Fable 5 tears open a rift: When the engineer's AI bill exceeds his own salary

On June 9, Anthropic released the strongest public model ever, Claude Fable 5. Benchmark tests showed a complete superiority over the previous generation, with autonomous coding capabilities far ahead of its peers. According to the pattern of the past two years, each time a powerful new model is released, the market would see a rally—investors believe that AI continues to progress and the future looks brighter. On the day of the launch, this pattern held true, with the overall market remaining stable and faith in AI-driven trading enduring.

However, Fable 5 was only available online for three days. On June 12, the US Department of Commerce, citing export controls, required Anthropic to ban access for any foreign nationals. Since Anthropic was unable to screen users in real time based on nationality for technical reasons, they had to suspend access for everyone at once, resulting in an 18-day shutdown.

During these 18 days, market sentiment shifted. By late June, chip stocks came under continuous pressure due to concerns over rising “AI infrastructure costs,” and investors began taking profits from semiconductor sectors that had gained more than 80% in the first half of the year. By the time Fable 5 was re-listed on July 1, it was greeted not with applause, but with a cold shoulder: the Nasdaq fell 0.66% that day, with chip stocks leading the decline—Micron dropped by 10%, Broadcom by 2%, and Nvidia by about 1%. The strongest model in history returned online, but the market’s response was continued selling.

This contrast is worth pondering. For the same model, the market was still willing to “buy” the notion of better AI on the initial release; but just three weeks later, with the relaunch, investors’ attitude shifted from faith to calculation. The market is no longer unconditionally cheering for advances in capability and has begun to ask a sharper question: how much does stronger AI really cost?