AMD CEO Lisa Su went on damage control Wednesday after the chipmaker's stock tanked 13%, telling CNBC that AI demand is accelerating faster than she imagined and data center business has stepped up significantly from Q4 to Q1. The remarks come just a day after AMD posted solid earnings but disappointed investors with what many saw as conservative guidance, sending shares tumbling despite the company projecting $9.8 billion in Q1 revenue above Wall Street's $9.38 billion estimate.
AMD CEO Lisa Su spent Wednesday afternoon firefighting investor concerns after the company's stock got hammered despite beating earnings expectations. The chipmaker's shares plunged 13% following Tuesday's fourth-quarter results, with Wall Street fixating on what it saw as underwhelming guidance rather than the solid numbers AMD actually delivered.
"What I would tell you from someone on the inside is AI is accelerating at a pace that I would not have imagined," Su told CNBC's Squawk on the Street on Wednesday. She stressed that demand continues to outstrip compute supply, and that AMD has seen a meaningful step up in orders over the past two to three months.
The core issue? AMD guided for $9.8 billion in Q1 revenue, plus or minus $300 million. That's actually above the Street's $9.38 billion consensus, but analysts had been pricing in a far more explosive quarter given the AI spending frenzy sweeping the industry. Nvidia has been posting blowout numbers quarter after quarter, and AMD's relative conservatism left some wondering if the company is losing ground in the AI chip wars.
Su pushed back hard on that narrative. She said AMD's data center business has actually accelerated from the fourth to first quarter, and that demand for the company's central processing units is "going gangbusters" as businesses rapidly scale up compute infrastructure for AI enterprise workloads. That's a crucial distinction - while Nvidia dominates GPU-based AI training, AMD is making inroads on the CPU side where enterprises need balanced systems.
The disconnect between AMD's actual guidance beat and investor disappointment traces back to sky-high expectations set in Q4. The company announced major partnerships with OpenAI and Oracle last quarter, with Oracle committing to deploy 50,000 AMD AI chips. Those megadeals led analysts to model aggressive near-term revenue contributions that AMD's guidance didn't quite deliver.
But Su's comments suggest AMD is playing a longer game. She told CNBC the company will hit an "inflection point" in the second half of 2026 as it begins shipping Helios, its new integrated server-scale AI system designed to compete more directly with Nvidia's full-stack offerings. Getting Helios out the door on schedule could be the catalyst that finally convinces investors AMD is a legitimate alternative to Nvidia for large-scale AI deployments.
The timing of Su's media blitz matters. AMD's stock had already been under pressure heading into earnings, with shares down about 8% year-to-date as investors questioned whether the company could maintain momentum against Nvidia's relentless execution. Tuesday's 13% drop represented a crisis moment that required immediate executive intervention to prevent further erosion in confidence.
Wall Street's reaction also reflects broader nervousness about AI infrastructure spending sustainability. After two years of exponential growth in data center buildouts, analysts are starting to ask when the spending will normalize. AMD's cautious Q1 guidance, even if it beats consensus, feeds into fears that the AI gold rush might be moderating faster than bulls anticipated.
Su's emphasis on sequential quarterly acceleration from Q4 to Q1 is her strongest counterargument. If AMD's data center revenue is actually growing quarter-over-quarter in an environment where some expected seasonal softness, that suggests underlying demand remains robust. The "going gangbusters" characterization of CPU demand isn't typical CEO speak - it's Su signaling she sees concrete order flow, not just pipeline speculation.
The question now is whether investors will buy Su's narrative or keep punishing AMD for not delivering Nvidia-style blowouts. The company's historical challenge has been converting partnerships into revenue fast enough to satisfy Wall Street's impatience. Those OpenAI and Oracle deals are real, but the revenue recognition timelines don't match the market's desire for immediate gratification.
AMD also faces execution risk on Helios. The company has a history of strong product roadmaps but occasionally stumbling on launch timing or supply ramp. If Helios slips or doesn't deliver the performance enterprises need, AMD's window to capitalize on AI infrastructure spending could narrow as Nvidia and emerging competitors like custom silicon from hyperscalers eat into the addressable market.
What's clear from Su's Wednesday interviews is AMD isn't conceding the AI chip race despite the market's skepticism. The CEO's willingness to go on camera hours after a brutal selloff shows she believes the guidance controversy is a communications problem, not a fundamental business issue. Whether she's right will become apparent when AMD reports Q1 results and investors can compare actual performance against her accelerating demand claims.
AMD's post-earnings damage control reflects the brutal calculus facing AI chip challengers in an Nvidia-dominated market. Su's insistence that data center demand is accelerating sequentially gives bulls a narrative to cling to, but the 13% stock drop shows investors want proof, not promises. The real test comes in six months when Helios ships and AMD either validates Su's confidence or proves the market's skepticism was warranted. Until then, AMD remains caught between solid execution that beats estimates and sky-high expectations that assume every AI infrastructure dollar flows equally to all chip vendors. Su's betting that enterprise customers will diversify their AI chip purchases beyond Nvidia as workloads mature - a reasonable thesis that just needs time and flawless execution to play out.
