Tencent (腾讯) "Lobster" Crash Flags Idle Compute after RMB 281M Traffic Push and Emergency 10× Expansion

Outage and immediate fallout

Tencent's (腾讯) CodeBuddy team apologised after WorkBuddy — nicknamed the "Tencent lobster" (小龙虾) — triggered a service overload shortly after its domestic public launch. The team said the domestic rollout drove user traffic more than tenfold, producing an immediate outage and prompting a temporary 10× emergency compute expansion; affected users were reportedly given 5,000 credits each as compensation. It has been reported that many testers opened dozens of parallel AI-agent windows — generating posters, reports and even letting agents "chat" with one another — behaviour that blew past the team's traffic estimates.

A usage paradox, not just an engineering bug

This is more than a capacity planning failure. It lays bare a behavioural-economic problem: when inference costs approach zero, human behaviour changes. It has been reported that Tencent also spent roughly RMB 281 million (2.81亿元) on traffic and marketing for its DeepSeek/元宝 push to drive mass adoption, swelling MAU quickly. The result? Massive compute consumption for ephemeral or exploratory tasks — what economists would recognise as a modern Jevons paradox: greater efficiency lowers marginal cost and drives much higher aggregate use, often with little corresponding productivity gain.

Strategic and geopolitical stakes

The material reality behind those "tokens" matters. GPUs, energy, rare metals and data‑centre capacity are physical and scarce — especially under current U.S. export controls on advanced chips. Emergency 10× scaling is not just a software switch; it implies rapid reallocation of precious hardware and power, and potential asset misallocation if the surge proves transient. Is this a one-night peak or the new baseline that requires long‑term capital expenditure? That question is now strategic, not merely technical.

What comes next for an infrastructure provider

Tencent is not a niche app maker; with WeChat (微信), QQ and enterprise WeChat (企业微信) it controls massive distribution and a meaningful "compute valve" in China's digital economy. The WorkBuddy incident sharpens the policy choice: throttle user‑side excesses and guide compute toward industrial and public‑good use, or keep subsidising consumer experimentation at the expense of scarce infrastructure. Regulators, operators and platform owners will need to decide whether to put brakes on “digital idling” or to continue funding a new kind of virtual exuberance — because compute is now strategic, geopolitical and material, not merely elastic.