The Question That Reshapes an Entire Portfolio
There comes a moment in every investor's journey when deviating from consensus is not a romantic stance, but a matter of basic economics. A venture capitalist specializing in artificial intelligence has reached that point, documenting it with a clarity seldom seen in the sector: he has stopped asking which AI company deserves backing and started to inquire which problems are being systematically ignored by major players.
This reorientation is not merely semantic. It fundamentally alters the asset evaluation process. When analysis begins from gaps in the market rather than from available technological capabilities, the universe of opportunities expands into territories that larger firms have dismissed for reasons unrelated to business viability, but tied to their own opportunity costs. What is inefficient for a $500 million firm can be the foundation of a profitable business for someone willing to operate with smaller structures.
This has a direct consequence on the business models deserving attention today. It's not those that scale faster on existing infrastructure, but those that create demand where no organized supply exists.
The Problem with Building for Customers Who Already Have Credit Cards
The dominant narrative in enterprise AI points toward a very specific customer profile: medium or large companies in English-speaking markets with established management systems, in-house tech teams, and a willingness to pay monthly subscriptions in dollars. This profile feels comfortable because it reduces customer acquisition costs, facilitates integration, and produces predictable retention metrics.
The problem is that this profile is already saturated. Every week sees a new layer of SaaS tools built on the same foundational models to serve the same customer. The investor mentioned in Fortune refers to this as a phenomenon that keeps him awake at night—not due to fears of those companies failing, but because of the success of a model that systematically excludes the vast majority of the world.
Africa, with over 1.4 billion people and mobile adoption rates that surpass many Western markets in growth speed, represents the strongest argument against that logic. The problems that AI could solve there—from access to financial services to medical diagnosis in rural areas—do not require less technological sophistication but rather an entirely different product architecture and monetization model.
And therein lies the trap that most teams fall into when trying to expand into these markets: they bring the product they built for San Francisco or London and look to adapt it with a layer of localization. That’s not validation; it’s the exportation of assumptions. The result is almost always a technically functional product that nobody buys because it solves the wrong problem at the wrong price.
When the Business Model Precedes the Product
What distinguishes this investor’s thesis is the order of the questions. Before assessing whether a technology works, he asks if there is a viable payment mechanism for the targeted segment. That may sound obvious, but in practice, most product teams reverse this process: they build first and think about monetization later, when they have already committed too much capital to pivot without pain.
In emerging markets, that misordered sequence is fatal at a faster pace. Feedback cycles are shorter, margins for error are tighter, and user trust is built incrementally, not all at once. A model that doesn't demonstrate tangible value in initial interactions does not get a second chance.
What works, according to this perspective, are models that start with a minimal, verifiable, and repeatable transaction. Not a freemium model with expected conversion in six months, but a real transaction that confirms the user understands what they are buying and is willing to pay for it from the first contact. That metric—the initial sale in a new market with different conditions—holds more value than any prior market research.
The difference between an investor who backs this approach and one who adheres to consensus isn't in the appetite for risk. It's in where they seek evidence to reduce that risk. The first looks to user behavior against a real price; the second relies on projections built around untested markets.
The Next Cycle Isn’t Won by Those with the Best Language Model
There is a superficial reading of this story that reduces it to an argument about geography: investing in Africa because developed markets are saturated. That interpretation misses the most important point.
What this investor is describing is a shift in the logic of value creation. Over the last three years, capital has flowed to those with preferential access to foundational models who can build product layers on top with speed. That window is closing as access to those models has become democratized and technical differentiation has compressed.
What remains as a sustainable advantage is a deep understanding of a user segment that major players have no incentive or structure to serve well. That understanding can’t be bought with capital; it’s accumulated through field iterations. The only way to accumulate it is to be physically close to the problem, with a product light enough to modify quickly and a monetization model that’s tested from day one.
The companies that will define the next stage of the industry won’t be those that build the most powerful infrastructure. They will be those that reach the ignored markets first with a model that works on a local scale before attempting to scale globally. Geography is a symptom. The underlying principle is that **sustained business growth happens when one abandons the illusion of the perfect plan and embraces constant validation with the customer that no one is yet watching.
