A look at the US open-weight AI model ecosystem, as VCs question whether Arcee AI, Reflection AI, Poolside, and Thinking Machines Lab can generate real revenue
Silicon Valley startups are setting up open models, with some operating on shoestring budgets because of limited interest from VCs
Wall Street Journal
Context & Ripple Effects
Open-weight models have been gaining strategic importance as a shared foundation for the AI ecosystem, while US startups have also adopted cheaper, customizable Chinese alternatives. That makes the funding scrutiny around domestic open-model builders a test of whether the US can support businesses around the model layer itself.
Arcee, Reflection AI, Poolside, and similar startups face a harder financing environment: investors are questioning how open weights translate into durable revenue, while some teams must operate with leaner budgets.
The companies must make a clearer commercial case beyond releasing models, such as paid services, customization, hosting, or other complementary offerings.
Second-order effects
Funding pressure favors open-model companies that can tie distribution and support to a customer need, rather than compete primarily on training scale against better-capitalized labs.
Enterprise buyers gain leverage as cheaper and customizable model options broaden their ability to mix providers; that cost focus has already contributed to pricing pressure on frontier labs.
Third-order effects
If capital continues to concentrate in closed frontier labs while open-weight builders struggle to monetize, the open ecosystem may shift toward a complement economy—tools, deployment, and services built around models—rather than stand-alone model vendors.
The outcome matters strategically: open weights are increasingly framed as ecosystem infrastructure, so weak commercial support for US builders could leave that layer more dependent on externally developed models.
The trend: AI is moving from a race to release capable models toward a contest over who can capture recurring value from the infrastructure, services, and distribution around them.
The Race to Build an American Alternative to Cheap AI From China—Silicon Valley startups are setting up open models, with some operating on shoestring budgets @KateClarkTweets @samschech https://www.wsj.com/... https://www.wsj.com/...
“There is a vast, vast degree of want for an American company producing the most-capable open-source artificial intelligence,” said Jason Warner, co-founder of Poolside, which in July released a new version of its open-weight model Laguna S 2.1. https://www.wsj.com/...
“Every tier-one VC pretty much said no,” said Mark McQuade, CEO of Arcee, a little-known Silicon Valley startup that in late 2025 bet much of its remaining cash on building the strongest open-weight AI model it could. https://www.wsj.com/...
The new batch of open-weight companies, which includes Arcee, Reflection AI and Poolside, are betting on an increase in demand for capable open models that offer the efficiency of China's without the geopolitical complications. https://www.wsj.com/...
One problem some open-weight American AI pioneers are facing is funding. Despite growing calls for the U.S. to develop stronger American open-weight AI to compete with China, investors are reluctant to back some of those startups. https://www.wsj.com/...
@benthompson has the clearest take here, worth reading to understand the underlying dynamics. It's hard for US open models to be better because they must follow big lab ToS. This is a structural disadvantage. https://www.wsj.com/... [image]
There is growing concern in Silicon Valley and Washington that China's open-weight models are close to matching top U.S. counterparts and could threaten the profitability of AI companies. A race has begun among a few homegrown players to compete. https://www.wsj.com/...