Custom artificial intelligence (AI) chips are a lucrative business; just ask Broadcom (AVGO 2.66%) and Marvell Technology (MRVL 2.87%). As two of the industry’s major competitors, they continue to impress with their eye-popping earnings. Broadcom and Marvell both design application-specific chips that are increasingly preferred over general graphics processing units (GPUs) in the new era of AI workloads.
As sales continue to surge for both companies, the question for investors is which stock offers greater upside going forward?
Image source: Getty Images.
Broadcom, the blue chip behemoth
Broadcom is undoubtedly the market leader in custom AI accelerators. It commands more than 70% market share in that area alone. Its customers are the world’s biggest hyperscalers, including Alphabet, Meta Platforms, OpenAI, Anthropic, and more.
The numbers Broadcom has posted over the past year are simply staggering, and 2026 is shaping up to be no different. For the first quarter of 2026, Broadcom’s revenue exceeded $19 billion, a 29% increase from the same period last year. Most impressively, AI semiconductor revenue grew a whopping 106% in the first quarter.
Today’s Change
(-2.66%) $-8.23
Current Price
$301.18
Key Data Points
Market Cap
$1.4T
Day’s Range
$298.88 – $307.50
52wk Range
$138.10 – $414.61
Volume
5.8K
Avg Vol
26M
Gross Margin
64.96%
Dividend Yield
0.82%
Broadcom’s second-quarter guidance was just as astonishing, as the semiconductor giant announced revenue expectations of $22 billion. This would be a 47% increase from 2025.
The stock is actually down more than 7% since the start of 2026. This recent drop has made Broadcom’s valuation more attractive, as its forward price-to-earnings (P/E) ratio is now below 30. The company also pays a quarterly dividend worth $0.65 per share.
It’s unlikely that Broadcom’s massive growth will last forever, but the end doesn’t yet seem to be in sight. The company’s financials and outlook remain strong, and its competitive moat is enviable.
Marvell looks marvelous
Don’t overlook Marvell. The company is smaller but growing fast and furiously. Its stock is also more attractively priced than Broadcom’s. Marvell delivered record revenue in fiscal 2026, reporting nearly $8.2 billion, a 42% increase from the previous year. Its earnings per share rose an incredible 81% as well.
Marvell has big ambitions. The company’s CEO and Chairman, Matt Murphy, said the business expects year-over-year revenue growth to accelerate in each quarter of fiscal year 2027. All told, Marvell anticipates growth of around 30%. Demand from data centers and recent acquisitions should keep the company’s revenue growing quickly.
Marvell’s biggest customer is Amazon‘s (AMZN 3.89%) cloud computing arm, Amazon Web Services (AWS). This poses a somewhat elevated risk of concentration for Marvell, and the company needs to diversify its customer base going forward.
Today’s Change
(-2.87%) $-2.80
Current Price
$94.88
Key Data Points
Market Cap
$83B
Day’s Range
$93.40 – $99.63
52wk Range
$47.09 – $102.77
Volume
253K
Avg Vol
17M
Gross Margin
50.10%
Dividend Yield
0.25%
As Marvell grows, it has a real opportunity to capture a bigger piece of the market. The company is targeting 20% market share, according to Murphy.
Marvell’s stock has risen by more than 15% year to date. As for tech stocks, the company’s valuation metrics remain compelling. The trailing P/E ratio is 28, and the company’s price/earnings-to-growth (PEG ratio) ratio is right around 1. This means the stock is fairly priced, even with its recent run.
Which is the better stock?
The better stock depends on who you are as an investor. Broadcom is going to look more like a blue chip company compared to Marvell. Broadcom’s sheer size and market scale will keep it the dominant player for years to come.
Marvell, however, could have substantially more upside if it executes its growth plan and expands its market share. The current price of Marvell is also more appealing than Broadcom’s. Broadcom is a much safer investment than Marvell, but Marvell is a bolder one.
Both stocks are subject to the same types of risk and would be greatly affected by any slowdown in AI infrastructure spending. Google’s TurboQuant (a compression method that reduces the amount of memory required to run large language models) could be a threat, but not likely in the near-term.
