OpenAI's Billions in Losses Are the Price of Building the Future, Not a Warning Sign
🕒 Published on AI Momentum: June 30, 2026 · 03:40
Internal documents reportedly show OpenAI lost $5 billion in 2024, with projected losses of $38 billion for 2025, driven mainly by the soaring cost of training AI models. The numbers look alarming—but they tell the story of an infrastructure race, not a failing business.
According to internal documents, OpenAI lost roughly $5 billion in 2024 and is projected to lose as much as $38 billion in 2025. The company itself attributes the bulk of this red ink to the enormous cost of training ever-larger AI models—the compute, the chips, the energy, and the talent required to push the frontier.
It's worth being precise here: these are figures reported from leaked internal documents, not audited public accounts, and "losses" of this kind are not the same as a company running out of money. They reflect aggressive, front-loaded investment in capacity. Building frontier models today resembles laying railroads or fiber-optic networks in earlier eras—the capital is spent years before the returns arrive, and the early balance sheets look brutal precisely because the ambition is so large.
In the short term, the impact is real and uncomfortable. Numbers like these intensify pressure to monetize fast, raise questions about how sustainable the current pace of model training really is, and concentrate the AI race in the hands of whoever can absorb the deepest losses the longest. That dynamic deserves scrutiny, not applause—it could entrench a handful of players and make the field more fragile if funding tightens.
Our reading: this is what the expensive, messy transition phase of a transformative technology actually looks like. The cost of training is falling per unit of capability even as total spending rises, and the infrastructure being paid for now is the foundation for tools that, over the coming decade, can accelerate drug discovery, extend healthy lifespans, and lower the cost of expertise toward zero. The losses are a bet on that future. The honest question isn't whether the spending is huge—it plainly is—but whether it converts into durable value and broadly shared abundance, rather than just a moat for the few. That's the metric to watch, and it's why measured optimism, not euphoria, is the right posture.