A company that did not exist at the start of the year is trying to raise money at a valuation of about $50 billion. Discovery Loop, an artificial intelligence startup founded in California by a group of prominent Google veterans, is in talks with investors for a new round at that figure, according to people familiar with the matter.
The number is a sharp jump even by the standards of a market accustomed to rapid re-ratings. As recently as last month, the company was seeking to raise about $1 billion at a valuation of roughly $10 billion, the people said. The terms could still change, and there is no guarantee the round closes at the higher figure.
Discovery Loop’s pitch rests on its founders. The group includes Jeff Dean, one of Google’s most senior engineers and a central figure in the company’s machine-learning efforts, along with other notable AI and engineering figures from Google, according to the people familiar with the matter. Their track record is the asset investors are being asked to price.
Dean has spent more than two decades at Google, where he helped build its earliest search and storage infrastructure, co-designed the TensorFlow machine-learning framework and played a leading role in the company’s AI strategy. People who have worked with him describe a figure whose technical judgment carries weight well beyond Google’s walls.
The company is organized as a public benefit corporation, a structure that allows it to pursue goals beyond shareholder returns. The same form is used by Anthropic and OpenAI, and it has become a signal in the industry that a lab takes safety and long-term concerns seriously enough to write them into its charter.
What Discovery Loop actually does is less precise. The company is focused on breakthroughs in machine learning, science and engineering, according to people familiar with its plans. The description is broad enough to cover model development, scientific applications and tools, and the company has offered limited public detail about its first products.
The speed of the re-rating is the notable part. A jump from a $10 billion target to $50 billion in a matter of weeks implies investors are competing for access rather than scrutinizing revenue, which the company has not demonstrated at scale. Analysts said the pattern echoes the early rounds of the largest labs, when access mattered more than metrics.
The fundraising climate has cooled in recent weeks. A public debate over whether AI development should slow has rattled valuations across the sector, and investors have grown more selective, favoring companies that can show revenue. Discovery Loop is asking the market to set those concerns aside and price its founders’ pedigree instead.
The company’s Google roots cut both ways. The founders’ credibility opens doors that a new entrant would otherwise not reach, and it allows Discovery Loop to command terms normally reserved for established labs. But the same history raises questions about how much of the company’s technology is genuinely new and how much is a repackaging of work its founders led elsewhere.
The round arrives as the largest AI companies are pulling further ahead. Anthropic has told investors it expects to go public at a valuation near $2 trillion, and OpenAI has cemented its position as the sector’s best-known name. Against that backdrop, a $50 billion valuation for a new entrant is modest, which is part of the argument its backers are making.
The company is also competing for the same resource as everyone else in the field: people. Researchers and engineers with frontier experience are scarce, and a new lab led by recognized names has an advantage in recruiting. The capital being raised is as much for salaries and compute as it is for any single project, according to people familiar with the plans.
Compute is the other line item. Training frontier models requires access to large quantities of accelerators, and the largest labs have locked up much of the near-term supply through multi-year contracts. A new company must either buy its way into that market or negotiate capacity, and a $50 billion valuation would give it the currency to do both.
The public benefit corporation form carries obligations that a conventional Delaware company does not. Directors must balance the stated mission against shareholder returns, which can complicate later decisions about pricing, partnerships or an eventual sale. Some investors accept that friction; others have privately questioned whether it fits a company seeking capital at this scale.
Little is known publicly about what Discovery Loop will build first. People familiar with the company said its work spans areas where machine learning meets scientific discovery, including applications that could bear fruit only over years. That long horizon is easier to fund with a public benefit charter than with the quarterly demands of a public market.
People familiar with the talks cautioned that the figure is a target, not a commitment. Venture rounds at this scale are negotiated over weeks, and terms can shift with market conditions. The AI sector’s recent volatility makes the outcome harder to predict than it would have been even a month ago.
For the broader market, the round is a test. If Discovery Loop raises at $50 billion, it will confirm that capital is still chasing founding teams regardless of the slowdown debate. If the round reprices lower, it will show that the market’s appetite for speed and pedigree has limits after all.
