China's low-cost DeepSeek allegedly has Meta in panic mode

Meta is reportedly in full-blown panic mode over DeepSeek, a low-cost Chinese AI project that’s making waves in the tech world. According to a post on TeamBlind by an anonymous Meta employee, engineers at the company are scrambling to reverse-engineer DeepSeek’s breakthroughs, especially after the release of its third version, DeepSeek v3, which has outpaced Meta’s Llama 4 in key benchmarks.

The kicker? DeepSeek v3 was trained on a budget of just $5.6 million, a fraction of what Meta spends on its AI projects. This has caused internal frustration, especially with Meta’s leadership team, who are now under pressure to justify their ballooning budgets. Some people have even pointed out that the salaries of certain AI leaders at Meta exceed DeepSeek’s entire training budget.

Things got even more intense with the release of DeepSeek r1, rumored to include innovations that could disrupt the AI industry. While specific details remain under wraps, other insiders say this development has shaken not just Meta but also rivals like Google, OpenAI, and Anthropic. DeepSeek’s unique approach to reinforcement learning and other AI techniques is forcing competitors to rethink their strategies.

Meta employees are venting online about the chaos within the company’s generative AI division. The OP even described it as a “too many cooks” situation, with conflicting priorities between engineering and product teams. Another employee said, “looks like all these clowns are about to get Zucked.”

This news comes just a couple of days after President Trump announced a $500 billion AI project dubbed Stargate, “to power the next generation of AI,” Meanwhile, China seems to be pulling ahead with a fraction of the cost. While Meta scrambles to catch up, one thing is clear: the AI race just got a lot more interesting.