Back to feed

Broadcom or Marvell? Valuation and Growth Math in AI Chips

The host compares Broadcom and Marvell shares on sales, profitability, return on capital and valuation, concluding that cheaper Broadcom beats expensive Marvell for the long run.

Imported to Nodesdaily: (UTC+03:00)
Watch on YouTube — 0c-MflvThsA
Reading options

Device speech is unavailable in this browser.

Concept lens

Choose a technical term in this view to read its general definition, teaching example and use in the article.

No terms from our glossary were found in this view. The glossary does not cover every term yet.

Money is pouring into artificial intelligence stocks while demand still runs ahead of supply. The host opens with a clear thesis: the chip shortage stretches years out, and two custom-chip makers stand among its biggest winners, Broadcom and Marvell . Both design alternatives to off-the-shelf Nvidia silicon, yet they play at opposite ends of the stack. The question is as simple as the title: which stock is the better buy for the long run?

Every analysis starts with sales, because without sales there is no business. Broadcom shows off here: annual revenue of around 35 billion dollars in early 2024 has grown to nearly 90 billion over the trailing twelve months. Part of the jump comes from AI demand, part from folding in the 61-billion-dollar VMware acquisition. In its own investor release Broadcom reports 29.6 billion dollars of revenue for the third quarter of fiscal 2026 alone and guides to 34.8 billion for the fourth quarter.

Custom chips versus connectivity: two different fronts

Broadcom's story revolves around custom AI accelerators . Giants such as Alphabet co-develop tailored silicon with Broadcom instead of depending on Nvidia, pouring tens of billions into these programs. According to Anthropic, it agreed with Google and Broadcom on multiple gigawatts of next-generation TPU capacity coming online from 2027; on the OpenAI front a 10-gigawatt custom-accelerator collaboration with Broadcom is underway. Broadcom is setting standards on both the compute and the networking side of AI infrastructure.

Marvell plays the connectivity slice: chips that link and switch every data-center component into one giant working cluster. Its Marvell investor page advertises a mid-September AI Infra Summit showcase of the end-to-end connectivity portfolio, complemented by early-August memory and storage announcements on the same theme. No wonder Nvidia chief Jensen Huang named Marvell a trillion-dollar candidate. Still, the scale gap is enormous: Marvell's market value sits below 300 billion dollars while Broadcom exceeds 1.6 trillion.

Converting sales into profit sharpens the picture further. Broadcom's operating margin has climbed to 48.6 percent, up from about 5 percent in 2017. Marvell earns roughly a third of that at 16.8 percent, though the direction is up. The host concedes the smaller Marvell could expand margins through economies of scale as it grows. At Broadcom, the high-margin software arm offsets the lower-margin chip business and keeps the blended margin at the top.

Valuation: expensive growth or cheap quality?

The third yardstick is return on invested capital , what the company earns on each dollar of assets tied up in the business. Both firms outsource manufacturing, so asset-light models limit capital-intensive risk. Broadcom earns 25.75 percent, more than twice its weighted average cost of capital. Marvell manages 12.61 percent, barely covering its own cost of capital at roughly half of Broadcom's level.

And then the price tag: Marvell trades at a forward price-earnings ratio of 39 against 18 for Broadcom on the same basis, more than twice as expensive. The host's own discounted cash flow model puts Broadcom's fair value at 523 dollars versus a 353-dollar market price, implying about 48 percent upside over 12 to 18 months. For Marvell the math flips: 207 dollars of fair value against a 266-dollar quote, a 22 percent downside risk. The SeekReturns comparison shows the same profitability and valuation gap between the two stocks. Micron's September 30 earnings, formally announced on the Micron investor page, will reveal the true state of memory demand just after the video's release. The host states his position openly: he owns Broadcom shares and none of Marvell.

Visualization: nodesdaily AI
MetricResult
SalesBroadcom ~$90B, Marvell ~$8B
MultipleForward P/E: 18 versus 39
VerdictModel: 48% up vs 22% down
MetricBroadcomMarvell
Sales (trailing 12m)~$90B~$8B
Operating margin48.6%16.8%
Return on capital25.75%12.61%
Forward P/E1839
Market value~$1.6T<$300B

Key moments

  1. Opening thesis: demand ahead of supply, two candidates
  2. Sales check: Broadcom from 35 billion toward 90 billion
  3. Product split: custom accelerators versus data-center links
  4. Profit gap: 48.6% margin and 25.75% return on capital
  5. Valuation: 39 versus 18 multiple and the cash-flow model
  6. Verdict: Broadcom in the portfolio, no Marvell

AI commentary

"The real lesson here is the multiples: Marvell's growth story is already priced in, while Broadcom's quality still trades at a discount. Just remember the host owns Broadcom stock while making that call."

AI assessment

The strongest counterargument for Marvell skeptics is the low-base effect: small sales, thin margins and a return near the cost of capital leave wide room to grow. As AI clusters scale, the connectivity bottleneck deepens, and Marvell's switching and optical portfolio plays directly into it. Even if Huang's trillion-dollar remark sounds stretched, the strategic value of the connectivity layer is hard to deny.

Three caveats apply to the host's math. First, discounted cash flow results are highly sensitive to assumptions; small tweaks to growth or discount rates swing fair value wildly. Second, long-term contracts give visibility into 2027 and 2028, but beyond 2028 the demand picture turns into guesswork. Third, the video carries a sponsorship and the host is a Broadcom shareholder, so his distance from Marvell aligns with his own interest.

The practical takeaway splits by investor type. Patient buyers of quality at a fair price will find Broadcom sturdier, with its strong margin and return on capital. High-risk growth hunters should not rush into Marvell; the 22 percent downside math and a multiple of 39 argue for waiting for a pullback. Either way, holding off on sizing up before Micron's September 30 report is a reasonable excuse to stay cautious.

Sources

6 links; no other published story cites them. Stories sharing a link do not confirm each other; a source's origin is not inferred from how often it is cited.

broadcom · marvell · ai chips · stocks

Follow the topic

Before this story

A short reading order from earlier stories linked to this event by an editor.

Evidence and sources

Review permitted source passages, versions and origins.

KAYNAKLARLA OKU

Bu haberi açalım.

Hesap kontrol ediliyor…