A deal potentially worth $120 billion would normally be the kind of announcement investors celebrate.
For Marvell Technology (MRVL), it produced the opposite reaction.
Shares of the chip designer dropped more than 8% to $221.60 in early trading Aug. 28, putting the company on pace to lose more than $17.4 billion in market value.
The company performed well. Management raised its revenue forecast, according to Reuters. Analysts lifted price targets. Marvell has one of the biggest new artificial intelligence opportunities in the semiconductor industry in a custom-chip deal with Alphabet (GOOGL) subsidiary Google.
But Wall Street dumped the shares.
The reason provides a useful lesson for ordinary investors chasing the AI boom: A giant contract is not the same thing as giant revenue today.
The Google opportunity will be much larger in FY 2029, Marvell CEO Matt Murphy said in a post-earnings call. Marvell’s previous guidance through fiscal 2028 also included some of the Google-related revenue.
So investors who had expected the $120 billion headline to immediately translate into dramatically higher forecasts found that part of the payoff is years away.
And when a stock trades at nearly 59 times forward earnings, waiting becomes expensive.
Marvell’s $120B number has a timing problem
Marvell has emerged as one of Wall Street’s biggest beneficiaries of the rush to build custom AI chips.
Big tech companies are increasingly designing their own custom semiconductors to boost performance and cut the enormous costs of AI computing.
Reuters reported that Big Tech’s AI spending is expected to exceed $740 billion this year. That spending surge helped Marvell shares nearly triple in 2026 before the latest selloff. Those gains also altered investors’ expectations for the company.
Exceeding expectations simply wasn’t enough.
The new custom-chip arrangement for Marvell with Google could generate as much as $120 billion in revenue through fiscal 2033, Reuters reported. That sounds like a huge amount compared to Marvell’s current business.
Related: Buffett’s Berkshire is doubling down on Google
But Murphy told investors that the company’s existing custom revenue targets through fiscal 2028 already included some Google revenue.The much bigger contribution is expected to begin in fiscal 2029.
That was the key difference in the post-earnings debate.
Expectations were higher mostly because of the Google deal, Morgan Stanley analysts said, and its contribution was largely already priced into earlier company guidance. Thus, Wall Street wasn’t discovering a whole new $120 billion opportunity after earnings.
Investors were getting additional information about when a known opportunity would actually be reflected on the financial statements.
Marvell is growing fast, but Wall Street wanted faster
The selloff is even more striking considering that Marvell’s underlying growth outlook is not weak at all.
The company expects revenue growth of about 45% for fiscal 2027.
Revenue is projected to reach around $18 billion in fiscal 2028, aided by continued growth in its data-center business.
Those are big numbers for many semiconductor companies. But for Marvell, they ran counter to expectations that had grown even faster than the business itself.
That’s an increasingly important distinction for the AI business.
Investors aren’t asking if AI semiconductor companies will grow anymore. They are asking themselves whether those companies can grow faster than the assumptions already baked into their share prices.
More Tech:
- Anthropic-powered AI model sends shocking message to employee
- Rocket Lab clears 1st hurdle in its biggest satellite deal
- Apple’s $54 billion iPhone machine may be about to break its biggest ritual
That is a particularly high bar given the valuation of Marvell.
Marvell was trading at 58.41 times its forward earnings. Competitor Broadcom (AVGO) trades at 32.15x. That means Marvell trades at an earnings multiple some 82% above Broadcom’s forward multiple.
Investors could justify such a premium if they believe Marvell’s future earnings will grow much faster.
But it also allows for less disappointment.
When a stock trades at nearly 60 times earnings, a good quarter can still be a bad day if the next wave of profits is later than investors expected.
Investors can learn something from Marvell’s plunge
The lesson here extends well beyond semiconductor stocks. Suppose a company announces a gigantic contract. The first number investors naturally focus on is the total potential value. With Marvell, that number is up to $120 billion.
But there are several other questions that matter just as much:
- How many years will the revenue be spread across?
- How much was already included in Wall Street forecasts?
- When does the largest contribution actually begin?
- What profit margins will the work generate?
- How expensive is the stock before the new revenue arrives?
Marvell’s selloff is a case study in what happens when investors pay a lot of attention to the first number and then get less exciting answers to the rest. That’s a far cry from a $120 billion opportunity that extends through fiscal 2033 to $120 billion of incremental revenue coming in over the next several quarters.
And management has now said that the Google business doesn’t become meaningfully more important until fiscal 2029.
That doesn’t necessarily make the opportunity any less valuable for long term holders. But for a market that has nearly tripled Marvell’s stock this year, it alters the timing of the payout.
Analysts still see a far bigger earnings opportunity
What’s remarkable about the selloff is that Wall Street isn’t suddenly giving up on Marvell.
At least eight brokerages raised their price targets after the results, Reuters reported.
The median analyst target increased to $275. That was around a 13.8% gain from Marvell’s previous close.
Melius Research was also bullish on the longer-term opportunity, according to Reuters.
Marvell’s analysts said its custom-chip business, potential business with Microsoft, and further growth in AI connectivity could eventually result in a big expansion of earnings. They wrote, “$20 in EPS power before the end of the decade look[s] realistic.”
That figure offers another way for investors to see why Marvell merits such a lofty valuation.
The market is not necessarily valuing the company on what it makes today. It’s valuing Marvell at what its AI business could look like in a few years. That can lead to explosive gains when expectations rise. Even a slight change in the timeline can also result in violent sell-offs.
Bloomberg / Getty Images
The AI boom has created an expectations problem
A bigger problem is surfacing in the semiconductor industry.
Hundreds of billions of dollars are being invested in AI infrastructure by the big tech companies.
Chip designers relying on that spending are posting huge growth rates. And investors have rewarded some of those businesses with fantastic valuations. But eventually those stocks run into a mathematical problem. The better the story, the more future success is baked into the shares.
Marvell’s stock has more than tripled before falling Friday, according to Reuters. Revenue is expected to increase by 45% in fiscal 2027. It pulls in roughly $18 billion in revenue in fiscal 2028. It has a $120 billion Google opportunity through fiscal 2033.
Those aren’t the numbers of a company that has lost its AI chance. Still, the stock plunged more than 8%. That raises serious concerns for investors about the state of the AI trade. Wall Street increasingly wants growth, and growth sooner than expected.
Marvell’s Google deal may still pay off, just not fast enough for everyone
So it’s the fiscal 2029 timetable that investors might want to pay closer attention to than the $120 billion headline.
If Google’s contribution lives up to Marvell’s expectations, the company may eventually have an excuse to explain much of the optimism already priced into the stock.
Melius Research thinks the Google relationship, Microsoft prospects, and connectivity business could produce roughly $20 in earnings per share by the end of the decade.
This wave of price-target increases shows analysts remain broadly constructive.
But Marvell’s selloff reveals the flipside of owning an expensive AI stock.
A company can win a giant customer. It can raise its forecasts. It can project 45% annual revenue growth.
And Wall Street can still wipe out more than $17 billion of its market value because investors had expected the good news sooner.
Perhaps that’s the most valuable number in Marvell’s earnings story for the average investor.
The company didn’t lose its Google opportunity in a day.
Investors just found out some of the $120 billion future they thought they were buying is further away than they thought.
Related: Ackman just walked away from Google
