Advanced Micro Devices reported a strong third quarter as demand for artificial-intelligence computing helped lift its data-center business to record levels. Reuters reported data-center revenue of $8.55 billion, up 97% from the comparable period a year earlier. Total quarterly revenue increased 45% to $13.38 billion, while adjusted earnings reached $2.09 per share. The company also issued a better-than-expected near-term revenue forecast and an ambitious outlook for 2027, adding to the debate over whether the AI infrastructure boom can sustain its rapid pace.

What AMD actually reported
The important first step is separating quarterly results from projections. Data-center revenue of $8.55 billion and total revenue of $13.38 billion describe the completed third quarter. The outlook for future periods is management's forecast, which can change as demand, supply and customer buying patterns evolve. A strong financial article should label those figures clearly rather than place actuals and estimates together as though they describe the same period.
Reuters reported that AMD's data-center business grew much faster than the company overall. That shows how heavily its growth is currently influenced by customers building AI infrastructure. The segment includes products used by data centers and cloud computing providers, and its performance is therefore particularly relevant to investors tracking demand for increasingly powerful compute systems.
Adjusted earnings per share are another useful measure, but they are not interchangeable with GAAP profit. Companies often report adjusted figures alongside statutory accounting results, excluding specified charges or other items. Readers comparing AMD with another chip company should identify the reporting basis and review the reconciliation rather than assuming that every quoted earnings number follows identical rules.
A 2027 forecast attracts attention
AMD told investors that it expects revenue to grow more than 80% in 2027, according to Reuters. The reported analyst consensus was lower, around 77%. The difference matters because stock prices respond not only to the absolute size of a company's business but also to whether its latest expectations exceed the forecasts embedded in market valuations.
Management also projected approximately $17 billion in revenue for the fourth quarter. Again, that is a forecast, not sales that have already been completed. Executives may have strong visibility into current purchase commitments and product shipment plans, but forecasting demand for fast-changing technologies still involves uncertainty.
Chief executive Lisa Su said she expects data-center products to account for around four-fifths of sales next year. Such a concentration would be a major strategic shift for a company historically identified with consumer processors, gaming products and servers. It could strengthen AMD's position in AI hardware, but it would also leave the business more exposed to changes in a relatively small group of large infrastructure buyers.
Why rack-scale systems are important
The race in AI hardware is no longer only about selling a faster individual chip. Large AI models are trained and operated using groups of processors connected through high-speed memory, networking and power systems. Performance depends on how effectively those pieces operate together, not simply on a benchmark measured on a single component.
Reuters reported that AMD is developing its Helios rack-scale systems to compete more directly with Nvidia's integrated offerings. Initial systems were being tested at Meta, and the company described broader deployment plans. That is commercially important because buyers want machines that can operate reliably at data-center scale and integrate into their existing software and infrastructure.
A new platform's early testing should not be confused with broad production deployment or a guaranteed future market share. The timing of shipments, the cost of installation, software compatibility and the ability to procure components can all affect adoption.
OpenAI and Meta agreements introduce another layer
The company has also built sizable relationships with OpenAI and Meta. Reuters separately reported that AMD took a $1.07 billion noncash accounting charge related to warrants granted to those customers. The warrants give eligible customers the potential to acquire shares when contractual milestones are met. The agreements are connected to large chip-purchase commitments.
That charge is a distinct topic from the quarter's sales growth. It is not accurate to describe the entire $1.07 billion as cash that AMD paid out during the quarter. Accounting treatment for equity-linked incentives can affect reported results even when the underlying transaction is not an immediate cash expense.
Investors must also consider possible dilution if customer warrants vest and are exercised. On the other hand, such arrangements may help secure future demand and deepen long-term commercial partnerships. Whether they represent good economics for current shareholders depends on the value of future purchases, margins and the amount of equity ultimately issued.
Risks hiding behind the strong numbers
AMD is not immune to constraints. Reuters identified a global shortage of high-bandwidth memory, a critical component of modern AI accelerators. A company can have customers willing to buy chips while still facing difficulty delivering enough fully assembled systems to meet demand. The bottleneck may move between memory, packaging, networking and other supplies as the industry expands.
The company also reported weakness outside its data-center business. Reuters said client and gaming revenue fell 8%, and AMD anticipates a double-digit decline in personal-computer shipments next year. That contrast matters because an excellent headline in one segment does not imply that all parts of the company are growing equally.
AMD's competition with Nvidia remains significant. Nvidia has a substantial installed base, established tools and software relationships, and integrated platform products. Higher AMD revenue demonstrates strong demand, but it does not by itself prove that AMD has displaced the market leader.
What analysts and customers should watch next
The next meaningful evidence will come from actual fourth-quarter revenue, gross margins and the pace of data-center product deployment. Analysts will also watch whether planned systems are delivered on schedule and how customers use them in live workloads. Supply constraints and the cost of high-bandwidth memory may be especially important.
For enterprise customers, procurement questions differ slightly from investor questions. Buyers evaluate total system cost, reliability, compatibility and long-term support. The company reporting the fastest growth is not automatically the cheapest or easiest platform for every AI use case.
The third-quarter announcement gives AMD a compelling growth story: record data-center sales, a higher-than-expected outlook and deepening partnerships with major AI buyers. But the responsible interpretation keeps reported earnings, forecasts, warrant accounting and future system deployments distinct. Those distinctions make the story more useful than a simple claim that the AI chip race has already been decided.
Sources
Reuters, AMD quarterly results and 2027 outlook
Reuters, warrants and noncash accounting charge
Smashology Editorial Team
The Smashology Editorial Team is responsible for reviewing and publishing content at Smashology Media, overseeing sourcing, fact-checking, corrections, and editorial standards across coverage of internet culture, technology, entertainment, news, and crime.