AMD and Cerebras Join Forces to Speed Up AI Inference
AMD partners with Cerebras to deliver ultra-low latency AI inference, while its new EPYC 'Venice' CPU claims a lead over rival chips.
Advanced Micro Devices is sharpening its AI ambitions with a two-pronged strategy: a high-profile hardware partnership and a next-generation server processor that AMD says outperforms its closest competitors. The moves signal that AMD is no longer content to chase Nvidia solely on GPU performance — it wants to own the entire AI compute stack, from model training to real-time inference.
At the center of the announcement is a collaboration with Cerebras Systems, the maker of the wafer-scale AI chip that has drawn attention for its unusual, full-wafer design. Together, AMD and Cerebras will combine AMD's Helios GPU rack with Cerebras' inference silicon to create what they describe as an ultra-low latency AI inference platform. CEO Lisa Su framed the partnership as a direct play for enterprise and hyperscale customers who need AI responses delivered at speed, not just at scale. The joint solution is expected to reach customers through Cerebras Cloud in the second half of 2026.
Read more Micron Stock Slides Toward Worst Monthly Loss in Over a Decade →
Separately, AMD touted its new EPYC "Venice" server processor, claiming it delivers 20% higher performance than Nvidia's CPU offering and maintains a meaningful advantage over Arm-based server chips. Crucially, Venice is already in full production — not a roadmap promise — with systems expected to begin shipping in the fourth quarter. Strong customer demand, AMD said, remains intact. That matters because server CPUs are a quieter but highly profitable arena where AMD has steadily taken market share from Intel over recent years.
Markets responded with some ambivalence. AMD shares fell roughly 3% on the day, settling near $535 after briefly touching $525, though they remained above key technical support levels around the 100- and 200-hour moving averages near $532 — a level analysts watch as a short-term momentum signal. Cerebras shares, by contrast, climbed about 4% to roughly $218, continuing a recovery from a post-IPO low of $160 hit in late June. The stock's next meaningful technical resistance sits near $245.
The broader takeaway is strategic: AMD is deliberately positioning itself as the infrastructure layer beneath the AI boom rather than a single-product company. By pairing inference-optimized silicon with Cerebras' wafer-scale architecture and reinforcing its CPU credibility with Venice, AMD is building a story investors and enterprise buyers can follow across multiple product cycles — not just the next GPU quarter. Continue reading at Forexlive.
