Author: Azuma

Reuters reported exclusively on April 18 that three insiders revealed to it that Zuckerberg's Meta plans to launch the first round of large-scale layoffs this year on May 20, with further layoffs to follow.

One insider stated that Meta will cut about 10% of its global workforce (approximately 79,000 employees), which amounts to about 8,000 people, in the first round of layoffs. Another insider indicated that Meta also plans to implement further layoffs in the second half of this year, but the specific timing and scale have not yet been finalized. As observations of AI capability development continue, Meta's executives may adjust their plans.

In another report by Reuters last month, it was revealed by insiders that Meta was considering layoffs of 20% or even more.

As of the time of publication, Meta declined to comment on the timing and scale of the layoffs.

Ten days ago, Meta had just caught up with the AI big league.

Just ten days ago, the AI development team 'Meta Superintelligence Labs' (MSL), led by the talented Chinese-American Alexandr Wang, who was poached at a high price, released its first self-developed AI model, Muse Spark.

Alexandr Wang disclosed that in the past nine months, MSL has rebuilt the entire AI technology stack from scratch. Muse Spark is a native multimodal reasoning model that supports tool invocation, visual chain of thought, and multi-agent orchestration. This is the most powerful model that Meta has released to date. During the training process, MSL observed that the model showed a predictable scalable improvement during pre-training, reinforcement learning, and testing phases.

Muse Spark also supports 'Contemplating Mode,' which utilizes multiple parallel reasoning agents specifically designed to tackle complex scientific problems and reasoning tasks. In tests, MSL found that its performance could compete with extreme reasoning models like Gemini Deep Think and GPT Pro.

As the first substantial product after Meta's heavy investment in AI and shift to a closed-source model, Muse Spark is widely regarded by the market as the beginning of Meta's pursuit of the first-tier AI companies like Anthropic, OpenAI, and Google. Although Meta also acknowledges that the model is still inferior to the flagship models of the top three companies in some capabilities, for Zuckerberg, who has long lagged in the AI race due to the failure of the Llama route, Muse Spark and its subsequent models in the same series are sufficient as chips for his return to the AI table.

The market also gave positive feedback signals for Muse Spark, with Meta's closing price on that day at $612.42, up 6.5%, and it continued to rise over the next 10 days (of course, also influenced by the overall market rally), with the closing price yesterday reaching $688.55.

The blade of AI has first landed on the employees.

From the end of 2022 to the beginning of 2023, Meta initiated a controversial 'year of efficiency' plan, resulting in the largest layoff in the company's history, cutting about 21,000 positions. This time, it is likely to become the largest round of layoffs at Meta since the 'year of efficiency.'

Compared to the 'year of efficiency,' Meta faced significant stock price declines and adjustment pressures following excessive growth during the pandemic. Today's Meta is clearly more financially stable, but the future envisioned by executives is one of fewer management layers and an organizational structure where AI-assisted employees can bring higher efficiency.

Business Insider reported last month that it learned from leaked internal Meta documents that Meta is pushing employees to use AI tools more actively, with a goal set for mid-2026 that 65% of engineers need to have more than 75% of their code written with AI involvement.

According to the self-media Official Layoff (@LayoffAI), which focuses on layoffs at large companies (without its sources, not guaranteed), 'Starting this year, Meta has incorporated 'AI-driven impact' into the performance evaluations of all employees, making it a core metric. Without using AI, promotions cannot be obtained. Meta has become the first large tech company to officially link AI usage to promotions.'

AI iteration of white-collar workers is no longer an isolated case.

Using 'AI iteration of productivity' as a reason for layoffs is no longer an isolated case.

In October last year, Amazon cut up to 30,000 jobs across its logistics, payments, video games, and cloud computing departments. The company's CEO Andy Jassy had previously hinted at this round of layoffs: 'As the company increasingly uses AI to perform tasks that were originally done by humans, the workforce at Amazon may shrink.'

At the end of February this year, Jack Dorsey (also the founder of Twitter) announced that his fintech company Block would cut 4,000 positions, reducing the total number of employees from over 10,000 to less than 6,000, to promote a more streamlined, flat, and AI-centric organizational structure. Block's CFO and COO Amrita Ahuja revealed that after the company announced the layoffs, a large number of corporate executives proactively contacted Block, seeking to replicate this 'script.'

Earlier this week, Snap, a direct competitor of Meta's core product Instagram, also cut about 1,000 jobs. Its CEO Evan Spiegel stated, 'AI will enable our team to reduce repetitive work, improve efficiency, and better support our community, partners, and advertisers.'

Now, the same wind has blown to Menlo Park, California, and Zuckerberg has raised the sword in his hand.

Oh, and there is one more thing worth mentioning. Although Jack Dorsey publicly stated at the time of the layoffs that 'the rapid development of AI is iterating the traditional productivity growth paradigm,' shortly after the layoffs at Block, many of the laid-off employees received rehire invitations.

The AI iteration of white-collar workers may eventually become a reality, but being as hasty as Block to cut 40% of the workforce can easily lead to 'overreaching and making mistakes.'