Meta AI layoffs: cutting 10% to fund a superintelligence bet
Meta plans to lay off roughly 8,000 employees, or about 10% of its workforce, and close around 6,000 open roles as it doubles down on artificial intelligence. The cuts, effective from May 20, are framed as a drive for efficiency and a way to “offset the other investments we’re making,” according to an internal memo. Those investments are enormous: Meta’s capital expenses for data centers and other AI infrastructure reached USD 72.2 billion (approx. RM333.1 billion) in 2025 and are projected to climb to at least USD 115 billion (approx. RM531.0 billion) this year. Management has explicitly tied this AI infrastructure spending to an ambition to deliver “personal superintelligence for everyone” and to build AI that surpasses human intelligence. The Meta AI layoffs therefore signal not a retreat, but a reallocation of resources from headcount toward long‑term AI capabilities and competitive positioning.
Inside Meta’s AWS Graviton deal and the rise of CPU‑heavy AI workloads
Beyond GPUs for model training, Meta is quietly reshaping its compute stack with a massive multi‑year deal to run on tens of millions of AWS Graviton CPU cores. These ARM‑based chips will underpin AI infrastructure focused on more CPU‑intensive, “agentic” workloads—systems that can plan, reason and execute multi‑step tasks autonomously. As Meta’s head of infrastructure put it, diversifying compute sources has become a strategic imperative as the company scales its AI ambitions globally. Graviton processors are designed to deliver high performance at lower cost for cloud workloads, with the latest Graviton 5 generation offering materially higher compute performance than its predecessor. By leaning into AWS Graviton, Meta is signalling that the next wave of AI will not be only about giant foundation models, but also about orchestration, tool‑use, and large‑scale inference—areas where efficient CPUs can be just as critical as cutting‑edge GPUs.
Siemens shows industrial players are joining the AI infrastructure race
The new wave of AI infrastructure spending is not limited to consumer internet giants. Siemens is committing USD 285 million (approx. RM1.3 billion) to expand manufacturing capacity and build specialised AI data centers in North America, creating more than 900 jobs in the region. The move reflects both the pull of supportive US incentives and frustration with what Siemens sees as heavy European red tape. At industry events, Siemens has pushed an “Industrial AI for the real world” narrative, demonstrating how AI can be woven into production lines to boost efficiency and reduce dependence on scarce skilled labour. Its healthcare subsidiary Siemens Healthineers is following a similar strategy by shifting production closer to US innovation hubs. Together, these decisions underline how AI data centers growth is becoming a central pillar of capital allocation, even for companies better known for turbines, factories, and medical devices than for social networks.
Cloud rivalry, specialised chips and the next phase of big tech AI strategy
Meta’s pivot from headcount to AI capex is intensifying competition among cloud and chip providers. By choosing AWS Graviton at massive scale, Meta strengthens Amazon Web Services as an AI infrastructure partner while still racing against Amazon, Google, Microsoft and OpenAI on consumer‑facing AI. At the same time, Microsoft is reportedly pursuing voluntary buyouts for a slice of its US workforce as it channels billions into AI, mirroring Meta’s efficiency‑driven restructuring. Across the board, big tech AI strategy now hinges on owning or influencing the compute layer—from custom CPUs and GPUs to optimised data centers. This raises the stakes for semiconductor makers and could pressure margins as capital demands soar. Yet it also opens the door to more specialised chips tuned for inference, agentic AI, and industrial applications, reshaping how value is captured across the AI supply chain.
What the new AI capex cycle means for workers and the wider ecosystem
For workers, the latest Meta AI layoffs are part of a broader pattern: companies are trimming staff while arguing that AI will automate tasks and maintain or even raise productivity with leaner teams. Meta, Amazon, and others increasingly present AI tools as substitutes for certain operational roles, even as they aggressively hire for AI research, infrastructure engineering, and specialised software positions. In regions where AI data centers growth is accelerating, such as North America, that can translate into new opportunities in data center construction, cloud operations, cybersecurity, and industrial AI deployment. However, the transition is uneven; many affected employees do not automatically match the skills in highest demand. As more enterprises follow Meta and Siemens in reallocating capital towards AI infrastructure spending, policymakers and educators will face growing pressure to support reskilling, while workers navigate a labour market where AI shapes both the risks and the new career paths.