- AI stocks plunged on Wednesday after AMD reported guidance that put a cloud over future AI chip demand.
- AMD said it expects its MI300 AI chips to generate $4 billion in revenue in 2024, which was below some Wall Street forecasts.
- Shares of Super Micro Computer, AMD, and Nvidia erased a collective $143 billion in market value.
AI stocks plunged on Wednesday after AMD offered analysts 2024 revenue guidance for its AI chip that was below analyst estimates.
Shares of AMD fell 10%, helping drag down its peer Nvidia by about 5% in Wednesday trades. Meanwhile, Super Micro Computer, which sells AI-enabled servers using products from Nvidia and AMD, dropped 15% after it reported mixed earnings results.
The three companies wiped out a collective $143 billion in market value in Wednesday's trading session.
AMD's failure to meet Wall Street's lofty AI sales expectations led to concerns that AMD is suffering from a demand problem, though AMD CEO Lisa Su dismissed those concerns and said that supply is the bigger issue for its AI chip sales.
"We are tight on supply. So there's no question in the near-term that if we had more supply, we have demand for that product," AMD CEO Lisa Su said on the company's earnings call.
AMD raised its 2024 revenue guidance for its MI300 AI chip by $500 million to $4 billion. Despite the increase, that figure failed to meet lofty investor expectations, with some expecting upwards of $8 billion in revenue from the AI chip this year, according to Bloomberg.
While supply constraints seem to be a headwind for AMD's AI chip business, so too could be competition. The company's MI300 chip is a direct competitor to Nvidia's immensely popular H100 chip, and AMD claims that its chip outperforms Nvidia's.
But according to Goldman Sachs, Nvidia's next generation H200 chip is 2-times more powerful in inference than its older H100 chip, so AMD may already be behind the ball with its AI chip offering.
Nvidia CEO Jensen Huang sure seems to think so.
In an interview last month, Huang said his competitors would essentially have to pay its customers to use their AI chips as they aren't comparable to the compute power offered by Nvidia's GPUs.
"Our total cost of operations is so good that even when the competitor's chips are free, it's not cheap enough," Huang said at the Stanford Economic Summit.
Whether Nvidia will be able to meet Wall Street's lofty sales and profit expectations won't be known until the company reports earnings on May 22.