banks are using chinese ai models as leverage against anthropic's pricing

@sqs, who built sourcegraph and now runs coding agent @AmpCode, on what's actually happening inside enterprise ai procurement:

big banks have quietly approved chinese open source models – not because they prefer them, but because they needed a credible alternative. when anthropic comes to negotiate a large commit, having deepseek or qwen approved gives you a seat at the table.

but with meta open sourcing muse spark, that dynamic shifts. an american open model at a lower price point means banks can drop the chinese approvals over time. no geopolitical risk, no disclosure headaches.

his projection: 80% of enterprise spend stays on anthropic and openai. 20% moves to american open models – and that share grows.

what makes this real: switching is fast. this isn't infrastructure migration. it's a config change.

meta's open source play isn't about beating anthropic on benchmarks. it's about making sure anthropic can't price like a monopoly.

full segment on the video via link. source: @theinformation