Advanced Micro Devices, Inc. (NASDAQ:AMD) shares dipped slightly in Thursday's premarket trading, but the real story isn't the stock's tiny move. It's the numbers AMD just put on the table at its Technology Leadership Forum 2026. And those numbers are, frankly, enormous.
Nasdaq futures were up 0.09% and S&P 500 futures gained 0.22%, so the market backdrop was mildly positive. AMD's stock, however, edged down 0.44% to $480.80 in premarket, according to market data. Investors were clearly digesting the company's ambitious AI and data center plans.
At the forum, AMD management laid out an acceleration in its data center strategy, pointing to AI inference and agentic computing as major growth drivers. They also detailed Helios customer commitments, the expanding server CPU roadmap, software improvements, and efforts to secure additional supply. Let's break down what they said.
Server And AI Growth Outlook Accelerates
Matt Ramsay, AMD's Corporate Vice President of Financial Strategy and Investor Relations, shared some eye-popping growth numbers. The server business grew more than 50% in the first quarter and more than 70% in the second quarter. Both cloud and enterprise revenue increased more than 70%.
And that's just the beginning. AMD expects server revenue growth to exceed 80% in the second half of 2026. For 2027, the company provided an early outlook of at least 70% growth. The broader data center business, including AI, is expected to grow well above 100% in 2027.
To put that in perspective, Ramsay said AMD's early 2027 server revenue outlook is roughly 20% larger than the entire server market was in 2025. Let that sink in. The company's projected server revenue alone would dwarf the whole market just two years ago. That's not just growth; that's a paradigm shift.
AMD also raised its estimate for the server CPU market to $220 billion by 2030, and it continues to target more than 50% revenue share. Ramsay noted that agentic computing could account for roughly two-thirds of the market as AI spending shifts from training to inference. In other words, the future isn't just about building bigger models; it's about deploying them to do things.
Helios Anchors AMD's AI Expansion
The centerpiece of AMD's AI push is Helios, its rack-scale AI system. AMD plans to begin shipping components for Helios AI racks to manufacturing partners in September. Those components include MI450 accelerators, Venice CPUs, and Pensando networking products.
Management expects a significant revenue ramp in the fourth quarter, followed by another increase in the first quarter. That's a fast timeline, and it's backed by serious customer commitments.
AMD has secured 1-gigawatt commitments from OpenAI and Meta Platforms, Inc. (NASDAQ:META). Anthropic has committed to 1 gigawatt, with an ambition to reach 2 gigawatts. These are not small pilot programs; these are massive, infrastructure-scale deployments.
AMD also expects deployments through Oracle Corporation (NYSE:ORCL) Cloud Infrastructure, Microsoft Corporation (NASDAQ:MSFT) Azure, and other cloud providers. Ramsay said customer demand remains strong, although deployment speeds will also depend on the availability of land, power, infrastructure, and capital. In other words, even with all this demand, the real world still has constraints.
ZT Systems Supports Rack-Scale Push
AMD acquired ZT Systems to add system-level expertise as it expands beyond individual chips into complete rack-scale AI systems. Ramsay said ZT Systems has helped strengthen the Helios design and reduce execution risks.
AMD plans to begin the Helios ramp with a limited number of manufacturing partners before expanding more broadly. That's a cautious approach, but it makes sense when you're building something this complex.
Ramsay said AMD has moved beyond major design concerns. The focus has shifted to producing working racks at scale and helping customers quickly move production workloads onto the systems. It's one thing to design a great product; it's another to actually ship it in volume and make it work in the real world.
AMD Says ROCm Is Closing The Software Gap
Software has long been NVIDIA Corporation's (NASDAQ:NVDA) moat, with its CUDA ecosystem being the industry standard. AMD has been working to close that gap with its ROCm platform, and Ramsay says they're making real progress.
He said improvements over the past 18 months have reduced friction for large customers. The competitive gap has narrowed enough that software is no longer a central concern in discussions with leading AI companies. That's a bold claim, but if true, it removes a major barrier to AMD's adoption.
AMD plans to launch ROCm AI alongside Helios and continue updating the platform with future accelerator generations. So the software and hardware are being developed in tandem, which should help ensure they work well together.
Venice Expands AMD's CPU Roadmap
AMD is sampling its next-generation Venice server CPUs to customers and plans to use them in both Helios systems and its broader server portfolio. Venice scales to 256 cores and 512 threads, targeting workloads ranging from traditional cloud computing to AI infrastructure and agentic systems.
That's a lot of processing power, and it's designed to be flexible. AMD plans to follow Venice with its Florence CPU family in 2028. Ravenna is also under development with customer involvement.
Ramsay said AMD's chiplet-based approach allows the company to develop different server configurations more quickly for a wider range of workloads. Instead of designing a new chip for every use case, they can mix and match chiplets to create tailored solutions.
AMD Targets x86 And Arm Rivals
AMD faces competition from both its traditional x86 rival, Intel, and processors based on technology from Arm Holdings plc (NASDAQ:ARM). But Ramsay said AMD doesn't frame its strategy around the x86-versus-Arm debate. Instead, the company focuses on building the best processor for each workload.
He said AMD has increased its share of the x86 server market from 0.4% to the high-40% range. That's a remarkable turnaround, and the company believes its experience in security, reliability, and enterprise workloads will strengthen its position as agentic computing expands.
For agentic systems, Ramsay said customers are increasingly focused on metrics such as agents or computing threads per megawatt rather than the underlying instruction set. In other words, what matters is how much work you can get done per unit of power, not whether it's x86 or Arm.
Supply Remains The Key Constraint
Here's the interesting part: AMD says demand is not its biggest near-term challenge. Instead, the company is focused on securing enough manufacturing and advanced packaging capacity to meet expected growth.
AMD continues to work with Taiwan Semiconductor Manufacturing Company Ltd. (NYSE:TSM), where it ranks among the largest customers. AMD has also announced a $10 billion investment in Taiwan's technology ecosystem, largely focused on back-end capacity. That's a serious commitment to ensuring they can actually build the chips they need.
The company is also working with server manufacturers, cloud companies, and other partners to secure sufficient memory supply. Memory is a critical component, and it's been in short supply in the AI boom.
Ramsay said AMD continues to pursue additional near-term capacity. Longer planning horizons should also give AMD and its suppliers greater flexibility to support expected growth in 2027 and 2028. In other words, they're thinking ahead, not just scrambling to meet current demand.
So, what does all this mean for investors? AMD is betting big on AI, and the numbers are staggering. But the real test will be execution. Can they ship Helios on time? Can they scale production? Can they close the software gap? If they can, the growth potential is enormous. If not, well, that's the risk.
For now, the market seems cautiously optimistic. AMD's stock is down slightly in premarket, but the long-term story is one of massive expansion. As always, we'll be watching to see if the company can deliver on these ambitious promises.
