Artificial intelligence (AI) has supercharged the semiconductor industry over the past three years. In fact, the demand for chips capable of supporting AI systems has been so strong that the PHLX Semiconductor Sector index has clocked a gain of 197% during this period.
That's well above the tech-focused Nasdaq Composite index's gains of 114%. Nvidia was one of the first notable beneficiaries of the AI chip revolution thanks to its graphics processing units (GPUs), which are the leading general-purpose parallel processors. It is still dominating the AI chip market, but there is now an increasing possibility that it could face stiff competition from Broadcom (AVGO 1.24%).
Image source: Getty Images
Broadcom's AI business has gained terrific traction, but the stock's valuation is too rich
Broadcom manufactures custom processors that are proving to be more efficient than GPUs in terms of computing performance and power consumption while handling the specific AI inference tasks they are designed for. As a result, Broadcom has been winning sizable contracts from the likes of OpenAI, and it already boasts a solid customer base that includes major hyperscalers.
The company is on track to end its fiscal 2025 with just under $20 billion in AI revenue, which would be a 63% increase over fiscal 2024. However, analysts are expecting Broadcom's AI revenue to double each year for the next couple of fiscal years. That target seems quite achievable considering that Broadcom reported a revenue backlog of $110 billion as of the end of its third quarter.
NASDAQ: AVGO
Key Data Points
That figure is substantially higher than the $60 billion in revenue it has brought in over the past 12 months. Throw in the lucrative contracts that Broadcom has won recently, and the odds look good for the company to keep growing rapidly for years.
Not surprisingly, Broadcom stock has taken off in the past couple of years, with its shares rising 339% during this period. But if you've missed out on Broadcom's incredible rally thus far and are wary of buying it right now owing to its rich valuation -- the stock trades at 87 times earnings -- it may be a good idea to instead start buying shares of Qualcomm (QCOM 3.58%).
Qualcomm has set its sights on the AI inference market
Qualcomm recently announced its AI200 and AI250 rack-scale data center solutions, which are built to run AI inference workloads in servers. The chip designer points out that these solutions are a part of a "multi-generation data center AI inference roadmap with an annual cadence." In other words, it expects to bring out fresh AI solutions every year.
The AI200 will be commercially available next year, while the AI250 will follow in 2027. Importantly, Qualcomm has landed its first notable customer for its AI chip solutions. Saudi Arabian AI company Humain, which is backed by that country's Public Investment Fund, will start deploying the AI200 and AI250 solutions next year.
Humain intends to deploy Qualcomm's AI chips to power 200 megawatts (MW) of AI inferencing infrastructure in Saudi Arabia. This partnership unlocks a new revenue stream for Qualcomm. That's because building a 100 MW data center could cost around $1 billion. A report from the consultants at McKinsey notes that 60% of the outlay in constructing an AI data center goes toward chips and computing hardware. As a result, Qualcomm could witness an addressable opportunity of $1.2 billion from the Humain contract.
Even better, this partnership could open a much bigger revenue opportunity for Qualcomm in the long run. Humain is aiming to deploy 1.9 gigawatts (GW) of AI data center capacity by 2030, which could go up to 6 GW by 2034. So, Qualcomm could eventually be looking at a multibillion-dollar opportunity in Saudi Arabia alone thanks to its relationship with the state-backed AI company.
NASDAQ: QCOM
Key Data Points
At the same time, investors will do well to note that the demand for AI inferencing is likely to exceed the demand for AI training. Market research firm IDC is expecting investment in AI inference infrastructure to overtake investment in AI training by the end of this year. Meanwhile, Gartner pointed out last year that AI inference computing capacity is likely to grow at an annualized rate of 40% to 50% through 2027.
Considering the tremendous growth potential available, Qualcomm is making the right move by targeting the AI inference market. Shareholders have appreciated this strategy as well, as it could give the company a much-needed growth boost.
In its fiscal 2026, which has just begun, Qualcomm's revenue and earnings are expected to grow by just 2%. That slow pace helps explain why Qualcomm trades at a cheap 16 times earnings and 4 times sales. But if its AI revenue starts taking off on the back of new customer wins following the vote of confidence from Humain, it would be able to clock much stronger growth.
Throw in the company's ability to capitalize on the growing demand for edge AI devices, automotive chips, and the Internet of Things, and there is a good chance that its growth rate will pick up. As such, investors on the lookout for the next Broadcom could consider accumulating a stake in Qualcomm, as its AI-related growth potential could turn out to be a big tailwind for the cheaply valued semiconductor stock.
