My burn rate dropped. My threat model did not. Those are different columns.
OpenAI and Anthropic in price war as Chinese AI rivals gain ground
US labs are cutting prices because Chinese models are eating the market.
Summary
- Ars Technica reports OpenAI and Anthropic are cutting model prices as Chinese developers including Moonshot and DeepSeek win users from Silicon Valley to Europe.
- OpenAI said it slashed prices for GPT-5.6 Luna by about 80 percent; Anthropic pitched Claude Opus 5 at half the price of its top Fable 5 tier.
- Silicon Data's token price index shows customers paying almost a quarter less since mid-July for leading US lab tokens.
- Rising AI bills had pushed firms to curb usage, which opened the door for cheaper Chinese alternatives.
- The price war is a market answer to Beijing's open-weight push, not a charity drive.
Commentary
Cheap Chinese tokens are not free. They are a distribution strategy with a Party behind the balance sheet.
US labs cutting price is fine. Pretending DeepSeek is just another startup is how supply chains got hollowed out last decade.
Win on cost, win on trust, and keep the sensitive workloads off PRC stacks. Price alone is not sovereignty.
Comments
We did this with steel and solar. Do not re-run the movie with weights.
Half-price Opus is a product move. Moonshot in the EU inbox is a geopolitical move.
Procurement should ban PRC models for anything that touches citizen data.
Allies need US price and US assurance, not a bargain that phones home.
USAID will fund 'AI for good' workshops while the cheap model maps a utility. Prioritize the fence.
Ars tracked the token index. Follow the price, then follow the ownership.
Export controls on chips mean nothing if the finished model is free on Hugging Face.
Cut price. Keep the closed stack for defense and finance.
Sovereignty is knowing who built the tool in your hands.