AMD’s data-center priority leaves Radeon supply questions open | Magica
AMD puts data centers first, but gaming’s decline tells only part of the story
Editorial Team
••📖6 min read
AMD chief Lisa Su calls the chipmaker a data-center-first company as server and AI sales surge. Gaming revenue fell 40%, but PC-chip sales grew, and her remarks leave the supply outlook for expensive Radeon cards unresolved.
AMD expects data-center revenue to grow 50% from Q3 to Q4 2026.
Gaming revenue fell 40% from a year earlier, while PC-chip revenue rose 7%.
Reported Radeon prices remain elevated; the interview announces no consumer GPU production cut or price-relief timetable.
Advanced Micro Devices (AMD), the chip designer behind Ryzen PC processors and Radeon graphics cards, is putting data centers first. Chair and chief executive Lisa Su, who became CEO in October 2014 after serving as chief operating officer overseeing product strategy and execution, sales and operations, set out that priority in a November 5 interview. Her appointment announcement documents that earlier remit.
Su said in the interview that AMD’s data-center business would grow 50% sequentially between the third and fourth quarters of 2026. She also described an ambition to more than double the business heading into 2027, while cautioning that AMD must assess progress quarter by quarter. The discussion concerned data centers, rather than a promise to double the entire company.
The stakes reach beyond a corporate slogan. Data centers now generate nearly two-thirds of AMD’s revenue, giving the company a powerful incentive to serve large computing customers. But expensive gaming cards and falling gaming revenue do not, by themselves, show that AMD has diverted a specific amount of consumer production to AI.
AMD’s company-reported revenue by business, Q3 2025 versus Q3 2026. Gaming includes discrete GPUs and semi-custom products and services. Source: AMD Q3 2026 results.
Data centers bring scale and higher segment margins
AMD’s November 3 results show where the growth is coming from. Its EPYC server processors and Instinct accelerators—GPUs used for AI computing—drove data-center revenue to $8.55 billion in Q3, up 97% from a year earlier. Total company revenue grew 45%, not 97%, to $13.38 billion.
A federal judge found Anthropic’s Claude Max usage advertising plausibly misleading because subscribers could read the multipliers as monthly promises. Contract and misrepresentation claims survive, but several other claims were dismissed with leave to amend.
Editorial Team
The distinction corrects an error in coverage of the interview that attached the data-center growth rate to company-wide sales.
AMD revenue
Q3 2025
Q3 2026
Year-over-year change
Total company
$9.25 billion
$13.38 billion
+45%
Data Center
$4.34 billion
$8.55 billion
+97%
Client: PC chips
$2.75 billion
$2.95 billion
+7%
Gaming
$1.30 billion
$773 million
−40%
Figures are rounded from AMD’s tables. Data centers accounted for about 64% of Q3 revenue, calculated from $8.546 billion divided by $13.377 billion. AMD’s 2024 results already showed an established shift: annual data-center revenue rose 94% to $12.6 billion, roughly 49% of $25.8 billion in company sales. Those shares describe the revenue mix over different reporting windows, not comparable annual growth rates.
Profitability reinforces the incentive. In Q3 2026, AMD reported $2.88 billion in Data Center operating income, versus $573 million for the combined Client and Gaming segment. Dividing each by its segment revenue gives operating margins of about 34% and 15%, respectively. These are segment measures, before expenses AMD assigns to its “All Other” category; they do not reveal the margin on an individual Radeon card or AI accelerator.
Gaming’s fall does not describe the whole PC business
AMD attributes the 40% gaming decline to lower semi-custom and graphics revenue. Gaming includes both discrete GPUs and semi-custom system-on-chip products and development services. The combined figure cannot establish a 40% fall in standalone Radeon sales or measure manufacturing capacity transferred to AI products. The 2024 results also show that gaming’s weakness predates this interview: annual gaming revenue fell 58% that year, primarily because of lower semi-custom revenue.
The PC business tells a different story. AMD said the Client business’s 7% revenue growth was driven by Ryzen mobile processors. Its latest results also list expanded availability of systems using Ryzen AI Max PRO 400 processors and the availability of its Ryzen AI Halo platform for running models locally.
A report questioning AMD’s consumer priorities points to the absence of a new Ryzen CPU or Radeon GPU generation during 2026 by early November. That criticism concerns generational cadence; it should not obscure the PC products and system availability AMD lists in its results. Neither a gap between generations nor growing PC-chip sales settles how much AMD will invest in the next gaming graphics lineup.
Radeon remains a cheaper alternative, at elevated prices
A retail-price report places the Radeon RX 9070 XT above $900, about $300 above its launch price, and the RX 9070 around $750, about $200 above its suggested price. It also puts the 16GB RX 9060 XT above $700 and even 8GB cards above $500, extending the concern beyond the top models.
AMD nevertheless retains a relative price advantage in that report: it places the RX 9070 XT about $500 below Nvidia’s competing RTX 5070 Ti. Switching to Nvidia, the rival graphics-chip supplier, therefore does not necessarily solve the affordability problem. These are reported retail snapshots, not AMD list-price announcements, a comprehensive market index or a like-for-like performance test.
The supply mechanism remains less certain than the prices. Su said AMD has resources to expand supply and serve customers, complicating a simple account of a fixed pool being redirected from gamers. AMD’s earnings release identifies dependence on third-party manufacturers, memory and other components, distributors and add-in-board partners as risks. Those dependencies matter to delivery, but the disclosures do not quantify which bottleneck caused a particular card’s markup or establish a Radeon allocation cut.
AMD’s company-reported operating cash flow from continuing operations and free cash flow. These company-wide figures do not isolate AI supply costs. Source: AMD Q3 2026 results.
The next test is delivery, availability and cash
AMD’s immediate target is the 50% sequential data-center growth Su forecast for Q4. Separately, its earnings release projects total quarterly revenue of about $17 billion, plus or minus 3%, with the midpoint representing approximately 27% sequential growth. The different forecasts should not be conflated.
Expansion also has a cash cost. AMD’s Q3 operating cash flow from continuing operations was $1.20 billion, down from $1.79 billion a year earlier. Its defined free cash flow—operating cash flow from continuing operations less capital expenditures—fell to $554 million from $1.53 billion. The quarter’s cash-flow statement records a $3.0 billion inventory increase absorbing cash, compared with $636 million a year earlier. These company-wide figures do not isolate the cost of supplying AI customers, but they show why revenue growth alone is an incomplete test of execution.
Asked about investor concerns over AI infrastructure spending, Su defended investing three to five years ahead and said she wanted AMD to be the best partner to the largest computing users. That is the conviction of a supplier with a financial stake in continued spending; it does not establish the eventual returns for its customers.
For Radeon buyers, the evidence needed is more specific: shipment volumes, the next graphics generation and sustained retail availability closer to suggested prices. AMD’s data-center results will test its growth forecast. Neither forecast answers when Radeon cards will become easier to buy at their suggested prices.
Thailand’s proposed data-centre framework would scale requirements by power demand, screen sites for electricity and water access, and assess domestic benefits. It also covers existing facilities, while conflicting application and approval schedules remain unresolved.
PJM’s independent market monitor attributes about $11 per megawatt-hour of first-half 2026 wholesale costs to data-center demand in capacity auctions. Energy costs drove more of the overall increase, while construction barriers and an October regulatory rejection complicate the response.
DataVita now says its proposed Lanarkshire AI data centres will be fully grid served, with direct renewable connections only a future possibility. The change in public presentation raises questions about promised power costs and community benefits, although the DV4 planning application already assumed grid supply.
A reported U.S. intelligence assessment links Megaspeed executive Alice Huang to China’s Ministry of State Security, raising questions about access to Nvidia’s leadership and chip-export policy. Whether the alleged operation succeeded remains unclear.
Ocient’s $100 million Series C backs tailored AI analytics on an established database platform, with national-security investors and deployment economics shaping its expansion.
Anthropic plans IPO marketing next week and a listing in the week of November 16, according to people familiar with the plans. Buyers are discussing valuations of $1.2 trillion–$1.3 trillion, with some weighing $1.1 trillion, as its filing reveals faster-growing compute costs and extensive infrastructure commitments.
India's IT lobby chief warns that AI could make technology jobs harder for graduates to secure. Hiring is below earlier peaks, but some firms favour AI-equipped beginners over costly experienced staff. The unresolved question is how junior workers will build expertise.
Internet scans and leaked databases expose the spread of ARTEX, a security-testing framework used in cyberattacks. Researchers documented largely autonomous compromises, but the records also show failed cashouts and uncertain counts of independent deployments.
Meta's strategy chief and former CFO is joining Fireworks AI ahead of a reported fundraising push targeting at least a $30 billion valuation. The inference company announced a $17.5 billion valuation in July, making the hire part of a much larger expansion than its older funding figures suggest.
Sand.ai says its $200 million Series A+ round lifts its valuation above $1 billion. Co-lead investor Wangsu plans inference and deployment cooperation, while its filing shows substantial losses behind the startup’s video AI ambitions.
An advocacy group estimates Interior Secretary Doug Burgum’s AI holdings at $18.1 million–$64.4 million as his department advances data centers on public land. His ethics agreement allows retained investments with recusals, while pending permits will determine both the application of those safeguards and the public’s return. Interior denies his decisions are driven by personal investments.
Microsoft is evaluating fewer neocloud partners while expanding its own infrastructure. Delivery-dependent contracts underpin supplier financing, while IREN reports rising revenue alongside a $6 billion funding target for its broader expansion.
Physical Superintelligence says its AI system found a proof of a conjecture limiting quantum query advantages on random inputs. Public proof files enable scrutiny, but do not establish the private system’s cost or repeatability.
OpenAI is expanding its relationship with Hacktron, whose researchers previously reported an attack chain into employee accounts, to test third-party connections with greater access to cybersecurity models.
Google has acquired Crusoe’s southeast Wyoming data center site, where it plans a long-term $17 billion investment. Crusoe describes completed pre-development work; independent reporting points to customer setbacks. Power delivery and carbon capture remain key tests.
The team behind document-parsing service SoMark has released OXR-1.0 code and weights. A benchmark retest supports a strong result, but speed and accuracy were measured separately, and users above funding or revenue thresholds need a commercial license.
OpenAI is reportedly developing a Business Hub with daily mention and click charts, product updates and business agents. The unannounced feature could give verified businesses direct visibility into ChatGPT responses, without revealing prompts or competitor rankings.
Palantir and Nvidia have specified an integrated stack for running AI in customer-controlled environments. The architecture builds on existing software and infrastructure offerings, with published configurations spanning 32 to 640 GPUs, while pricing and workload economics remain unproven in the launch materials.
Democrats’ incoming congressional majorities put hearings and investigations within reach, but AI rules and crypto legislation still face bipartisan bargaining and a Republican president.
