AMD investors have changed their minds to panic over Nvidia and Intel alliance. What risks does it bring?

Shares of chipmaker AMD, which collapsed by almost 6% during trading on September 18, managed to recover losses by the end of the day and closed down by only 0.8%. After the emotional reaction of the market to the news about the partnership of AMD's main competitor in the market of AI processors Nvidia with AMD's main rival in the segment of central processing units Intel, analysts said that it is not worth thickening the colors yet.

Shares of chipmaker Advanced Micro Devices ended Thursday trading down 0.8% to $157.9, although they fell 5.9% during the day. Investors considered the deal between its two rivals, Nvidia and Intel, as not good news for the company. Their shares rose 3.5% and 22.8%, respectively.

AMD is best known as the second company after Nvidia in the fast-growing market for graphics processing units (GPUs) for artificial intelligence. AMD also competes with Intel in the large market of central processing units for notebooks and servers - CPUs.

Analysts explain that so far the step to save Intel, taken by Nvidia, does not lead to a decrease in sales at AMD: rather, orders, so far in theory, may lose Taiwanese TSMC, if Nvidia will transfer some contracts to Intel factories.

"The two biggest competitors joining forces is, to put it mildly, not a good outcome for AMD," said J.Gold Associates principal analyst Jack Gold in a note cited by MarketWatch. He said the problem isn't just Intel's consolidation, but potentially the loss of market share in the future to chip startups. In addition, enterprise customers who already use Intel products are likely to prefer Nvidia over AMD, Gold added.

The deal with Nvidia solves only some of Intel's problems, but does not address the main cause of its woes: the company's manufacturing technology is still inferior to Taiwan Semiconductor Manufacturing, writes Barron's. The company has lost ground in several markets that are important to it - CPUs for personal computers and GPUs for data centers. Its Gaudi AI chips have only a small share of the market, with sales under $500 million in 2024.

AMD, on the contrary, shows progress quarter after quarter. Gold said the company should maintain a significant advantage in AI "because Nvidia is supply constrained," meaning it is unable to bring as many chips to market as it needs. "If someone needs something and one vendor doesn't have it, customers will go look at another vendor," he added, referring to AMD.

AMD chips also show "consistent performance" at a lower average price than Nvidia's, "making them a popular alternative amid Nvidia's high prices," Brian Mulberry, senior portfolio manager at Zacks Investment Management, told MarketWatch in a commentary.

Intel would benefit from partnering with Nvidia, but "that doesn't automatically eliminate the fact that the company is experiencing significant delays in building new factories," Mulberry added. In the long term, if capital spending levels continue, "there will be enough work for everyone developing AI infrastructure," he said.

Mizuho analyst Jordan Klein believes that investors should stay away from Intel shares for now. "I admit that in the short term investors will buy Intel securities on the back of momentum, but for me this is no reason to revise the company's valuation or significantly raise its growth forecast," Barron's quoted him as saying in a note.

Nvidia will invest $5 billion in Intel and become one of its largest shareholders, taking about 4%. Nvidia said the companies are focused "on seamlessly connecting Nvidia and Intel architectures" through NVLink technology, which provides scalability to handle demanding AI and high-performance computing tasks. NVLink allows multiple GPUs to be connected to each other or to CPUs for faster and more efficient data transfer in processing centers.

Intel will develop custom CPUs based on its x86 architecture for data centers, which Nvidia will integrate into its AI infrastructure.

The semiconductor market is now expanding due to the growing use of artificial intelligence to perform inferencing (outputting finished results to the end user) and the shift to edge devices - edge computing on the user's device rather than in a data center, MarketWatch writes. In this market, Nvidia doesn't have such a clear advantage, according to Jack Gold of J.Gold Associates. Edge devices, which process and store data locally rather than through the cloud, also need AI gas pedals, but they mostly run on CPUs, the analyst noted. It was Nvidia's relative weakness in CPUs that he believes was one of the reasons for the partnership with Intel.

The support of Nvidia opens new prospects for Intel after years of unsuccessful attempts to carry out a transformation, notes Reuters. At the same time, the deal carries a risk for Taiwanese TSMC, which now produces flagship processors Nvidia: in the future, part of this business may go to Intel. However, for this Intel still has to eliminate the technological lag from TSMC.

Contract chip manufacturing remains Intel's biggest challenge, writes Barron's. The company launched this ambitious initiative four years ago, hoping to start producing advanced chips on demand from independent developers around the world. The idea was to offer Apple, Samsung, Nvidia and others an alternative to TSMC. However, so far this project has been a failure, the publication states.

After a recent quarterly report, Intel's new CEO Lip-Bu Tan recognized that the costs of the next generation of technology are too high to be covered by Intel's own sales alone. The company needs external customers to keep this business growing. And that's exactly what Nvidia has yet to promise. At Thursday's press conference, Tan and Nvidia CEO Jensen Huang were asked several times about that possibility. Both executives avoided giving direct answers, but Huang took a moment to express his admiration for TSMC's manufacturing capacity, calling it "real magic."

This article was AI-translated and verified by a human editor