Meet the Artificial Intelligence (AI) Inference Stock That Could Deliver the Biggest Gains Over the Next 3 Years (Hint: It's not Nvidia or Broadcom)

Artificial intelligence (AI) compute workloads in data centers are now shifting toward inference. Deloitte estimates that inference workloads will account for two-thirds of AI-focused computational workloads in data centers this year, up from 50% in 2025.

This has led to a strong jump in demand for chips capable of executing inference workloads cost-effectively. Nvidia currently leads the market for AI inference chips. That isn't surprising, as the company is aggressively reducing the cost of running inference workloads with its chip systems. On the other hand, custom AI chip designer Broadcom is also witnessing phenomenal acceleration in revenue and earnings.

However, both semiconductor stocks have delivered paltry, single-digit gains this year. They have underperformed the PHLX Semiconductor Sector index, which has jumped 86% this year, by a huge margin. Their poor performance can be attributed to their valuation. Broadcom trades at an expensive 61 times earnings and 24 times sales. While Nvidia stock is significantly cheaper at 30 times earnings and looks like a bargain, its sales multiple of 18.6 is on the expensive side.

However, there is another company -- Qualcomm (NASDAQ: QCOM) -- which has started making a dent in the AI chip market thanks to its inference-focused chips and trades at a really attractive valuation right now. Let's see why Qualcomm stock could be one of the biggest winners in the AI inference era, potentially outperforming Nvidia and Broadcom over the next three years.

Image source: The Motley Fool.

AI is poised to significantly boost Qualcomm's data center revenue

Qualcomm currently gets its revenue from selling chips for smartphones, automotive, and the Internet of Things (IoT). These three end markets comprise the company's Qualcomm CDMA Technologies (QCT) semiconductor division. It has another business segment -- Qualcomm Technology Licensing (QTL) -- through which it collects fees and royalties from customers using its intellectual property to design and manufacture wireless products.

The QCT business is Qualcomm's bread and butter, generating 86% of its revenue in the second quarter of fiscal 2026 (which ended March 29). The company's QCT revenue fell 4% year over year in fiscal Q2 to $9.1 billion, primarily due to the weakness in the smartphone market. As a result, Qualcomm's overall revenue fell 2% year over year in fiscal Q2 to $10.6 billion.

The smartphone market is likely to remain constrained by limited memory supply and higher component costs at least until next year. However, demand for chips used to run AI workloads in data centers and edge applications is poised to grow at a healthy pace. Qualcomm has been trying to make a dent in this market for some time, and the latest update from the company suggests that it may have finally made a breakthrough.

At its recently held Investor Day 2026, Qualcomm revealed that it anticipates at least $15 billion in data center revenue by fiscal 2029. That's an impressive start for a company that's making an entry into this fast-growing niche. The company will be offering custom AI processors and inference-first chips aimed at lowering the total cost of ownership (TCO) for data center operators. Even better, the company notes it will upgrade its AI chips annually.

Qualcomm is all set to offer a broad portfolio of AI systems to customers, including liquid-cooled rack-scale servers, custom processors, connectivity solutions, purpose-built data center CPUs (central processing units), and high-bandwidth memory. What's worth noting is that it has already landed a notable customer in the form of Meta Platforms. The tech giant will deploy Qualcomm's Dragonfly C1000 server CPU in its servers starting this year. More importantly, Qualcomm and Meta have a multi-generation agreement, suggesting that the former could witness a solid long-term revenue stream.

Qualcomm CEO Cristiano Amon indicated that Meta isn't the only customer for its AI chips when he pointed out that the company is bringing its "high-performance, low-power computing into the data center, with multi-year, multi-generation agreements with leading customers." So, the company seems well-positioned to achieve its data center revenue growth target over the next three years, which could supercharge its growth.

The solid acceleration in the company's earnings will send the stock soaring

Qualcomm estimates that its earnings per share could exceed $18.00 in fiscal 2029. For comparison, the company's earnings per share are on track to drop by 10% in fiscal 2026 to $10.80. So, Qualcomm's bottom line could increase at an annual rate of 18.5% for the next three years, which seems quite achievable.

Assuming Qualcomm's earnings indeed reach $18.00 per share in fiscal 2029 and it trades at 26.3 times earnings at that time (in line with the Nasdaq-100 index's forward earnings multiple), its stock price could soar to $473. That's a potential upside of 150% from current levels, suggesting that investors should consider buying this AI stock before it steps on the gas.

What's more, Qualcomm is trading at just 17 times forward earnings and 4.6 times sales. So, any step-up in Qualcomm's growth rate could be rewarded with a premium valuation, which could pave the way for greater stock price upside than I have assumed above.

Harsh Chauhan has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Broadcom, Meta Platforms, Nvidia, and Qualcomm. The Motley Fool has a disclosure policy.