Why OpenAI May Never Generate ROI 

By Eugene Malobrodsky

As AI companies with no product raise billions at monster valuations, OpenAI’s Sam Altman himself has called this moment an AI “bubble.” But the companies that spearheaded the AI wave are themselves at great risk of not returning any profits to investors.

At its current valuation, there’s a real risk OpenAI is not going to return money to later investors. Its cash burn by 2029 will reportedly reach a whopping $115 billion (despite a projected ARR of $20 billion for 2025). Early backers have already locked in the windfall, but the current expectation of future profit that motivates investors may be misguided.

We’re in a race to the bottom

Foundational AI models — aka the LLMs used by most AI tools — are now in a phase of commoditization. When GPT first burst into our lives, it was a true technological leap forward, presenting businesses with transformational, profitable applications.

But since then, each new advance has been less and less impactful. OpenAI’s GPT-5 rollout was deeply disappointing to users, who demanded a return to the previous version.

With the major foundational AI companies — Google, xAI, Anthropic, Meta, OpenAI — working on comparatively marginal improvements to their base models, the real sell now is price point. Switch costs between the major providers are extremely low for a lot of AI users, and OpenAI is looking to acquire and retain customers by aggressively driving prices down with new models.

And yet, their operational costs are going to remain sky-high for the foreseeable future, as their huge infrastructure demands continue to grow.

The real winners

In fact, the clearest winners of the current AI era are the Googles, Microsofts and Nvidias providing the picks and shovels. OpenAI and others have been financial boons for these industry giants, with OpenAI recently committing to buy $300 billion in Oracle computing power over the next five years.

Unless infrastructure costs or compute requirements somehow plummet — and I don’t see that happening in the near future, quantum breakthroughs or otherwise — the billions of realized profits are going into the pockets of the providers of GPUs, energy and other resources, not the foundation model providers.

The other real competition for profits is happening among companies making AI tools that will be practically useful to people’s lives and workplaces, trained on industry-specific data from the companies they serve.

Consumers and business users will logically move from generalized AI platforms to more specialized tools that target specific needs. But in this field, OpenAI is competing with an infinite amount of startups and tech giants. Their early advantage as a first mover is far less relevant than having an exceptional team working on UX, or reinforcement learning, or rather than having a vast pool of historic user data.

And while GPT and other for-profit LLMs power many third-party applications, another existential threat here is free, open source models like DeepSeek offering a viable alternative.

OpenAI would need to break into this application layer with force in order to own a significant portion of the profits that come with more specialized AI products. But until now, they’ve not proven that they have a clear edge compared to competitors in major arenas (OpenAI’s Codex vs. GitHub’s Copilot, DALL-E vs. Midjourney).

Future of AI investment

So why are investors sinking hundreds of millions into OpenAI and others?

It’s partly ego. That, and they hope their “horse” ultimately wins all the pie. I don’t think that a single player is going to come out on top. The separation between different foundational AI providers is becoming too insignificant to offer investors a clear winning scenario.

Even in the likelihood that the foundational AI market consolidates around a few players, these still won’t be returning at the multiples that will generate ROI to the later-stage investors.

Users will be spending their cash on AI applications that significantly impact their life. I can see a lot of the profits coming from the AI companies that are hyperfocusing on a specific industry niche and transforming its outcomes — such as new drug discovery, vertical enterprise applications, healthcare and even consumer applications.

Eugene Malobrodsky is managing partner at One Way Ventures, a VC firm funding exceptional immigrant founders. He is also the creator of the One Way Summit, a one-day gathering to celebrate the immigrants building the most disruptive companies on the planet.

Illustration: Dom Guzman