain at 7% of GDP, and (3) oil prices were capped at $120 during the recent Middle East crisis but didn't get there, so inflationary bust from energy spike is now unlikely. In an inflationary boom, the winning assets are value stocks, metals, and financials.

“I think the highest odds to be honest remains the inflationary boom. uh it remains the inflationary boom because we had a chance at an inflationary bust if oil prices moved to 120 bucks but we didn't go there for reasons we can discuss if you want uh we we didn't end up going there um and I think the odds of us going there now that the whole situation seems to be resolved the odds of us going there are pretty low so if we don't get 100 to $120 on oil and if the US government continues to run budget deficits of 7% of GDP I think the highest odds are, you know, you're going to be in an inflationary boom.”

Currently, international bank shares are not flashing red signals of a deflationary bust, and most countries' bank performances do not suggest imminent recession, making a deflationary bust the least likely of the four economic scenarios in 2026.

“the bank share performance isn't pointing in that direction right now. Um and you know in most countries that the performance of bank shares don't seem to be flashing big red signals. Um so I think that of all the the four so that's the first reason.”

There is an arms race in US AI development to be the best performer and reach AGI quickly, driven by beliefs that being number one captures the entire market, based on previous business models (Amazon, Uber).

“I think the big question on the AI is is the one you just highlight. I think in the US there's a sort of arms race to have the the best performing uh and to try to get to AGI as quickly as possible and and to dominate that and and a feeling uh perhaps because of the previous business models of Amazon, of Uber, etc. that if you're number one uh then you'll capture the whole market.”

For Chinese consumers to recover, they need reprieve on their balance sheets and real estate to stop falling, which is a prerequisite for broader consumption recovery.

“I think for for China to for the consumer to really recover, he needs reprieve on the his balance sheet. He needs he needs the basically you need the real estate to stop falling.”