Qualcomm Announces Meta as First Major Customer for Data Center Chips Qualcomm has moved to reposition itself from a mobile chip specialist to a serious AI data center contender, aligning a long-term hardware roadmap with a marquee customer commitment from Meta and a major software acquisition.

In the run-up to its investor day in New York, reports indicated Qualcomm was nearing completion of a roughly $4 billion all‑stock acquisition of AI software startup Modular, which builds the Mojo programming language and MAX inference engine to run models across chips from Nvidia, AMD, Intel, and Qualcomm. The deal, now confirmed at about $3.9 billion in stock, is expected to close in the second half of the year and is intended to provide the software layer needed to make Qualcomm’s new data center hardware more attractive to cloud operators.

At Wednesday’s event, Qualcomm filled in details of its “Dragonfly” strategy, a new brand spanning data center CPUs, AI inference accelerators, and custom silicon for hyperscalers. It introduced the AI300 accelerator, joining the previously announced AI200 and AI250 chips, and argued that decades of mobile design give it an edge in power efficiency as energy-hungry AI workloads strain global electricity grids.

The pivotal announcement came when Qualcomm revealed that Meta will be the first named customer for its Dragonfly C1000 general‑purpose server processor, the strongest signal yet that the company is “serious about competing in the AI infrastructure market.” Meta has committed to using the C1000 and its successors across its data centers, though the chip will not be available until 2028, making this a forward‑looking bet rather than an immediate deployment.

From Wall Street’s perspective, the combination of a clear data center roadmap and Meta’s endorsement as the first Big Tech customer for these chips, alongside a higher revenue outlook, boosted Qualcomm’s shares by as much as 15%. Still, analysts note that Qualcomm faces entrenched competition from Nvidia and others, with its performance at hyperscale yet to be proven.

Continue reading https://foxvector.com/stories/019efc1a-83b7-1b61-719b-2c94d9b45c9a