A Wall Street analyst on Thursday downgraded AMD stock after tests showed the chipmaker badly lags Nvidia in developing chips for artificial intelligence.
Jefferies analyst Blayne Curtis lowered his rating on AMD stock to hold from buy and cut his price target to 120 from 135.
On the stock market today, AMD stock fell 3.2% to close at 106.65.
Data from the Jefferies graphics processing unit (GPU) benchmarking report showed that AMD's MI300x series AI processors lag Nvidia's older-generation H200 processors in performance, Curtis said in a client note.
The finding supports a comment Nvidia Chief Executive Jensen Huang made at the company's GTC conference last week. At the show, Huang said, "Everybody's still trying to catch up to Hopper, and I haven't seen a competitor to Hopper yet."
Hopper products, including the H200, are from Nvidia's previous-generation series. Nvidia's current-generation AI product is the Blackwell series, with Blackwell Ultra due out later this year.
Curtis said he expects AMD to fall further behind Nvidia with Blackwell and next year's Rubin processors. AMD has "limited traction in AI" and Street estimates are too high, he said.
Further, Curtis expects competition in the space from Intel under its new CEO Lip-Bu Tan.
"Our proprietary benchmarking report suggests real-world throughput of Nvidia's H200 across a range of open-source models is substantially higher than AMD's MI300x, despite MI300x's higher advertised TFLOPs (trillion floating-point operations per second) and memory bandwidth," Curtis said.
"A valid pushback would be that the AMD solution is not as well optimized for the models we tested, but we view this as exactly the point," he said. "These results underscore the importance of Nvidia's mature software stack for managing GPU efficiencies and this gap will widen as system performance (NVL 72 vs. 8 GPU HGX form factor) becomes more important."
AMD Stock, Others Retreat On Demand Concerns
AMD stock and others exposed to the AI megatrend have fallen recently amid concerns about a pullback in spending on AI data centers.
On Tuesday, Joe Tsai, chairman of Chinese e-commerce giant Alibaba, suggested that the AI data center market is in a bubble. He said companies were investing astounding amounts ahead of actual demand.
On Wednesday, investment bank TD Cowen said recent data center lease cancellations and deferrals by Microsoft were "more pervasive" than initially thought. Cowen's incremental channel checks indicate that the list of third-party data center operators affected by Microsoft's lease cancellations has expanded.
In the last six months, Microsoft has walked away from more than 2 gigawatts of data center capacity that was in the process of being leased in the U.S. and Europe, Cowen said. The company also deferred and canceled existing data center leases in both the U.S. and Europe in the last month, the firm said.
"We continue to believe the lease cancellations and deferrals of capacity point to data center oversupply relative to its current demand forecast," Cowen analyst Michael Elias said in a client note.
Meanwhile, Alphabet's Google has stepped in to take capacity that Microsoft walked away from in international markets. And Facebook parent Meta Platforms has taken additional capacity in the U.S., Cowen said.
Semiconductor Stocks News And Analysis
Follow Patrick Seitz on X, formerly Twitter, at @IBD_PSeitz for more stories on consumer technology, software and semiconductor stocks.
