Advanced Micro Devices forecast fourth-quarter revenue above market estimates on Tuesday, betting on the multibillion-dollar expansions of data center infrastructure to boost demand for its artificial intelligence chips.
AMD has received significant investments in AI hardware from companies such as ChatGPT maker OpenAI and the U.S. Department of Energy, as investors expect spending on advanced processors will continue. But fears remain that the AI boom has created mounting risks as it is still unclear whether the investments will generate the cash needed to sustain expectations.
The company expects revenue of about $9.6 billion for the quarter, plus or minus $300 million, compared with analysts’ average estimate of $9.15 billion, according to data compiled by LSEG.
The company’s shares were down about 3 per cent in choppy extended trading. Its shares have more than doubled in value this year, with AMD’s stock gains outpacing those of market leader Nvidia, even as the larger rival’s market valuation scaled the $5 trillion mark.
AMD’s shares may be falling in extended trading because short-term traders could be taking profit, said Michael Schulman, chief investment officer at Running Point Capital. Investors are concerned about the overall economy and chip stock valuations that may take time to grow into, he said.
AMD reported third-quarter sales of $9.25 billion, compared with analysts’ average estimate of $8.74 billion.
Fears of an AI bubble have lately rippled through Wall Street and AMD’s results were keenly anticipated.
In AMD’s key data center segment, which houses its AI chips, revenue in the September quarter rose 22 per cent to $4.3 billion, beating estimates of $4.09 billion.
Tech giants have all consistently committed to investing heavily in AI hardware, with a large portion of this spending funneled to chips, benefiting suppliers like AMD.
AMD customer Microsoft reported a record capital expenditure of nearly $35 billion for its fiscal first quarter last month, with roughly half of this spent primarily on chips.
Alongside major Big Tech spending on AI chips, the intense focus on expanding data center footprints has also increased demand for server central processing units (CPUs) – a market where AMD has consistently been gaining share against Intel.
Analysts widely expect this to boost sales of AMD’s traditional data center chips which are used alongside AI GPUs in servers.
Sales in the client segment, which caters to the personal computer market, rose 46 per cent to $2.8 billion in the third quarter.
End-market demand in the PC market has risen as consumers upgrade to AI-enabled PCs while a Windows upgrade cycle has further spurred sales.
Preliminary results from research firm Gartner show an 8 per cent boost in worldwide third-quarter PC shipments.
OPENAI DEAL
For the fourth quarter, AMD expects adjusted gross margin of 54 per cent, compared with estimates of 54.5 per cent. In the third quarter, it recorded an adjusted gross margin of 54 per cent, which slightly topped estimates.
At the same time, AMD is expanding production of its latest chips to meet swelling demand and has also scaled manufacturing up to full AI systems, following suit after Nvidia, resulting in a costly ramp up.
Adjusted for stock-based compensation among other items, AMD reported per-share profit of $1.20 in the third quarter, beating estimates of $1.16.
AMD last month said it would supply AI chips to OpenAI in a multiyear deal that would bring in tens of billions of dollars in annual revenue and give the startup the option to buy up to roughly 10 per cent of the chipmaker.
The deal covers the deployment of hundreds of thousands of AMD’s graphics processing units (GPUs), roughly equivalent to the energy needs of 5 million U.S. households, or about three times the amount of power produced by the Hoover Dam.
This has raised optimism around AMD’s position in the unrelenting race to build the fastest, most widely adopted AI chips.
However, Nvidia’s dominance of the lucrative GPU industry is still largely secure, analysts say, even as AMD competes for a bigger piece of the AI chip pie.