A shift is happening in the artificial intelligence (AI) computing space. Hyperscalers and AI companies are transitioning from training large language models to putting those models to work on real-world data to unlock the productivity gains that this technology promises.
A lot of money has been spent on data center infrastructure that's optimized to train models using large datasets. Inference, on the other hand, doesn't require as much parallel-processing power as the training phase, as it needs to act on specific inputs to respond to queries in real time. However, Deloitte estimates that rising inference workloads will create demand for more computing power.
The consulting giant estimates that inference workloads will consume two-thirds of all AI computing power in 2026, up from 50% last year. This will create a need for more AI chips of the varieties that are best for inference. That's great news for Intel (INTC +4.36%) investors. In fact, it won't be surprising to see Intel eclipsing the likes of Nvidia and Broadcom (AVGO 0.52%) in the AI inference era.
Image source: Intel
AI inference demand has given Intel a big shot in the arm
The different computing demands of AI inference explain the strong demand for chips designed specifically for these tasks. Application-specific integrated circuits (ASICs), for instance, have been in healthy demand from hyperscalers and AI companies because of the advantages they can offer over general-purpose parallel processors such as graphics processing units.
NASDAQ: INTC
Key Data Points
ASICs are more power efficient, and they help reduce overall computational costs when they are handling the highly specific tasks that they are designed to. This explains why Broadcom, the biggest player in the ASIC market (with an estimated 60% market share) is forecasting a huge increase in its AI revenue.
The company is confident it will exceed $100 billion in AI chip revenue in 2027, a huge jump from last year's $20 billion. Even Nvidia is gearing up its upcoming Vera Rubin processors for inference-focused tasks, promising significant reductions in inference costs. However, I believe that investing in Intel could be one of the best ways to capitalize on the growing demand for AI inference chips.
That's because the company is the dominant player in the server central processing unit (CPU) market, with a share of just over 71%. Moreover, Intel is cutting its teeth in the ASIC market as well. Its ASIC revenuealmost doubled year over year in the first quarter, while sequential growth was 30%.
For comparison, Intel's custom ASIC revenue was up by 50% year over year in the fourth quarter of 2025, while sequential growth was 26%. Clearly, Intel has seen a significant acceleration in its ASIC sales lately, and the business now has an annual run rate of more than $1 billion. It won't be surprising to see further acceleration in this business, as hyperscalers and chip companies have lately signed multiple contracts for Intel's custom AI chips and server CPUs.
Alphabet's Google, for instance, has a multiyear contract with Intel to deploy Intel's ASICs and Xeon server CPUs. Even Nvidia is integrating Intel's Xeon server CPUs into its Rubin rack-scale server systems. In all, AI inference seems to be heralding a new era of growth for Intel.
CEO Lip-Bu Tan's remarks on the latest earnings call make it clear why the company is well-placed for further growth in the inference era. He noted:
Artificial intelligence is now moving into the real world toward more distributed inference and reinforcement learning workloads, like agentic, physical AI, robots, and edge AI.
This shift is now beginning to show up in our results, and I want to spend some time on this today. For the last few years, the story around high performance computing was almost exclusively about GPUs and other accelerators. In recent months, we have seen clear signs that the CPU is reinserting itself as the indispensable foundation of the AI era.
Market research company Futurum estimates that the market for CPUs in AI data centers could increase at an annualized rate of 28% through 2028, though it wouldn't be surprising to see their sales increase at a faster pace because of the boom in AI inference demand.
Chipzilla's improving growth prospects point toward a bright future
Intel started turning its business around last year, and those efforts are bearing fruit in 2026. While the company's revenue remained almost flat at $53 billion in 2025, analysts anticipate consistent double-digit percentage growth over the next three years.
INTC Revenue Estimates for Current Fiscal Year data by YCharts.
However, Intel can surpass those expectations, driven by the company's focus on enhancing its manufacturing capacity, boosting production yields of its advanced process nodes, and the lucrative partnerships it is striking in the AI chip space. Stronger-than-expected growth could lead investors to reward Intel stock with a premium valuation, as compared to the 8.7 times sales it trades at right now.
That sales multiple is slightly higher than the U.S. tech sector's average of 7, but it isn't all that expensive. Assuming Intel trades at 10 times sales and achieves $71 billion in revenue in 2028, its market cap would reach $710 billion. That would be an upside of 48% from current levels. Don't be surprised to see this AI stock deliver bigger gains, however, as the demand for its chips is exceeding supply, suggesting that it could indeed grow at a faster pace than analysts are anticipating.
