Competition Is For Losers In The AI Era: A Geographic Perspective

The future of innovation is global. We discuss it here.

This post was first published in my Forbes column.

Last month I had two discussions that are actively shaping my thinking in AI investing.

A VC, when discussing AI applied to one of the hottest fintech sectors, accounting, told me flippantly: “no self-respecting VC does not have an AI accounting play yet.”

Afterwards, a founder pitched me an AI-native insurance broker for small businesses. It was a strong team with promising early traction. My first reaction, said out loud: there will be a hundred companies doing this in the US (upon verification, I’ve already found many!).

Peter Thiel argued that competition is for losers. His take is that lasting value comes from building monopolies. This has become Silicon Valley gospel - despite many similar AI companies getting funded anyways.

The good news: escape has a geography.

The US, and particularly the Bay Area, dominates access to venture capital. The United States captured 57% of global venture funding in 2025, and the Bay Area alone took 60% of global AI funding. AI is radically lowering the barriers to build and scale startups.

The same strategies applied in other geographies may just be more interesting. Two structural advantages of building fintech (and startups more broadly) outside the SF-NY axis are about to be amplified by the AI rebuild moment. The first is less competition, and the second is product breadth.

A Hundred Versus Three

Pick a fintech category: AI accounting, AI underwriting for SMBs, AI-native insurance brokerage, payroll, spend management, etc.

In the United States, each of these has multiple funded competitors. The startup graveyard is littered with fourth-place finishers who raised big rounds, yet still lost.

Ask coinvestors and operators in Latin America, Southeast Asia, MENA, or Sub-Saharan Africa which fintech categories are actually contested, and the counts come back small: typically two or three serious contenders in Brazil and/or Spanish-speaking LatAm, and similar or even fewer in Southeast Asia, MENA or Sub-Saharan Africa.

The standard venture model assumes that the best companies attract the most competition, and that winners emerge from that crucible. That holds in capital-abundant, talent-saturated markets. Because the venture pool is smaller globally, the dynamics are different. In a market with three serious contenders, the diligence work that gets you to the right answer is the same, yet the probability of picking the right one is dramatically higher.

I have lived this directly. At previous firms, I partnered with Neon, a digital bank serving underbanked Brazilians, when the country had a handful of serious neobank attempts. They did not win the market (behind Nubank) but still became a unicorn. Similarly, in BNPL, I got to work with Kueski in Mexico and Kredivo in Indonesia, the category leaders in markets with only a handful of credible players.

The competition math will turbocharge in the AI era.

The Compound Startup

Parker Conrad coined the term “compound startup” to describe what Rippling does: build multiple integrated products in parallel rather than focus on one wedge.

He argued, against Silicon Valley conventional wisdom, that breadth was an advantage. The argument was contrarian because of competition - every piece of the US business stack already had a specialized, well-funded incumbent: Stripe for payments, Gusto for payroll, Ramp or Brex for corporate spend and so on.

In fintech outside the United States, the compound startup is often the natural outcome. In Out-Innovate, I wrote about the horizontal stack, but this opportunity is even more powerful in the AI era. For example, in Korea, one of the world’s largest financial services markets, Toss grew from peer-to-peer transfer in 2015 into more than 70 financial products in a single app. In Kazakhstan, Kaspi combined banking, payments, an e-commerce marketplace, and government services into one super-app that handles around 65% of the country’s digital payments.

Why AI startups more readily compound outside the US

Three conditions support the compound startup, and AI strengthens them.

First, is the cost of building product. In most of the world, the incumbent in any given financial product is a 40-year-old bank, a state-owned operator, or nothing at all. There are of course startups, but density is often lower. What changed is that a team of ten with AI-native tooling can now credibly build the second and third products that used to require a second and third company. And because fintech adoption relies on trust, it is much easier to sell the next product to an existing customer than acquire a new one.

Second, the data advantage of the 3Ds (distribution, data, delivery) is particularly powerful in the age of AI. Executing well on the first products and generating data exhaust catalyzes the second or third - much easier to build AI underwriting if you’re already managing corporate spend, for instance.

Third, regulation favors scale. Compliance is a fixed cost, and spreading it across more products improves unit economics; AI then compresses the marginal cost of compliance for platforms that already carry the fixed cost.

AI is empowering a rebuild. In the United States, that rebuild is happening in the most crowded conditions in software history. But, in global fintech, the same rebuild is happening with these two structural advantages.

And there is a third, under appreciated advantage: a head start on the platform clock. The hyperscalers and frontier labs will eventually turn to every market, but their roadmaps start in English and in the US. Local players globally are building AI-native products on local rails, in local languages, with local licenses and with local context, years before that attention arrives.

Perhaps when it does, a similar pattern will repeat: Amazon reached Latin America and found Mercado Libre entrenched; Uber met Grab in Southeast Asia and Didi in China and ended up selling to both.

For the first time, the tools to build are as available in Mexico City and Jakarta as in San Francisco. The competition is not (yet).