Qualcomm just pulled a move that caught Wall Street’s attention. The chipmaker’s stock jumped more than 11% after it revealed two new processors, the AI200 and AI250, aimed squarely at artificial intelligence workloads. It’s a big swing for a company known mostly for smartphone chips, and this time, it’s stepping into territory dominated by Nvidia and AMD.
But Qualcomm isn’t competing on its terms. While Nvidia builds the hardware that trains massive AI models, Qualcomm is focusing on what happens next: inference, the process of running those models once they’re built. That’s where most of the real-world computing happens, and it’s where data center operators spend huge amounts of power and money.
The AI200 is expected to launch in 2026, followed by the AI250 in 2027, which promises 10 times the memory bandwidth of its predecessor. Qualcomm has already lined up its first major customer, Saudi Arabia’s Humain, which committed to 200 megawatts of compute capacity. The company is also betting on efficiency as its key advantage. Each rack draws about 160 kilowatts of power, a figure comparable to Nvidia’s systems, but Qualcomm says it can offer better performance for the same cost.
How AI infrastructure demand is driving the global semiconductor industry
A relentless appetite for AI infrastructure is driving an increase in worldwide semiconductor revenue growth rate.
For Qualcomm, this marks a second attempt to enter the data center market. Its first effort, back in 2017, failed to gain traction. This time, the company is entering a market where customers are desperate for alternatives. High energy costs, GPU shortages, and mounting demand for inference have created an opening that Qualcomm believes it can fill.
That timing is also hard to ignore. Just three weeks ago, OpenAI signed a multibillion-dollar deal with AMD that could give it up to 10% ownership through stock warrants. Nvidia also invested around $100 billion in OpenAI and secured a large supply agreement. These overlapping partnerships show how deeply interconnected the AI hardware industry has become, and how little room there is for stagnation.
Inside OpenAI’s trillion-dollar web of deals keeping the AI industry afloat
From Nvidia to AMD, a handful of tech giants are fueling each other’s growth in a cycle that could either sustain the AI boom or break it.
Outside of AI, the chip industry is seeing a similar reshuffling. AMD now holds about 42% of the gaming CPU market on Steam and close to 40% of the server market after starting from nearly zero in 2017 (via overclock3d report). Nvidia, though, still dominates AI training with more than 90% market share and a market value of $4.5 trillion, but tech giants such as Google, Amazon, and Microsoft are designing their own chips to reduce reliance on outside suppliers.
What this all signals is a shift away from the era of one company dominating an entire segment. The industry is fragmenting into specialized players, each competing through speed, efficiency, and partnerships. For Qualcomm, that might be the best opportunity it has had in years.
The company doesn’t need to dethrone Nvidia to succeed. If it can capture even a small share of the growing inference market, it could move from being a phone chip supplier to a serious player in the global AI infrastructure space.
