America Has a Comparative Advantage in Handling Global Disorder

In this issue:

• America Has a Comparative Advantage in Handling Global Disorder—Geography and demographics aren't the sole determinants of destiny, but the do set the boundaries of plausible destinies. For a while, there was a theory that energy independence would make the US more isolationist, but it just enhances the degree to which the US can afford to be interventionist while the rest of the world pays a relatively high share of the economic price.
• Where We Are in the Cycle—Small-time grifts and second-best deals aren't any early-cycle phenomenon, but they don't signal the end, either.
• Insurer of Last Resort—China, too, can offer the economic equivalent of insurance.
• Marketplaces and Routers—Anthropic opens a market.
• Getting a Bid—A software deal that would be more positive for the space if the company weren't protected by regulation, acquired by a strategic buyer, and, at least in share price terms, immune from the drawdown in the first place.
• Talent—Even if you don't care about AI safety, some of the best AI talent does.

This issue of The Diff is brought to you by our sponsor, MNX.

Chat with this post on Read.Haus.

The Diff March 9th 2026

0:00

/804.649796

1×

America Has a Comparative Advantage in Handling Global Disorder

In the late 2010s, Peter Zeihan started to get a lot of attention for making a real resource constraints-driven argument about great power competition. In his view, it was too limiting to attribute different national outcomes to different institutions—capitalist democracies had a good track record, sure, but plenty of countries were getting rich despite skeptical attitudes towards capitalism and implementations of democracy that were nonexistent (e.g. China) or mostly cosmetic (Japan during the peak MITI era).^[1] And it was simply naïve to have some Great Man theory, where the Berlin Wall was torn down because Ronald Reagan told Gorbachev to. In the Zeihan view, what matters is who has the key natural resources, who controls global shipping lanes, which country produces enough young people to realistically fight a war (or produce income to pay for retirees), which borders are defensible (he specifically calls out how Russia can always feel nervous about having a long border that's fairly easy to send tanks across), etc.

Zeihan argued that the US's increasing energy independence meant that it didn't have to be so engaged with the rest of the world, and, in particular, that it didn't need to keep the Middle East safe for oil companies. And that if the US withdrew from that responsibility, the entire global trade system would start to recede.

One reason this view got so popular was that it started coming true. He happened to publish the book just before fracking got crushed in late 2014, but that was just a blip: US oil production was back to all-time highs by late 2017, when he published a sequel. Meanwhile, trade barriers were going up, Russia had seized some Ukrainian territory and was making noises about taking more, and the world generally looked very different from the globalized, unipolar vision of the 90s. It complicated things a bit that these trends were downstream from institutional differences: the Republican party was vulnerable to capture by someone like Trump, who had a core of loyalists and a knack for driving up the unpopularity ratings of whoever else he was campaigning against; the Democratic party was similarly vulnerable to nominating Hillary Clinton, because the world's most obsessed Clinton-haters weren't voting in Democratic primaries. So, it's downstream from US party-level institutions that 2016 was a campaign between two of the most-disliked people in the country. And the Great Man Theory (values-neutral form of "Great," here) can also chalk up a win: even if it was somehow geostrategically inevitable that the US and China would disentangle their economies, or that Russia would expand its territory, those things happened under the rule of particular people who were carrying out their own idiosyncratic goals.

And now we're in a very confusing situation, where Zeihan's warning of a world with multiple trade zones, where there are different prices for oil and some places can't always get any, does look more likely than it did a few weeks ago, the way we got there was a series of decapitation strikes on two big oil-producing countries, Venezuela and Iran.

One way to reconcile this with the Zeihan headcount-and-resource-realism model is to note that these constraints compel only by providing constraints. Having lots of oil and gas is a necessary precondition for having a local industry devoted to extracting those resources, but it didn't mean that the Apache had a thriving oil industry by the time Europeans arrived. Resource endowments give countries more flexibility; they can be a pure raw materials exporter, or try to build the rest of the supply chain for more valuable end products. Sometimes, the exact specialization involves a mix of both: when Australia had cheap iron and China had cheap labor and had engineered a high savings rate, it made sense for Australian iron to be shipped to Chinese steel mills.

At one level, it's obvious enough to be tautological that copper will be mined in places with lots of copper, shipped to places that have copper-intensive manufacturing, and used in places with lots of people who can spend money on the resulting hardware and infrastructure. On the other hand, it is historically odd that resource-poor countries like Japan and China (with the exception of rare earths and coal) have achieved high growth; importing raw materials and processing them to sell value-added results is pretty new as a viable national economic strategy.

If you have a model where everyone who is willing to play by America's rules gets to happily trade with one another, and pursue their comparative advantage, you have a nice, stable equilibrium. But some countries have defected from that. Russia is a recent example, but Iran and North Korea have been disconnected from that system for a long time, and China had a closed economy that only gradually opened starting in the late 1970s. For countries that play by the rules, the main incentives are to keep doing so and to ensure that other countries do, too. But for the pariah states, there's a different range of outcomes. North Korea is actually a less worrying country than it looks, for example, because their domestic economy can't produce enough for them to survive, at least in the formal sector, but they can threaten their neighbors. (This is partly a form of internal politics, even though it involves interactions with other countries: running a nuclear weapons program keeps people busy, and periodically escalating tensions with neighbors makes it easier to paint coup plotters as Japanese or South Korean agents rather than as people who are sick of taking orders from Kim Jong Un.) In Iran's case, it's the reverse: because they have such a high breakeven price for oil, their whole economy is an out-of-the-money call option on oil, and volatility is likely to move it closer to the money. The ideal world for Iran is one where oil pipeline explosions and the closure of the Strait of Hormuz is a live possibility—just one that never quite materializes, so it's not their oil infrastructure blowing up or their shipments sinking.

Pariah states' main export is uncertainty, because they benefit from a world where the system they're in opposition to is shakier.

But the US, oddly enough, has its own version of pariah-state economics:

• The US is less trade-dependent than other big countries: trade is 27% of US GDP, compared to 57% for the next nine biggest economies. The US is worse-off, in an absolute sense, if we can't import cheap stuff from places with cheaper labor markets, but the countries with those labor markets are much worse-off.
• The US is a net energy exporter, and energy prices tend to be the most visibly responsive to higher risk in global trade. That doesn't mean that the US is autarkic, though; America's refinery footprint was built to handle oil from other markets, so there's a mix of imports and exporters to get the right grades to the right refineries.
• Anyone with memories of the 70s, or the 2000s, knows that higher prices at the pump are a brake on intervening in the Middle East. But gasoline expenditures as a share of consumption have been trending down for years: currently 2%, slightly lower than when The Economist had a cover story titled drowning in oil ($). Cheap oil is still better for consumers than expensive oil, but it's just not as pressing as it used to be. To hit mid-2008 levels, gas has to hit about $4/gallon, and stay there, with no change in consumption. (For what it's worth December futures are at $74 as of this writing, compared to the front month at $107. One thing that's pricing in is that hybrid work has probably made oil demand more elastic, mostly in the US but also in other places.
• Plenty of international financial actors are treating the US's reserve issuer status as more questionable than before. But their liabilities include dollar debts they took on in simpler times, and their usual reaction to any chaos is to hoard dollars, i.e. to make every other currency (and most asset classes) cheaper in terms of dollars.
• Donald Trump has no qualms about grading his own homework. Whatever happens in Iran, he's happy to paint a bullseye around it. This doesn't preclude a long, drawn-out conflict. But when's the last time anyone worried about America invading Canada or Greenland?

It's far from a positive development that the country with the biggest military and a highly improvisational foreign policy is also the one that bears the smallest relative share of the economic consequences of its actions. The economic story of the last decade and a half has been that the world grows, and the US grows faster. It would be tempting to think that the US is levered to growth, and would decline faster, too. That might be true in terms of initial equity price reactions, but over longer periods the US will probably hold up better.

A Word From Our Sponsors

Trillions of dollars are flowing into AI infrastructure and DRW’s Don Wilson recently predicted that “the total dollars spent on compute will exceed total dollars per year spent on crude oil in the next 10 years.” Yet unlike crude oil, there’s no futures market for training runs, no benchmark markets to bet on the speed of AI progress, and no public market for OpenAI shares despite a valuation nipping at the heels of Saudi Aramco. MNX is changing that. As compute eats the world, it needs the same financial infrastructure as every other commodities market. MNX is the first exchange built to trade the full AI value stack—from H100 rental rates to frontier model benchmarks to AI equity.

Try their testnet platform here.

Elsewhere

Where We Are in the Cycle

When there's an economic cycle driven by top-down capital allocation decisions, it's a mistake to look to fundamentals for guidance on where things stand. Those fundamentals are just sentiment from quarters earlier, reported on a lag. So it makes more sense to look at shorter-term indicators of sentiment. So you can collect datapoints like Oracle and OpenAI deciding that they're not so excited by a new datacenter, only for Nvidia and Crusoe to step in and try to market it to Meta. As long as there's a buyer for every asset, the market clears, and as long as there are two or more buyers, it can clear at a price that's okay for the seller.

The other interesting sentiment read to look at is fraud: an AI startup has accused its former founder of stealing data ahead of being ousted, and using it to start a competitor. Anyone who's perused a selection of 1999-vintage IPO prospectuses can internalize the idea that being right about the general theme doesn't mean making money from every expression of that trend. But if you look at SEC enforcement actions from that period, you don't see many high-profile dot com-related scams. Plenty of microcaps, but nothing beyond that scale. The big scandals tend to happen after asset prices have peaked, or more specifically after companies can't raise outside capital to meet their formal obligations or provide some semblance of the results their reported metrics imply. Small-scale alleged grifts are a late-cycle phenomenon, but not an end-of-cycle one.

Disclosure: long NVDA, META.

Insurer of Last Resort

One of last week's Diff articles talked about the Trump administration kicking around the idea of insuring cargoes that pass through the Strait of Hormuz ($), and how this is a fairly literal reading of the concept of government-as-insurance-company. The US government is not the only government that can be conceived that way, and is less dependent on that strait than many others. Some ships that pass through have broadcast messages indicating that they're Chinese. It's small-scale, and fairly temporary, but it's a notable change that in at least one body of water, and at least right now, there are two different countries who assume the right to protect international ocean-borne trade.

Marketplaces and Routers

Anthropic is introducing a new software marketplace, where customers can shop for software products, including vertical-specific AI tools that indirectly compete with Anthropic. This lines up with the Diff router thesis, which argued that models will route people to commercial endpoints, and that these endpoints can be companies, people, or other models. An LLM is an incredibly effective way to measure buyer intent—your customers are literally telling you exactly what they want and why—but the general-purpose ones won't necessarily capture the most value. For that, they want to delegate to specialists, who specialize both in what their specific AI application does but, more importantly, how they charge for it.

Getting a Bid

Talkspace, a virtual therapy company, is being acquired by Universal Health Services. This deal is a useful template for where companies nominally threatened by AI can still be valuable: therapy is a pretty popular use case for LLMs, but your insurance company doesn't have a way to fit a general-purpose tool like Claude into their billing system. So Talkspace is partly in the business of connecting people to therapists, and partly in the business of finding a good way to charge for therapy. It's also being acquired by a more full-stack company in the same supply chain, not by a pure software business, and not as a tuck-in deal by an AI company. But perhaps the most annoying feature of this deal, from the perspective of wondering when non-AI software stocks will finally hit their lows, is that Talkspace did that a long time ago. They're up 90% in the last year, mostly thanks to a big rally that took off around the same time the software sector started dropping. N-of-1, but investors correctly assessed that even if there was a product-level threat from AI, the companies with a completely different business model can ride that out.

Talent

AI safety has been a part of the AI discussion for a very long time; OpenAI was founded by people who were worried about the risks of powerful artificial intelligence, and Anthropic by people who were worried that OpenAI wasn't worried enough. For a long time, that issue was a theoretical one. But the more capable models get, the more there's a direct tradeoff between safety concerns and revenue. But that means there's also a tradeoff between immediate revenue and access to talent; OpenAI's head of robotics announced that she's leaving over the company's DoW deal. Because the AI safety movement has such a big head start in thinking that AI was a big deal, it's overrepresented among the (scarce) talent that AI labs really need to hire. So, short of the big labs being nationalized and their workers being drafted (as Truman once threatened to do with striking railroad workers), safety has a partial veto over how AI is deployed, even by the US government.

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• High-growth startup building dev tools to help highly technical organizations wrangle and autonomously test/debug complex codebases is looking for a senior design engineer to own their design system and build the visual abstractions customers rely on to simulate their software systems, find bugs, and quickly remediate them. A compelling portfolio, a rare blend of design and engineering chops, and a deep understanding of how the internet and browsers work required. (D.C.)
• A pre-IPO, next-generation chemicals company that’s manufacturing the mission-critical inputs for a sustainable American reindustrialization is looking for a CFO to own the capital raising roadmap and allocation strategy end to end. Experience turning corporate strategy into a data-driven narrative and advising on late stage capital raises and/or IPOs preferred. (Remote, Houston)
• Ex-Citadel/D.E. Shaw team building AI-native infrastructure that turns lots of insurance data—structured and unstructured—into decision-grade plumbing that helps casualty risk and insurance liabilities move is looking for forward deployed data scientists to help clients optimize/underwrite/price their portfolios. Experience in consulting, banking, PE, etc. with a technical academic background (CS, Applied Math, Statistics) a plus. Traditional data scientists with a commercial bent also encouraged. (NYC)
• A leading AI transformation & PE investment firm (think private equity meets Palantir) that’s been focused on investing in and transforming businesses with AI long before ChatGPT (100+ successful portfolio company AI transformations since 2019) is hiring experienced forward deployed AI engineers to design, implement, test, and maintain cutting edge AI products that solve complex problems in a variety of sector areas. If you have 3+ years of experience across the development lifecycle and enjoy working with clients to solve concrete problems please reach out. Experience managing engineering teams is a plus. (Remote)
• Series A startup that powers 2 of the 3 frontier labs’ coding agents with the highest quality SFT and RLVR data pipelines is looking for growth/ops folks to help customers improve the underlying intelligence and usefulness of their models by scaling data quality and quantity. If you read axRiv, but also love playing strategy games, this one is for you. (SF)
• Ex-Bridgewater, Worldcoin founders using LLMs to generate investment signals, systematize fundamental analysis, and power the superintelligence for investing are looking for machine learning and full-stack software engineers (Typescript/React + Python) who want to build highly-scalable infrastructure that enables previously impossible machine learning results. Experience with large scale data pipelines, applied machine learning, etc. preferred. If you’re a sharp generalist with strong technical skills, please reach out. (SF, NYC)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.