AI can change the world—if we change who it’s built for

In the first quarter of 2026, investors deployed 300 billion into 6,000 startups, up 150 from previous years. Only a single-digit sliver of that capital is directed toward AI applications focused on solving social and environmental challenges. That gap is not an accident. It reinforces who AI is being built for, and who it’s leaving out. And that failure is costing us some of the best and most-needed solutions being built outside of Menlo Park. Here is what I mean. When Temie Giwa-Tubosun started LifeBank in Nigeria, hospitals were running out of blood. She couldn’t build for ideal conditions because ideal conditions didn’t exist. Every assumption had to survive contact with reality: inconsistent power, fragmented logistics, traffic jams of mythical proportions, and hospitals that couldn’t afford to wait. The result was an AI-enabled delivery network coordinating supply and demand across unreliable infrastructure, using a mix of routing software, local logistics, and last-mile transport—to move blood, oxygen, and medical supplies where they were needed most. They are now serving 3,000 hospitals with 24/7 service and delivery in under 45 minutes. Rather than slowing her down, the infrastructure clarified what the solution had to be. PROXIMITY IS EXPERTISE That is the mechanism that funders so often fail to recognize. Proximity to a problem is a form of expertise—one that compresses feedback loops, eliminates false assumptions, and produces solutions truly suited to the environments they’re meant for. It should be a competitive advantage. Instead, it often functions as a disqualifier. These founders don’t have the right ZIP code, the right credentials, or the networks their counterparts in San Francisco navigate by default. So a category of high-impact, technically credible, and commercially viable work gets systematically undercapitalized. The losses are not abstract. Proximity matters not only in physical systems like health logistics, but in preserving knowledge systems as well. Consider what FLAIR—the First Languages AI Reality initiative (formerly known as International Wakashan AI Consortium)—is racing to prevent. More than half of the world’s languages are projected to become extinct or seriously endangered by 2100. Each one that disappears takes with it ecological knowledge, historical memory, and ways of understanding the world that cannot be reconstructed. FLAIR is using AI and immersive technology to help Indigenous communities revitalize their languages for living practice. The technical challenge is significant. The irreversibility of failure is total. The loss is one that all of humanity would endure. Scale looks different at this edge of the market, too. Amini built a data platform for smallholder farmers across Africa, delivering hyper-accurate agricultural insights via SMS to people with no smartphones and limited connectivity. Over a million people now have access to financial products, insurance, and climate information that were previously out of reach—built on a constraint (no reliable internet) that most AI developers treat as a reason not to build at all. By treating constraint as a design input, Amini is creating a new market, serving populations long excluded from digital infrastructure. Together, these examples point to a consistent pattern: When solutions are built in proximity to the problem, they are more adaptive, more efficient, and better suited to real-world complexity. THE CONNECTIVE BRIDGE The talent, the technology, and the market exist. What is missing is the connective bridge between these founders and the capital and networks that would let them scale. Venture capital’s model does not map well to solutions built in fragmented environments and where return is measured in people served per dollar. That gap is where impact investors, development finance institutions, and philanthropies have an urgent role to play. They can combine grant funding and investment capital to absorb early risk and unlock larger-scale financing, allowing such organizations to grow. Pairing philanthropic risk capital with private investment can give those organizations a longer lead time, reduce early risk, and make these models investable at scale. While the model exists, we need to get it deployed at the scale the moment demands. These founders are already building with a fraction of the resources, in conditions designed to stop them. It’s time for capital to catch up. Hala Hanna is executive director of MIT Solve.