AMD boosts AI chip forecast by $1.5 billion – not enough for Wall Street

By Max A. Cherney and Arsheeya Bajwa

(Reuters) – Chipmaker AMD on Tuesday boosted its 2024 forecast for its artificial intelligence (AI) processors by $1.5 billion. Investors wanted more.

AMD’s near doubling of its AI processor projections to $3.5 billion was not enough to meet Wall Street’s lofty projections for the company’s AI fortunes, or assuage the pain of a significant drop in its programmable chip and videogame businesses.

It also forecast first-quarter revenue below Wall Street estimates on Tuesday.

“AMD delivered an underwhelming quarter, with a notable miss on operating income and operating margins standing out the most,” analyst Jesse Cohn said.

“The company had little margin for error amid lofty expectations and investors were disappointed with the forward guidance provided for the current quarter.”

For the current quarter, AMD executives forecast revenue of $5.4 billion, plus or minus $300 million, compared with analysts’ average estimate of $5.73 billion, according to LSEG data. The company did not issue per-share earnings guidance.

Advanced Micro Devices stock has surged roughly 140% in the past year, outpacing the benchmark PHLX Semiconductor Index by nearly 90 percentage points. Shares of the Santa Clara, California-based company fell about 6% in extended trading after the results.

AMD Chief Executive Lisa Su said in the conference call that the company could sell more than the $3.5 billion worth of AI chips that it is now forecasting, once more capacity comes on line in the second half of the year.

But, analysts had previously issued projections for the AI segment ranging from $4 billion to $8 billion, and the stock valuation is pegged to those figures, according to Summit Insights analyst Kinngai Chan.

AMD’s data center segment, which includes its traditional server chips along with its AI processors, grew 38% from a year ago to $2.3 billion but demand from cloud computing companies remained “soft,” Su said in the conference call.

As companies look to develop and operate their own generative AI applications, enterprise budgets are being funneled into processors used in AI servers.

Large and small businesses are looking for alternatives to the advanced AI chips produced by Nvidia, which commands roughly 80% market share. AMD has one of the few viable alternative products in the market.


The market for programmable chips, which can be customized to perform a variety of functions, has faltered in past quarters as industries like automotive and industrial are being hit by a chip supply glut due to weak end-market demand.

AMD’s fourth-quarter embedded segment revenue fell roughly 24% to $1.1 billion.

Rival Intel also flagged corrections in programmable chip inventories across industries on a post-earnings call, which it said were expected to last through the first half of the year.

AMD’s gaming segment shrank 17% to $1.4 billion, as the company has hit its peak revenue from the chips it designs for Microsoft’s Xbox and the Sony PlayStation 5. AMD’s videogame console revenue, which represents a significant portion of the segment that also includes graphics cards for PCs, typically peaks four years after the launch of new systems.

After a pandemic induced trough for PC chip buying through 2023, AMD began to see signs of a return to the once seasonal buying pattern that used to be the norm. The company’s client segment surged 62% to $1.5 billion.

Su said she expected the market to grow “modestly” with more upside in the second half of the year as PCs with artificial intelligence capabilities built in ramp up.

In 2024, worldwide PC shipments are set to grow about 5% following according to data from research firm Canalys.

The company reported overall fourth-quarter revenue of $6.17 billion, slightly above analysts’ estimates of $6.12 billion. Adjusted for one time items such as stock compensation, earnings were 77 cents a share. Wall Street estimated 77 cents a share.

(Reporting by Arsheeya Bajwa in Bengaluru and Max A. Cherney in San Francisco; Editing by Stephen Coates)