Tech Coalition Pushes US to Back Open AI Models as Frontier Costs Keep Climbing
Updated
Updated · Spyglass.org · Jul 27
Tech Coalition Pushes US to Back Open AI Models as Frontier Costs Keep Climbing
3 articles · Updated · Spyglass.org · Jul 27
Summary
A broad tech coalition is coalescing behind downloadable “open weight” AI models, arguing Washington should promote them as a pillar of US AI leadership rather than restrict them.
Rising frontier-model costs are driving that push: training and serving top-tier systems is getting harder to finance, with enterprises and startups alike squeezed by heavy token and capital spending.
The coalition has widened quickly since the July 24 letter launch, with signatures reaching 50 and late additions including OpenAI and Google, while Amazon and Anthropic remain notable holdouts.
Anthropic and parts of the US government are the clearest skeptics, citing security, geopolitical leverage and the risks of broadly available models even as rivals say openness is needed to counter China.
China’s progress has sharpened the debate, with Moonshot’s Kimi K3 cited as a “good enough” open-weight model that underscores both the competitive pressure and the unresolved safety and provenance questions.