Qualcomm, long known for the chips inside smartphones, is making a serious push into the data center business that powers artificial intelligence.
According to Tom's Hardware, the company unveiled a new "HBC" near-memory architecture along with two AI accelerators, the AI250 and AI350. Qualcomm says the design delivers six times higher bandwidth-per-watt compared to HBM memory and 200 times the capacity of on-chip SRAM, and claims it has "broken the memory wall" — industry shorthand for the bottleneck that occurs when fast processors are forced to wait on slower memory.
Alongside the accelerators, The Decoder reports that Qualcomm is entering the data center market with its own processor, called the Dragonfly C1000. That move puts Qualcomm into more direct competition with the chipmakers that currently dominate AI infrastructure.
Qualcomm frames all of this as part of a broader plan. In its own announcement, the company described a comprehensive data center strategy aimed at diversifying its business, saying it expects "multiple inflection points over the next three to five years." In plain terms, Qualcomm is signaling that it sees data center hardware as a major new growth engine, not a side project.
Why it matters: as demand for AI computing surges, a credible new entrant in data center accelerators and processors could expand the options available to companies building AI systems and intensify competition in one of technology's most valuable markets.