Arcee x Emergence
Two years ago, we backed Arcee on the belief that enterprises needed AI built around their own data, goals and economics without sacrificing control. Since then, that question of control has moved from the enterprise to the frontier. Open weight models have become a matter of American AI competitiveness, as Chinese labs move quickly to put capable open models in developers’ hands while much of our domestic frontier remains closed. Arcee is building another path in the U.S. Now, at a $1B+ valuation, they’re scaling it.
Why we're doubling down: Emergence first backed Arcee when the company was helping enterprises build smaller, specialized models on their own terms. What’s happened since has expanded the scope of that original bet. Arcee built its entire 2025 model lineup, including its 400-billion-parameter Trinity Large model, for approximately $20 million. This serves as evidence that frontier-scale open-weight AI can be built domestically with a radically different level of capital efficiency. Arcee is now working with the U.S. Department of Energy and its 17 national laboratories on an open model for scientific computing. This extends Arcee’s models beyond the enterprise and productivity tasks.
What we're watching: Whether the U.S. can build an ecosystem that competes at the frontier without adopting frontier-lab economics. We recently joined leaders across the industry in arguing that open weights are critical to American AI leadership because they expand access, competition and ownership. Arcee is showing what that argument looks like in practice.
Arcee started by proving that enterprises shouldn't have to choose between performance and control. Now they're proving the U.S. shouldn't have to choose between openness and frontier capability.
— Joe Floyd, General Partner
Read why we backed Arcee from the start here.
