AMD’s New Helios AI Server Enters Full Production, Shipping Expected by Q3-End

AMD Scales Up Production of New Helios AI Servers to Challenge Nvidia

The artificial intelligence hardware race has just intensified in a major way. Advanced Micro Devices (AMD) has officially announced that its second-generation Helios AI servers have entered full production, with commercial shipments slated to begin before the end of the third quarter. Unveiled by CEO Lisa Su during a keynote address in San Francisco, these high-performance rack systems represent AMD's most aggressive push yet to capture market share from Nvidia in the booming data centre AI infrastructure space.

As enterprise adoption of generative intelligence continues to accelerate, major tech firms are desperately searching for high-performance alternatives to power massive machine learning workloads. With the Helios architecture, AMD is proving that it has both the hardware firepower and the manufacturing capability to go toe-to-toe with the industry's incumbent leaders.

Next-Generation Silicon and Breakthrough Performance

At the heart of the new Helios rack systems is a powerhouse of cutting-edge engineering. Built in partnership with Taiwan Semiconductor Manufacturing Co., the servers integrate AMD's advanced MI455X AI accelerators alongside brand-new Venice central processors running on the innovative Zen 6 architecture.

The performance metrics speak for themselves. According to AMD's benchmark figures, the Helios architecture delivers a notable 15% increase in overall compute performance, paired with an astonishing 50% boost in HBM4 memory capacity and bandwidth. More importantly, the system is engineered to deliver up to 30% more processing tokens per dollar compared to competing hardware. Because global AI utilisation is shifting heavily toward running live models and handling high-volume daily queries, AMD has tailored Helios specifically to dominate the lucrative inference computing market.

Major Partnerships With OpenAI and Industry Heavyweights

Building world-class hardware is only half the battle; securing ecosystem adoption is where market dominance is truly won. Fortunately, AMD has lined up heavyweight industry backing, headlined by a landmark partnership with artificial intelligence pioneer OpenAI.

During the San Francisco keynote, OpenAI's Vice President of Compute Strategy, Sachin Katti, stepped on stage to confirm that the company plans to deploy Helios at massive scale starting in late 2026 and accelerating aggressively through 2027. OpenAI also outlined long-term plans to integrate AMD’s upcoming MI500 chips into its infrastructure roadmap. This monumental commitment, bolstered by additional support from industry giants like Anthropic and Meta, signals a sweeping industry appetite for hardware diversity. As enterprises demand reliable alternatives to avoid single-vendor bottlenecks, AMD's bold gamble on Helios is positioning the company to reshape the future of global AI computing.