Prediction: Data Center Passes 70% of AMD's Revenue in 2027, Before the Helios Ramp Is Finished

AMD's data center revenue reached $6.7 billion, or 58% of total Q2 sales, after growing 107% year over year versus roughly 8% growth across its other businesses. Q3 guidance of about $13 billion implies data center could account for roughly 63% of revenue if non-data-center sales remain near $4.8 billion, while management expects the segment to more than double in 2027 as Helios AI systems ramp for OpenAI, Meta and Anthropic. The article projects data center could surpass 70% of AMD revenue by mid-2027, but flags Helios execution risk and a roughly 30x 2027 earnings valuation that already assumes a smooth ramp.

Analysis

AMD’s key investable question is no longer AI exposure but whether rack-scale systems convert into incremental gross-profit dollars rather than lower-margin, capital-intensive integration revenue. A larger systems mix raises exposure to HBM availability, advanced packaging yields, networking, power delivery and customer acceptance testing; any bottleneck can defer revenue recognition even if end-customer demand is intact. The critical proof point over the next 1-3 quarters is data-center gross margin and cash conversion, not segment mix.

The customer concentration implicit in a few hyperscale deployments creates asymmetric quarterly risk. Large installations are lumpy, and customer qualification delays or changes in cluster architecture can push revenue across reporting periods while leaving headline demand unchanged; at a valuation already discounting a clean multi-year ramp, that timing risk can compress the multiple quickly. NVDA remains the primary competitive hedge because software maturity, interconnect performance and delivered cluster uptime—not accelerator specifications—will determine whether AMD captures durable share.

Contrarian view: a faster mix shift is not automatically bullish if systems revenue carries materially lower gross margin or requires elevated inventory and receivables. Conversely, evidence that AMD can sustain data-center growth while expanding consolidated gross margin would force estimates higher and could validate a share-gain narrative versus NVDA. META’s benefit is indirect: a credible second supplier improves procurement leverage and potentially lowers AI infrastructure unit costs, but that benefit is unlikely to be material until deployments are operational.

Key Decisions for Investors

• Do not chase AMD solely on mix-shift headlines. Upgrade to a tactical long only after the next earnings release shows data-center revenue of at least $8.0B, stable-to-higher consolidated gross margin, and no material inventory or receivables build; target 15-20% upside over 3-6 months with an 8% stop on a guidance or margin miss.
• Use a conditional pair trade: long AMD / short NVDA in equal beta only if AMD demonstrates the above delivery and margin thresholds. The trade isolates a credible AI-share-gain catalyst; exit if AMD data-center growth falls below 15% quarter-on-quarter, Helios timing slips, or NVDA’s next platform adoption shows no customer diversification.
• Maintain NVDA as the cleaner core AI exposure versus AMD until independent deployment evidence confirms rack-level performance, software adoption and customer acceptance. The near-term risk/reward favors the incumbent where ecosystem advantages are more visible in reported margins and backlog conversion.
• Monitor MU, AVGO, VRT and SMCI as second-order read-throughs. Strong HBM pricing, high-speed networking demand and data-center power/cooling orders would corroborate a real rack-scale buildout; weak orders despite AMD shipment commentary would signal inventory staging rather than deployed compute demand.