The Wall Street Journal has highlighted a growing trend that is reshaping Silicon Valley: billion dollar artificial intelligence startups raising enormous sums of money despite having no product, no revenue, and often no clear plan to make money. Investors are pouring capital into what are increasingly called AI “neolabs,” research-first companies that prioritize long-term breakthroughs over commercial results. To supporters, this is how transformative technology is born. To skeptics, it looks more and more like a speculative bubble.
Unlike traditional startups, these AI labs are not selling software, services, or subscriptions. Many do not even have a public-facing tool. Their pitch is centered on research talent, novel training methods, and the belief that today’s dominant AI models may have already reached their limits.
Investors are betting that a small number of these labs could become the next OpenAI. The problem is that most of them may never produce anything that justifies their valuations.
Six High-Profile AI Labs With No Products or Revenue
Several companies stand out as examples of this trend, all raising hundreds of millions or even billions of dollars without shipping a product.
Flapping Airplanes
Founded by Stanford Ph.D. student Ben Spector, Flapping Airplanes is focused on long-term AI research inspired by biological systems. The company raised $180 million at a $1.5 billion valuation despite having just 11 employees and no product. Investors include GV, Sequoia Capital, Index Ventures, and Menlo Ventures.
Humans&
Humans& raised $480 million at a $4.48 billion valuation to build AI systems designed to help people collaborate. The company has not released a public tool or generated revenue and remains in the research phase.
Reflection AI
Reflection AI raised more than $2 billion at an $8 billion valuation to build an open-source frontier AI model. Like many neolabs, it is focused on model development rather than commercialization.
Periodic Labs
Launched with $300 million in funding, Periodic Labs aims to automate scientific research using AI. The company has no product on the market and no reported revenue.
Thinking Machines Lab
Co-founded by former OpenAI executive Mira Murati, Thinking Machines Lab has sought funding that could value it as high as $50 billion. The lab has not announced a product and has already lost several founding researchers to OpenAI and Meta, raising questions about talent retention.
Safe Superintelligence
Founded by former OpenAI chief scientist Ilya Sutskever, Safe Superintelligence has raised more than $3 billion at a $32 billion valuation. Sutskever has been explicit that the company may never produce a product or revenue, stating that the goal is to research safe superintelligence rather than commercial AI tools. Investors reportedly include Google, Nvidia, and Andreessen Horowitz.
Why Investors Are Funding Research Instead of Products
Investors say the traditional rules no longer apply. AI research is extremely expensive, and the belief is that whoever cracks the next major breakthrough could control a technology more important than any single product.
U.S. AI startups raised a record $222 billion last year, according to PitchBook. Venture capital firms are increasingly underwriting talent, conviction, and time rather than revenue projections. Some investors openly admit they are backing people, not businesses.
Expectations Versus Reality
Supporters argue that history favors this approach. OpenAI began as a research lab. DeepMind was once dismissed for lacking a business model before being acquired by Google.
Skeptics point out that for every OpenAI, there are dozens of well-funded labs that quietly fail. Ashu Garg of Foundation Capital warned that most neolabs will not cross the technical gap required to matter, ending up with results that are only incrementally better than existing models.
How Many Are Likely to Fail
While exact numbers are unknown, both investors and researchers acknowledge that the majority of these labs are unlikely to succeed. With dozens of neolabs and only room for a handful of winners, failure rates are expected to be extremely high. The risk is amplified by intense competition for talent and the willingness of Big Tech firms to pay hundreds of millions to hire researchers away.
Backers argue this is a necessary phase in AI development. They say meaningful breakthroughs require patience, freedom from short-term pressure, and massive capital. Some believe that refusing to rush products actually builds long-term trust and credibility.
Critics see troubling parallels to past bubbles, including the dotcom era and the 2021 startup boom. Valuations are disconnected from commercial reality, and there is no reliable way to measure progress without revenue or products. If market conditions tighten or investor sentiment shifts, these companies could face sudden and severe corrections.
A Bubble in the Making
The surge of billion dollar AI startups with no products, no revenue, and no clear path to monetization may represent a bold bet on the future. It may also be the clearest sign yet that AI investing has entered bubble territory. History suggests that when valuations float free of reality, gravity eventually returns.
FAM Editor: Anything that doesn’t get bought is irrelevant.
