AMD shares jump as AI-driven data-centre strength fuels growth outlook

Advanced ​Micro Devices (AMD-Q) rose 7.3 per cent in premarket trading after ​the U.S. chipmaker forecast faster growth ‌through 2027, reassuring investors that demand for AI infrastructure remains strong.

AMD, which recently joined the ranks of trillion-dollar companies, has gained ⁠about ​126 per cent this year and is poised to add about $58 billion in market value if premarket gains hold.

The company’s strong outlook highlighted growing momentum in its server processor ​business, signaling that spending on AI computing ‌was extending beyond Nvidia’s accelerators, which dominate the market.

“We see demand across the data-center business, across our CPU business, as well as our data-center AI business,” CEO Lisa Su said, adding ‌the company ​had been preparing ‌for that growth for several years.

Su said on Tuesday ​that the company expects overall revenue to ⁠rise more than 80 per cent in 2027, which is ⁠higher than the 77 per cent growth analysts estimate, according to LSEG data.

AMD’s ​product roadmap and expanding software ecosystem position it to capture a meaningful share of the AI market, while strengthening its role as the leading alternative to Nvidia and supporting further gains in server CPUs, ⁠said analysts at TD Cowen.

Last week, rival Intel’s shares plunged more than 20 per cent after the company reported weak PC demand, fueling concerns about its reliance on the PC market and relatively limited exposure to AI.

Analysts at ⁠Morgan Stanley are cautious about assuming current ​growth in the CPU server business will continue indefinitely amid ⁠an industry-wide shortage, but they believe AMD is executing better than competitors and ‌could continue to capture a larger share of the market.

The company’s ​third-quarter data-center revenue jumped 97 per cent to US$8.55-billion, topping analysts’ expectations of US$8.20-billion.

The Santa Clara, California-based company trades at about 31 times forward earnings, compared with 14 times ​for Nvidia (NVDA-Q) and 16.5 times for Broadcom (AVGO-Q) .