Qualcomm looks to take on Nvidia, AMD as it enters AI accelerator market

[Qualcomm headquarters sign in San Diego, California, USA.]
JHVEPhoto

Qualcomm (NASDAQ:QCOM [https://seekingalpha.com/symbol/QCOM]) announced its entry into the artificial intelligence accelerator market on Monday, as it looks to take on industry heavyweights Nvidia (NASDAQ:NVDA [https://seekingalpha.com/symbol/NVDA]) and AMD (NASDAQ:AMD [https://seekingalpha.com/symbol/AMD]).

The Cristiano Amon-led company unveiled its Qualcomm A1200 and AI250 chip-based accelerator cards and racks, all of which use the company's neural processing units technology.

The A1200, set to be released in 2026, supports 768 GB of LPDDR per card for higher memory capacity and lower cost, San Diego-based Qualcomm said. The AI250 — available in 2027 — will have an “innovative memory architecture based on near-memory computing,” which Qualcomm said will enable 10 times higher effective memory bandwidth and lower power consumption.

“With Qualcomm AI200 and AI250, we’re redefining what’s possible for rack-scale AI inference. These innovative new AI infrastructure solutions empower customers to deploy generative AI at unprecedented [total cost of ownership], while maintaining the flexibility and security modern data centers demand,” said Qualcomm executive Durga Malladi in a statement.

Qualcomm also teased that it would unveil another AI accelerator in 2028. The company has committed to an annual cadence, similar to what Nvidia and AMD have already done.

Separately, on Monday, Saudi artificial intelligence company Humain said it would use Qualcomm's AI accelerators. Humain is targeting 200 megawatts starting in 2026, the company added.

A Qualcomm spokesman declined to answer how much the accelerators cost or what foundry was making the processors. Samsung (OTCPK:SSNLF [https://seekingalpha.com/symbol/SSNLF]) and Taiwan Semiconductor (TSM [https://seekingalpha.com/symbol/TSM]) both count Qualcomm as a customer.

Qualcomm's announcement is notable, given the size of the AI accelerator market (AMD CEO Lisa Su has estimated [https://www.aol.com/finance/amd-ceo-lisa-su-says-202818166.html] it could be worth $500B or more) and the market shares of Nvidia and AMD.

In its most recent quarter, Nvidia's data center revenue soared 56% year-over-year to $41.1B, accounting for nearly 88% of its total $46.7B in quarterly revenue. Conversely, AMD's data center segment accounted for $3.2B out of a total $7.685B in revenue.

While Qualcomm still generates the vast majority of its semiconductor revenue from handsets — $6.33B out of a total $8.99B — the company has worked to diversify its revenue. In the most recent quarter, automotive revenue rose 21% year-over-year to $984M, while sales attributed to internet of things jumped 24% to $1.68B.

Qualcomm is set to report its fiscal fourth-quarter results after the close of trading on Nov. 5. A consensus of analysts expect [https://seekingalpha.com/symbol/QCOM/earnings/estimates?period=quarterly] the company to earn $2.86 per share on $10.74B in revenue.

MORE ON QUALCOMM, AMD AND NVIDIA

* Nvidia: The Big 3 Issues Before Earnings [https://seekingalpha.com/article/4832412-nvidia-the-big-3-issues-before-earnings]
* AMD's OpenAI Deal Won't Catch Nvidia, But It Could Still Power The Stock Higher [https://seekingalpha.com/article/4831411-amds-openai-deal-wont-catch-nvidia-but-it-could-still-power-the-stock-higher]
* AMD: You're About To Get Crushed [https://seekingalpha.com/article/4831137-amd-youre-about-to-get-crushed]
* Intel has an AI accelerator strategy. Here's what it is. [https://seekingalpha.com/news/4508270-intel-has-an-ai-accelerator-strategy-heres-what-it-is]
* IBM's key quantum computing algorithm can work with standard AMD chips: report [https://seekingalpha.com/news/4508128-ibms-key-quantum-computing-algorithm-can-work-with-standard-amd-chips]