Firing people to fund robots

In the same week, Meta announced it would cut 8,000 jobs, roughly 10% of its workforce, to fund AI investments. Tesla disclosed a $25 billion capital expenditure plan for 2026, triple its prior year, pouring money into Optimus robots, Cybercab, and AI infrastructure, and warned investors to expect negative free cash flow for the rest of the year. Microsoft offered voluntary buyouts to about 7% of its U.S. employees. The pattern is unmistakable: headcount down, compute up. Companies are literally converting human labor budgets into GPU budgets. And the market is rewarding them for it.

The new math

The numbers tell a stark story. Meta plans to spend between $115 billion and $135 billion in 2026, an increase of at least $42 billion from the prior year, with the bulk going to AI. At the same time, it is eliminating 8,000 positions and scrapping 6,000 open roles. The internal memo framed it plainly: the cuts are needed "to offset the other investments we're making." Tesla's shift is even more explicit. Its $25 billion capex figure is three times the $8.5 billion it spent in 2025. The money flows to data centers, in-house chips, manufacturing lines for humanoid robots, and the Cybercab robotaxi program. Elon Musk is building machines to replace the workers he does not need to hire. Microsoft, for its part, is offering early retirement packages to long-serving employees whose age plus tenure equals 70 or more. A softer approach than outright layoffs, but the direction is the same: fewer humans, more infrastructure. Across the industry, over 92,000 tech workers have lost their jobs in the first four months of 2026 alone. A quarter of U.S. layoffs in March explicitly cited AI as a factor.

The sanitized framing

Meta's language is carefully constructed. "Restructuring to invest in AI" sounds like a strategic pivot, not a headcount swap. But that is precisely what it is: a direct substitution of human labor budgets for compute budgets. Mark Zuckerberg said in January that "2026 is going to be the year that AI starts to dramatically change the way that we work," noting that individuals using AI tools heavily could now complete projects that previously required entire teams. Mustafa Suleyman, Microsoft's AI chief, went further, predicting that AI would be able to replace most white-collar work within 12 to 18 months. These are not throwaway comments. They are signals to investors, boards, and the labor market about what comes next. When a CEO says one person can now do a team's job, the unspoken conclusion is obvious: you do not need the team.

The incentive structure

The investor reaction tells you everything about the incentive structure at play. Jamie Dimon estimated that five hyperscalers, Microsoft, Amazon, Alphabet, Meta, and Apple, would increase their annual AI-driven capital spending from $450 billion in 2025 to $725 billion in 2026. Markets have broadly rewarded companies that announce this combination: cut humans, buy GPUs, signal commitment to AI. The formula is simple. Layoffs reduce operating expenses immediately. AI capex is capitalized, spreading the cost over years. The short-term effect on earnings is positive, or at least narratively compelling. The long-term bet is that automation will generate returns that dwarf what those workers were producing. This is not unique to tech. Oracle cut up to 30,000 employees while ramping AI data center spending. Block eliminated over 4,000 people, 40% of its workforce, in February. Snap cut about 1,000 jobs, 16% of its headcount. Atlassian cut 1,600 to "self-fund further investment in AI." The playbook has gone cross-industry.

The honest counterargument

It is worth engaging with the strongest version of the other side. Maybe these roles genuinely are being automated, and the capital reallocation is rational. Some of these cuts are in areas where AI tools are demonstrably more productive: code generation, content moderation, customer support, internal tooling. If an AI coding assistant lets a team of ten engineers produce the output that previously required fifteen, hiring those five extra people is not compassionate, it is wasteful. Companies that refuse to adapt risk losing to competitors that do. But there is a critical distinction between genuine automation and what analysts are increasingly calling "AI-washing." Genuine automation means the work is actually being done by machines now. AI-washing means using the AI narrative as cover for cost cuts that are really driven by slowing revenue, rising interest rates, over-hiring during the boom, or shifting consumer demand. A Forbes tracker noted that companies have been pointing to AI capabilities when announcing layoffs even in roles that have nothing to do with AI deployment. The honest answer is that both things are happening simultaneously, and it is hard to tell them apart from the outside.

The knowledge problem

Here is the part that does not get enough attention. When you cut people before the tools are ready, you lose the institutional knowledge the AI still needs to learn from. A workforce analytics study noted that companies eliminating positions are mostly "betting on future AI capability, not current AI performance." The productivity gains have not yet materialized on the timelines promised to boards. And the problem compounds: the people being let go are often the ones who carry the context, the client relationships, and the judgment that takes years to build. You cannot easily rehire what you let go. This creates a risk that is underpriced in the market. If the AI tools underdeliver on their two-to-three-year timeline, companies will find themselves understaffed, with degraded institutional memory, and no quick path to rebuilding.

What this looks like from Singapore

Singapore's labour market tells a more nuanced story. According to the Ministry of Manpower, resident employment growth in 2025 was stronger than in 2024, and retrenchment rates remained low at 1.5 per 1,000 employees. A parliamentary question on AI-driven automation in February 2026 drew a measured response: the government is paying close attention, but the data has not yet shown the crisis that headlines suggest. Global survey data suggests AI is actually a net job creator so far, with 77% of firms reporting new roles created compared to 46% seeing eliminations. But the composition matters. Technical roles in AI engineering and cybersecurity are booming. Entry-level and middle management positions are where the displacement hits hardest. The question for anyone in the region is whether Singapore's tech sector, heavily integrated with these same global companies, will follow the same substitution pattern with a delay, or whether smaller, more adaptive economies find a different equilibrium.

The question that actually matters

For the individual worker, the framing that matters is not "can AI do my job?" That question is too abstract and too binary. The better question is: am I working on the thing that gets the $25 billion budget, or the thing that gets cut? The capital allocation decisions being made right now are not subtle. They are trillion-dollar signals about which categories of work are valued and which are not. The companies are telling you, in their earnings calls and their layoff memos, exactly where the money is going. That does not mean the bet will pay off. It does not mean the cuts are justified. It does not mean AI will deliver what it promises on schedule. But it does mean that the incentive structures, from boards to markets to individual managers, now point overwhelmingly in one direction: fewer people, more compute. Whether that ends up looking wise or reckless is a question we will only be able to answer in hindsight. What we can say now is that the trade is being made, at industrial scale, with other people's livelihoods as the cost basis.