Open-weight model developer Arcee AI reaches $1B-plus valuation with new funding
SiliconANGLE Mike Wheatley ● Covered by 2 sources
Arcee AI says its new funding pushed it past a $1B valuation. The bet is on open-weight models, not the locked-down kind most U.S. firms sell.
Based on reporting by SiliconANGLE, Mike Wheatley — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Arcee AI says it has raised a new round of funding that values the company at more than $1 billion. The amount wasn’t disclosed, though one source told Fortune the deal was worth at least $150 million. Vista Equity Partners led the Series B, with Cambium Capital and Emergence Capital also at the front of the line. A10 Ventures, Hitachi, Microsoft’s M12, P7, Wipro and IAG joined in.
The pitch is straightforward: Arcee wants to be one of the U.S.’s top open-weight model builders. In that model, the company releases the core numerical parameters, so people can download the system, modify it and run it on their own infrastructure. Chief executive Mark McQuade says that matters because American companies should have more control over the AI they rely on. He framed it as a choice between frontier capability and control, and said organizations should not have to give up one to get the other.
That puts Arcee in an unusual spot. Most U.S. AI developers have gone proprietary, while many Chinese developers have leaned hard into open weights. McQuade says he wants Arcee to catch up with China’s leading AI builders, and he was blunt about the U.S. side: far ahead in closed-source, he said, but it “kind of dropped the ball” on open-source.
The company’s Trinity family is the proof point. Arcee says all of those models were released in the last year, and the biggest one, Trinity Large, has 400 billion parameters with 13 billion active for each token. The part that should make other AI labs wince is the cost: Arcee says it spent about $20 million on those models last year, covering compute, data, operations, and salaries. McQuade said the company had about $30 million when it decided to build its own models, and that almost 70% of its total capital went into Trinity.
The fresh money is going to more of the same, just bigger. Arcee plans to train a new generation of Trinity models, expand work with the U.S. Department of Energy and 17 national laboratories, and keep building tools for companies that want to customize, test and deploy open-weight models. It’s a neat little rebuke to the idea that AI progress has to come with maximum secrecy and maximum spending.
My take — AI-written commentary, not fact-checked reporting
This is the kind of funding round that says open-weight AI is no longer a hobby for idealists and grad students. The real story is the budget discipline: $20 million to build something that can sit in the same conversation as much pricier closed systems is the sort of number that makes boardrooms squirm. Silicon Valley loves calling everything strategic; this one actually is.
Read more about this at: SiliconANGLE