2026.04.24 — Big Tech's 'AI arms race' shakes up jobs · Tech AI News

Today's key takeaways
Google pledged up to $40 billion to Anthropic, tightening the AI infrastructure alliance another notch. On the same day, Meta and Microsoft announced large-scale layoffs and a voluntary retirement program, making it plain that 'AI investment money' is being carved out of payroll. The thread to watch today: the AI infrastructure race is now shaking the structure of the labor market itself.

1. Google to invest up to $40 billion in Anthropic — the AI infrastructure alliance accelerates

Source: TechCrunch / Axios / Bloomberg | 2026.04.24
What happened: According to multiple outlets, Google signed a deal to invest $10 billion in cash upfront (at a $350 billion valuation) and a further $30 billion if performance targets are met. The agreement also includes supplying Anthropic with 5 gigawatts (GW) of Google Cloud computing capacity over the next five years. Anthropic took an additional $5 billion from Amazon the week before, and is reported to have committed up to $100 billion in spending with Amazon for roughly 5 GW of compute.
Why it matters: Google has cemented its dual position as both a direct competitor to Anthropic and its core infrastructure supplier. It is a clear signal that the AI model race has moved entirely from 'algorithms' to 'securing compute'.

2. Meta to cut 10% of staff, about 8,000 people — "to cover AI investment"

Source: ZDNet Korea / Money Today / Seoul Shinmun | 2026.04.24
What happened: In an internal memo, Meta announced it will lay off 8,000 employees, about 10% of its workforce, starting May 20. It will also leave 6,000 planned openings unfilled, for an effective reduction of roughly 14,000 people. Chief People Officer Janelle Gale described the move as a way to run the company more efficiently and offset other ongoing investments. Meta has set this year's capital expenditure at $135 billion, nearly double last year's.
Why it matters: This is the largest cut since the 2022–2023 layoffs, and it is explicitly a restructuring to fund AI, in this case the push to build 'personal superintelligence'.

3. Microsoft introduces its first voluntary retirement program in 51 years

Source: AI Times / Korea Economic Daily / Seoul Economic Daily | 2026.04.24
What happened: Microsoft said it will offer a voluntary retirement program to US employees for the first time since its founding in 1975. Eligible staff are those whose age plus years of service total 70 or more, which foreign media report covers more than 7% (about 8,000) of its roughly 125,000 US employees. CPO Amy Coleman stressed the need to simplify the organization, saying the company has never moved with this much speed and urgency. Microsoft has committed to $140 billion in capital expenditure through the fiscal year ending in June.
Why it matters: This reads less as cost-cutting and more as a signal of an 'AI-first' reorganization. It also dovetails with Microsoft's push to reduce its dependence on OpenAI and accelerate its own model development.

4. Google unveils a wave of 'AI agent tools' at its cloud conference

Source: Bloomberg | 2026.04.22
What happened: Bloomberg reports that Google Cloud showed off a suite of tools at its annual Las Vegas conference for building enterprise AI agents and tracking their progress. It includes a dedicated inbox where virtual bots can post information and progress reports, plus updates across the Workspace productivity tools. It is a direct challenge in the enterprise agent market that OpenAI and Anthropic have led so far.
Why it matters: The vision that 'AI agents will fundamentally change the average office worker's day' arrived in the same week as the Meta and Microsoft layoff announcements, and that timing says a lot.

Today's insight

'AI budget = payroll savings' is becoming the formula. Meta's and Microsoft's layoffs and retirement offers, and the $100-billion-scale capex commitments made in the same period, are not separate events but one movement. Big Tech is settling into a structure where the money spent on AI infrastructure is effectively funded by workforce efficiency.
Compute is power. As the Google–Anthropic and Amazon–Anthropic cases show, the line between model companies and cloud companies is disappearing. Going forward, AI competitiveness is likely to be judged less by 'how good a model you build' and more by 'how much stable, gigawatt-scale power and compute you have locked in'.
This is not someone else's problem for office workers in Korea. Once the enterprise AI agent tools from Google and Microsoft spread in earnest, Korean companies will move past 'whether to adopt' to 'how fast to adopt'. It seems only a matter of time before Meta's and Microsoft's layoff rationale is quoted verbatim in HR meetings at Korean conglomerates.