The company confirmed plans to reduce headcount to 75,000, down from over 96,000, via attrition and targeted layoffs, many of which have already been executed.

This downsizing is just one piece of an overhaul led by newly appointed CEO Lip-Bu Tan, who stepped in earlier this year to confront years of strategic missteps and financial underperformance. 

Despite a short-term bump in revenue, Intel expects to post a larger-than-anticipated third-quarter loss of 24 cents per share, well above the 18-cent loss forecast by Wall Street analysts. Shares tumbled nearly 6% in Frankfurt following the announcement.

Tan, who is pressing for a more cost-conscious and focused Intel, made his intentions clear in a memo to employees: “There are no more blank checks. Every investment must make economic sense. We will build what our customers need, when they need it, and earn their trust through consistent execution.”

That message is a big difference from the company’s previous approach under former CEO Pat Gelsinger, who championed aggressive expansion plans and heavy investment in next-generation manufacturing processes like 18A and the upcoming 14A node.

Tan, however, appears sceptical of that strategy’s commercial viability, particularly when it comes to selling these technologies to external customers. Reuters recently reported that Intel may abandon plans to offer 18A tech to outside firms altogether.

In what he describes as a “disciplined approach,” Tan is halting construction of new chip plants in Poland and Germany, slowing down development in Ohio, and consolidating packaging operations from Costa Rica to more established hubs in Vietnam and Malaysia.

“I do not subscribe to the belief that if you build it, they will come,” he said in a call with analysts, adding that he will personally approve all major chip designs going forward.

Chief Financial Officer David Zinsner told Reuters the company has taken a “surgical” approach to job cuts, removing about half of Intel’s management layers in the process. The aggressive cost-cutting contributed to restructuring expenses of $1.9 billion in the second quarter alone.

Despite the grim forward-looking forecast, there were signs of short-term improvement. Intel’s second-quarter revenue held flat at $12.9 billion, breaking a year-long streak of declining sales and surpassing analysts’ expectations of $11.92 billion. 

Still, adjusted earnings showed a loss of 10 cents per share, in stark contrast to estimates of a 1 cent gain. The unadjusted figure was worse: a 67-cent-per-share loss.

Intel’s competitiveness has eroded significantly over the past decade. While Nvidia has surged to the top in the high-growth AI chip segment, and AMD continues to nibble away at its market share in both consumer and server chips, Intel has struggled to deliver products on time and budget. 

Its push into the chip foundry business—intended to rival industry leader TSMC—has largely faltered.

“They may have overspent on 18A … but I think this is the painted picture of a new fiscally disciplined base that they’re going to go from here. I think that’s the right approach,” said Ben Bajarin, CEO of Creative Strategies.

The global economic backdrop hasn’t helped. Although semiconductors have been spared from sweeping tariffs, customers remain cautious due to macroeconomic challenges, usually pulling forward orders or delaying long-term commitments.

On April 9, U.S. officials quietly informed Nvidia that shipping the H20 would now require a special licence. Five days later, Washington said this restriction isn’t temporary, but indefinite. 

Nvidia didn’t alert all its customers in time. By April 15, the damage was evident. Shares plunged 6% in after-hours trading. The company had no choice but to take the charge.

The H20 is Nvidia’s most advanced AI chip that’s still legal for sale in China. It’s a workaround product — designed after earlier export rules stopped sales of more powerful chips like the H100 and A100. 

While the H20 has slightly watered-down processing power, it still features rapid memory access and high-speed connectivity. Those features are good enough to raise eyebrows in Washington.

And here’s the thing, this chip was never really slow. It’s effective at inference — the stage where AI gives users answers. That’s where the market is heading. It was the perfect solution for Chinese giants like Tencent, Alibaba, and ByteDance, who were reportedly stocking up in bulk. 

Even startups like DeepSeek were buying in aggressively. Their V3 model? Trained on H20s, according to a policy group in D.C.

“The Commerce Department is committed to acting on the President’s directive to safeguard our national and economic security,” a spokesperson said.

Behind the scenes, Chinese customers were still expecting deliveries by year-end — unaware that the game had changed. Analysts now expect a surge in demand for Huawei’s alternatives.

“By restricting the H20 system, U.S. regulators are effectively pushing Nvidia’s Chinese customers toward Huawei’s AI chips,” said Nori Chiou of White Oak Capital Partners. “Huawei’s chip design and software capabilities are likely to advance quickly as it gains more customers and development experience.”

The consequences involve China accounting for about $17 billion in Nvidia’s sales last year. That’s 13% of its global revenue. Losing that market — or even just slowing it — changes the forecast for Nvidia’s dominance. CEO Jensen Huang had already warned that revenue from China was down to half its former levels.

Worse still, competition inside China is heating up. Nvidia now lists Huawei as a direct rival. And that’s not symbolic — it’s strategic.

There’s also the matter of U.S. politics. Some voices in Washington say Nvidia’s chips could support China’s military supercomputing. The H20 may not have been built for that, but its connectivity potential sparked concern. With DeepSeek’s use of the chip under scrutiny and Tencent allegedly using H20s to train large AI models, it became harder for Nvidia to make its case.

“At least one of the buyers, Tencent, has already installed H20s in a facility used to train a large model, very likely in breach of existing controls,” the Institute for Progress wrote.

In the same breath, Nvidia has been pledging major investments in America. Just a day before the new restrictions went public, the company announced plans to build AI servers worth up to $500 billion in the U.S. over the next four years. The timing didn’t go unnoticed.

No one at Nvidia is saying much beyond the official filing. But we know what’s next: the company’s quarterly results are due on May 28. Until then, investors and regulators will be watching every move. 

Right now, the world’s most valuable chipmaker is learning a tough lesson — sometimes, being caught in the middle of a geopolitical power play can cost you billions.