Technology

• Qualcomm is building server racks and CPUs, eyeing AI inferencing clusters in hyperscale data center
• Qualcomm's promised CPU might arrive in 2028, well after rivals have scaled infrastructure
• But investors weren’t excited by Qualcomm’s data center plans, as share prices dipped on earnings release

Qualcomm may finally be ready to move beyond its mobile stronghold and take a real step into the data center market.

During its recent Q3 earnings call, the company confirmed it is in “advanced discussions” with a hyperscale customer regarding new silicon targeted at server infrastructure.

This could mark a significant shift for the firm, which has long teased ambitions to bring ARM-compatible processors into cloud computing spaces but never quite followed through at scale.

Ambitions grow, but execution will be critical

Qualcomm's CEO Cristiano Amon framed the moment as an opportunity created by the evolving nature of AI workloads, arguing cloud service providers are increasingly focused on efficiency metrics like tokens per watt and tokens per dollar, not just performance.

This, coupled with a broader industry move away from x86 CPUs toward custom ARM-based alternatives, might finally give Qualcomm the entry point it has been looking for.

However, the market is already crowded with established players who have been building dedicated AI infrastructure for years, and Qualcomm will need more than ambition to earn meaningful traction.

Amon explained the company is developing “a general purpose CPU” aimed squarely at hyperscalers.

“While we are in the early stages of this expansion, we are engaged with multiple potential customers and are currently in advanced discussions with a leading hyperscaler,” he said, “If successful, we expect revenues to begin in the fiscal 2028 timeframe."

Amon also mentioned other projects, including accelerator cards and even full server racks, additions which are intended to support AI inferencing clusters, signaling Qualcomm is aiming to become more than just a CPU vendor.

Still, questions remain about the timeline and competitiveness of such efforts.

While Qualcomm expects revenue from this move to begin around fiscal 2028, that delay might leave it trailing behind more entrenched rivals like Broadcom, whose own custom accelerator business appears to be scaling aggressively.

Despite the optimistic tone of the announcement, investor reaction was muted.

Qualcomm's stock briefly dipped following the earnings release, suggesting that the market remains cautious about the viability of its expansion plans.

The company’s PC business, while showing modest gains, remains small - its share of premium Windows laptops has reached only nine percent, and there’s little indication yet that it holds the fastest CPU in any major computing category.

Competition also looms back in Qualcomm’s core segment, and its rival, Samsung, has indicated it is preparing to bring its own advanced SoCs into flagship mobile devices by 2026.

This rivalry suggests Qualcomm might be trying to diversify out of necessity, rather than from a position of strength.

For now, Qualcomm’s leap into the workstation and datacenter space is still mostly theoretical.

Whether it can deliver a credible, high-performance CPU and become a serious player in AI infrastructure remains to be seen.

Via The Register

You might also like

• China tried to upgrade the Great Firewall but may have left it vulnerable to attack
• We've rounded up the best monitors for photo editing on the market today
• And here are the best monitors for programming we've found

Here are hints and answers for the NYT Strands puzzle for Aug. 9, No. 524.

Here are some hints and the answers for the NYT Connections puzzle for Aug. 9, #790.

Here are some hints and the answer for today's Wordle for Aug. 9, No. 1,512.

Here are hints and the answers for the NYT Connections: Sports Edition puzzle for Aug. 9, No. 320

• Nvidia chips under investigation as China questions hidden access in H20 hardware
• Beijing summons Nvidia after US proposals ignited fears of remote chip surveillance capability
• TSMC is still building chips for Nvidia, despite regulatory heat and uncertainty in China

Nvidia’s position in the global AI hardware market could soon be under scrutiny following news of an investigation from the Cyberspace Administration of China.

The Chinese regulator has summoned the American chipmaker to explain potential “backdoor” risks in its H20 chips, developed specifically for China after US export restrictions disrupted prior sales of high-end AI processors.

The concern stems from US legislative moves proposing location verification systems on chips intended for export, which Chinese authorities fear could compromise data sovereignty and user privacy.

Mounting suspicion

While Nvidia has firmly denied the existence of any such vulnerabilities, the Chinese government’s decision to interrogate the issue introduces a new layer of uncertainty into the company’s already complex relationship with its second-largest market.

The regulator has not detailed any specific actions it plans to take, but the call to clarify potential security flaws suggests the company’s access to Chinese institutions could face added friction.

Nvidia’s official position has remained consistent: its chips do not contain any embedded features that could allow remote access or control.

In its own words, “Cybersecurity is critically important to us,” and no “backdoors” exist in Nvidia hardware.

However, this reassurance may not be enough to shift growing skepticism, especially as U.S. and Chinese policies around technology exports continue to diverge.

Meanwhile, Chinese analysts have suggested the move could be a political gesture, mirroring concerns the US has raised about Chinese tech in recent years.

What’s notable is that even amid rising tensions, Nvidia continues to see robust demand for the H20 chip within China.

The company has reportedly ordered 300,000 units from TSMC, reflecting the chip’s ongoing relevance to Chinese developers, research institutes, and universities, all of which rely heavily on high-performance AI chips to drive local advancements.

Even military and state-backed projects are known to use Nvidia technology.

Despite public optimism and high-profile visits by Nvidia CEO Jensen Huang to China, the broader regulatory environment is increasingly unpredictable.

The regulator is also looking into the acquisition of Israeli chip designer, Mellanox Technologies, claiming Nvidia violated some of the terms in the 2020 conditional approval of the deal.

Supply chain uncertainty, potential import limits, or changes in licensing rules could eventually impact hardware availability and cost.

As both nations dig deeper into their technology standoffs, Nvidia’s global leadership in AI hardware is no longer guaranteed to go unchallenged.

Via Cybernews and Reuters

You might also like

• A single ransomware attack has pushed this business into insolvency
• Check out the best 3D modeling software for 3D printing and more
• We've rounded up the best portable monitors available now