AI Startup Valuations Climb as VCs Split Rounds Into Two Prices

A seed round with more than one price

Right now, a startup in the AI venture market can be worth two very different amounts in the same funding round. Bloomberg reported on October 22 that venture firms are competing to back Emulate, a company started by former Google DeepMind researchers. Emulate is about three months old and has no product and no website, yet investor demand has pushed its valuation into the billions.11 The competition is strong enough that investors are buying into the same round at prices that rise quickly from one closing to the next. Bloomberg calls this a tranched, or two-step, round.1114

The Emulate deal is new, but the structure behind it has been building for most of 2026. Investors and founders are still arguing over whether it is a reasonable way to price scarce talent or a way to make headline valuations look bigger than they are. Our view, explained below, is that the second description is closer to the truth, though the reasons behind it are more interesting than outright deception.

How a two-step round works

In a standard venture round, every investor in a seed or Series A pays the same price per share, so the company has one valuation.3 A tranched round breaks that rule. The company closes a first slice at a lower price and leaves the round open. Later investors then buy into the same legal round at a much higher price, sometimes after the company makes a key hire or ships a small update, and sometimes just because more investors want in.3 The valuation that gets announced is usually the higher price.5

What goes unannounced is the blended valuation, meaning a weighted average of the tranche prices based on how much equity each tranche bought. Amplify partner Sarah Catanzaro told Forbes that this blended figure is usually the better measure of what the company is worth, and that it is rarely made public.

The numbers behind the headlines

The clearest documented example is David Silver's Ineffable Intelligence. Its $1.1 billion seed round was announced at a $5.1 billion valuation and promoted as the largest seed round in European history.2 Company filings reviewed by Forbes show the money came in two pieces. First, Ineffable raised $11 million from Sequoia and others at a pre-money valuation of about $55 million. Roughly a month later it raised $1.1 billion at a $4 billion pre-money valuation, with Lightspeed, Index Ventures and DST Global buying in, and Sequoia buying again at the higher price.2 That is a price increase of more than 70 times for the same company in a matter of weeks.15 A person familiar with the deal told Forbes that most of Sequoia's money went in at the higher price.5

The structure is not limited to research labs without revenue. AI infrastructure company Baseten raised $1.5 billion in two tranches, one priced at an $11 billion valuation and the other at $13 billion.2 Startup Fortune's account adds that Baseten was valued at $5 billion six months earlier, and that its annualized revenue run rate grew from about $200 million to $600 million over that time. Baseten can at least point to real business growth, which Emulate cannot.1

Earlier in the year, application-layer startups set the pattern. TechCrunch reported in March that Redpoint led a Series A for synthetic-customer research startup Aaru. Redpoint put most of its check in at a $450 million valuation and a smaller share at $1 billion, and other investors came in at the $1 billion price.4 Serval, an AI IT help desk company, announced that its $75 million Series B valued it at $1 billion, while Sequoia's lowest entry price was reportedly at a $400 million valuation.4 PitchBook found that Serval's round combined two classes of preferred shares priced at different valuations, and listed Lovable's Series A as another round built this way.10

Where the reporting on Emulate differs

The Emulate reporting is less consistent than the Ineffable reporting. A September report from the Financial Times, relayed by Tech Funding News, said Emulate was in talks to raise up to $700 million at a post-money valuation of about $3.7 billion, with Index and Lightspeed expected to co-lead.5 One Kenyan outlet presents those figures as the round's final tranches, putting the valuation at roughly $3.7 billion to $4 billion.3 A summary of Bloomberg's story, however, says the report itself does not give the round size, an exact valuation, or the participating firms.14 Readers should treat the September figures as reported talks, not a confirmed close, and the final price could be higher.

The same Kenyan outlet also says investors are giving up board observer rights and standard anti-dilution protections to secure allocations.3 None of the other coverage repeats that claim, so it is best read as plausible but unconfirmed.

Why investors accept two prices

The reporting mostly agrees on the mechanics and disagrees on motive. TechCrunch's account focuses on efficiency. Constant fundraising distracts founders, so lead investors combine what would have been two consecutive rounds into one.4 Primary Ventures general partner Jason Shuman described the structure as a sign of intense competition among VCs, and noted that a very large headline number can discourage rivals from funding the second- and third-place competitors.4

Some investors defend it more strongly. SignalFire venture partner Josh Constine told PitchBook that tranches reward investors who commit early and take more risk, and that the best founders now have enough leverage to set their own prices.10 When Mercor's Brendan Foody publicly accused Sequoia of using two-tranche rounds to inflate valuations, Sequoia partner Shaun Maguire said the firm only splits its check when rival investors are willing to pay more than Sequoia would for a competitive deal.10 Foody's criticism was sharp. He said he had seen about half a dozen such rounds in six months, and that participants act as if they only invested at the higher price.

Critics come back to the gap between headlines and economics. Foundation Capital's Jaya Gupta told Forbes that in a market driven by sentiment, a billion-dollar headline is worth more than an accurate one.5 FPV Ventures' Wesley Chan described the practice to TechCrunch as a symptom of a bubble.4 Bryan Yeo, GIC's group chief investment officer, has warned about hype in early-stage venture, where startups with an AI label are valued at large multiples of very small revenue.1

Who carries the risk

Looking across the reporting, the clearest conclusion is about who bears the risk, not about whether the practice is honest. The first tranche buys ownership at a low price. The second tranche produces the headline number.5 Forbes reported that a first-tranche investor such as Sequoia at Ineffable could show a paper markup of up to 70 times before its own investment had fully settled.5 Later investors pay the highest price, and their returns depend on the company eventually being worth at least that much.1

Corporate strategic investors complicate the picture. According to Forbes, the lead investor in the first tranche usually organizes the whole round and signals to Nvidia, Google or Microsoft that the startup is worth backing. Those corporate investors tend to care less about price, because their checks often turn into chip contracts and cloud deals. The headline valuation is therefore partly a figure for marketing and partly a cost of doing business for buyers who expect to get much of their money back as revenue.

Founders also face risk. Shuman warned that a company with a lower blended valuation is still expected to raise its next round above the headline price, or else face a damaging down round.4 A big headline helps with hiring now and raises the bar for the next financing.4

Why it matters for the wider market

The amounts are large. Menlo Ventures partner Deedy Das counted more than 63 so-called neolabs, together valued at over $300 billion, which had raised about $48 billion. Forbes calculated that this equals 16% of the roughly $283 billion invested in startups other than OpenAI and Anthropic over the past year. Public money is involved too. Ineffable received backing from the UK's new Sovereign AI fund and the British Business Bank, both funded by taxpayers.

It also fits a broader pattern in which AI valuations rise faster than normal funding cycles. In late 2025, Fortune reported that more than a dozen AI startups had raised two or more rounds within a year at rising valuations. Tranching essentially compresses that sequence into a single round, which is why Startup Fortune argues the structure has moved from a fix for difficult deals to a standard tool for the most sought-after labs.1

Our conclusion is that tranched rounds are not fraud, but they are opaque. They reward early conviction, satisfy strategic investors, and give founders a strong recruiting pitch. They also make headline valuations a much weaker indicator of what the market will actually pay for these companies. Until blended valuations are disclosed routinely, investors and readers should take each new billion-dollar seed round as the price the last buyer paid, not as a measure of the whole company's value. The real test will come when one of these companies tries to raise above its headline price and cannot.