AI Startup Lindy Ditches Anthropic for DeepSeek, Saving Millions as Cost Pressures Reshape Industry

AI startup Lindy has fully replaced Anthropic's Claude with DeepSeek, a Chinese model hosted on US infrastructure, saving millions of dollars. CEO Flo Crivello told CNBC the cost curve "crashed to the ground" after the switch, calling the previous spending "unsustainable."

For Lindy, a 25-person company, AI costs had surpassed personnel costs. Crivello said the change was "a matter of survival for the business," though he would return to Anthropic if prices dropped. The move comes as OpenAI CEO Sam Altman recently noted that AI costs have become a "huge issue" for companies adopting agentic systems, which burn through tokens rapidly.

Cost Pressures Drive Model Switching

Lindy's decision is not isolated. A Snowflake analysis found that Chinese models like GLM-5.2 are competitive with Anthropic's Opus 4.7 on price-performance, even if they don't match it on every benchmark. The trend is pressuring major AI labs like Anthropic, which is reportedly eyeing an IPO, and OpenAI, which has already delayed its own public offering.

• Lindy saved millions by switching from Claude to DeepSeek, hosted on US soil.
• AI costs for the startup exceeded personnel costs before the change.
• Snowflake's CEO found Chinese models competitive on price-performance for many tasks.
• OpenAI's Sam Altman warned that agentic systems are driving up token consumption and costs.
• Anthropic's IPO timing may be affected as customers seek cheaper alternatives.

European Ambitions and Political Spending

Meanwhile, Europe is pushing for its own AI capabilities, with WIRED reporting that the continent is "fed up" and wants to build a top-tier model, leveraging political dynamics around Donald Trump. Berlin-based Peec AI is targeting a $200 million valuation in a new fundraise, signaling continued investment in European AI startups despite the dominance of US and Chinese players.

The AI industry is also pouring millions into US elections, according to Blood in the Machine, as model capabilities now have real political consequences. TechCrunch noted that the competition is no longer just between Anthropic and OpenAI, but about how society collectively deals with the implications of advanced AI.

What comes next: As cost pressures mount, more startups may follow Lindy's lead, switching to cheaper models. Anthropic and OpenAI will need to balance pricing with performance to retain customers, while European and political investments reshape the broader AI ecosystem.