Intel reported second-quarter revenue up 22 percent from a year earlier on Wednesday, but the growth came with a round of layoffs in its data center and AI division, part of a restructuring the company says is necessary to turn itself around. The quarter’s brightest number was elsewhere: the yield rate at its foundry business, which makes chips for other companies, improved to 85 percent.
The combination of growth and job cuts captures the strategy Intel’s leadership has adopted. The company is trying to restore its position in personal computers and data centers while building a foundry business to compete with Taiwan’s TSMC, the dominant manufacturer of the world’s most advanced chips. That second ambition requires billions in spending on factories and process technology, and the company has been cutting costs everywhere else to pay for it. The data center layoffs are part of that trade.
Revenue growth of 22 percent was driven by a recovery in the PC market and by demand for Intel’s server chips, which had lost ground to AMD and to the custom processors designed by cloud companies. The company said its AI server products are gaining acceptance, though it remains far behind Nvidia in the market for the accelerators that power AI training. Intel’s leadership has acknowledged the gap and has said the company’s path runs through its foundry, where it hopes to manufacture chips designed by other companies, including AI chip designers.
The 85 percent yield figure is the number Intel’s executives emphasized. Yield — the share of chips produced on a wafer that work correctly — is the metric that determines whether a foundry is viable, and Intel’s foundry has struggled to reach the levels customers demand. The improvement, executives said, shows the company’s manufacturing process is maturing, and they argued it should give prospective foundry customers confidence. Skeptics note that yields are still below the levels TSMC achieves on its most advanced nodes and that Intel’s foundry has yet to sign the kind of large external customer that would prove the business model.
The restructuring is not finished. The layoffs announced with earnings are part of a broader plan the company has been executing for two years, involving multiple rounds of job cuts and the sale or closure of businesses outside its core focus. Intel’s workforce is down significantly from its peak, and executives have signaled more changes to come. The company is also managing a large debt load taken on during its expansion into manufacturing.
The market’s reaction was mixed. The stock moved up on the results, and Jim Cramer, the CNBC commentator, called Intel a “miracle stock” on air, citing the yield improvement and the revenue recovery. Other analysts were more measured, noting that Intel still loses money on its foundry operations and that the turnaround depends on customers no one has yet named publicly. The company said it expects losses in the foundry unit to narrow over the coming quarters.
The next test is the one Intel has faced for years: converting progress into profit. The PC recovery will not last forever, the data center market remains fiercely competitive, and the foundry business, for all its improvement, has yet to prove it can attract the marquee customers that would make it a real rival to TSMC. Wednesday’s numbers are the most encouraging Intel has delivered in some time. Encouraging, though, is not the same as done.
The layoffs deserve closer reading. Intel has said it is cutting jobs in its data center and AI division even as it hires in areas tied to its foundry strategy, a rebalancing that reflects where the company believes its future lies. Employees in the affected groups were told the reductions are part of the broader restructuring the company has described to investors over several quarters. The moves are painful inside a division that once defined Intel’s dominance, and they underscore how far the company has come from the era when it set the pace of the entire chip industry.
The foundry strategy itself is the largest bet in Intel’s history. Building advanced chip factories costs tens of billions of dollars, and Intel is doing it while competing in product markets against companies that do not carry that manufacturing burden. The company has argued that the bet is necessary — that the era when Intel could design and build its own chips in a closed loop is over, and that manufacturing for others is the only path to the scale its factories need. The 85 percent yield figure is the first hard evidence that the bet is technically working.
What the strategy still lacks is customers. A foundry’s economics depend on filling its factories, and Intel’s own products can only fill so much capacity. The company has said it is in talks with potential customers, including AI chip designers that currently rely on TSMC, but it has not announced the kind of large, named commitment that would transform the narrative. Until it does, the foundry business will remain a cost center that the rest of the company must subsidize.
The earnings call did not change that fundamental picture. The revenue recovery is real, the yield improvement is real, and the cost reductions are real. So is the distance remaining: Intel’s foundry still loses money, its market share in data center chips is a fraction of what it once held, and the competitive environment shows no sign of softening. The market’s reaction — a stock move up, cautious analyst notes — was appropriate for a quarter of progress that is not yet a turnaround.
