AI and humans competing for resources—start with insiders at technology companies

July 12, CCTV Finance cited a CNBC report: US employment consulting firm Challenger, Gray & Christmas released a monthly report. As of the end of June, in the first half of this year, cumulative tech-industry layoffs in the US have nearly reached 140k, mainly due to restructuring and automation driven by AI.

Amazon saw the harshest cuts. This year in January it laid off 16k people. Three months earlier it had just cut 14k. Across the two rounds, the total is about 9% of the company’s total employee count.

CEO Andy Jassy’s explanation was straightforward: Over the next few years, AI will significantly improve efficiency, so the company needs to reduce layers to stay flexible.

Meta also carried out multiple rounds this year. Its most recent one was in May, when it cut 8,000 people, accounting for 10%. Oracle and Cisco followed closely.

With 140k layoffs in half a year, this figure is close to all of 2025. And “AI replacement” has already become the largest share of US tech layoffs, reaching 40%...

At the same time, there have also been plenty of layoffs among tech companies in China.

Alibaba reportedly had 66k fewer layoffs in a year, formed a Token Hub business group, and canceled the 13th salary.

ByteDance’s internal leak said non-AI departments optimize in rolling fashion by about 20% every half year, while AI core departments are only 5%.

Tencent Docs’ Beijing team was reported in May to be entirely shut down, with employees required to leave by a deadline—pregnant women were not an exception. The official response said “internal transfers are open,” but the Beijing office location essentially disappeared in practice. The product logic changed too: it gave up traffic expansion and shifted to high-paying scenarios in Tencent Cloud’s To B AI ecosystem, such as AI writing and smart PPTs.

Baidu’s MEG business group saw optimization in some departments of 20%-30%, cutting from new graduates to employees with up to ten years of experience, and some teams disappeared in whole units.

Xiaohongshu was also conducting layoffs before going public. Rumors say the layoff ratio could be as high as 30%, targeting employees with performance scores below 3.5.

Meituan and NetEase have also reportedly seen a large number of layoffs...

Against the backdrop of layoffs, these tech companies’ profits are also rising in parallel—and the higher the layoffs, the harsher they are...

Based on this logic, it’s not that many tech companies are cutting because they don’t have money. Instead, employees’ future performance growth no longer seems to be having much effect.

And AI actually doesn’t save money compared with real human employees. Meta’s AI-related capital expenditures this year are $125-145 billion—nearly double last year’s. Microsoft expects $190 billion for the full year. Google has raised it to $180-190 billion. Amazon will burn $200 billion for the full year. The industry total is about $725 billion.

With about 140k people laid off in the first half of the US, using $250k per person, the annual savings would be about $35 billion—enough to cover only 5% of the silicon-based bill.

To raise money for AI, US tech companies have already accumulated $182 billion this year, up sharply by $13 billion versus the same period in 2025.

According to data compiled by Bloomberg, over the past five years, the five companies with the largest data center investment scale in the US (Alphabet, Amazon, Meta, Microsoft, and Oracle) collectively added about $350 billion in debt. Their debt size has doubled.

Among them, Oracle has even been downgraded by S&P to “BBB-” due to issuing too much debt, only one notch above the so-called junk tier.

In its rating report, S&P explicitly listed OpenAI as a “key credit risk” for Oracle. It noted that Oracle’s AI business cash burn far exceeded expectations. Capital expenditure forecasts have been raised dramatically from $60 billion previously to $95 billion in 2027, while corresponding revenues will not be realized until years later.

Among Oracle’s $638 billion in contractual obligations, OpenAI accounts for about half. S&P warned that if OpenAI faces an operating crisis, Oracle will be stuck with a large amount of data center capacity that cannot be absorbed, and financial pressure will be amplified sharply.

The expansion logic of other tech companies is basically consistent with Oracle too—it’s all about investing in data centers.

Once there is even a slight shift in optimistic expectations for future AI, the picture becomes almost unimaginable...