AMD Ryzen AI MAX PRO 400 launches in third quarter, only MAX+ 495 gets updated Radeon 8065S graphics

AMD confirmed its Ryzen AI Max PRO 400 lineup with three SKUs: Ryzen AI Max+ PRO 495, 490 and 485. The flagship 495 features 16 Zen 5 cores, 32 threads, 80MB cache, 40 RDNA 3.5 CUs and up to 55 TOPS, while the 490 and 485 top out at 50 TOPS with Radeon 8050S graphics. The platform supports up to 192GB of unified memory and up to 160GB of VRAM, with commercial systems from HP and Lenovo expected to launch in Q3 2026.

Analysis

The important signal is not the SKU list itself but AMD’s tightening of the halo/mobile workstation stack into a more clearly segmented commercial AI platform. That usually helps gross margin mix: the company is pushing OEMs toward premium, memory-heavy systems where platform attach rates for DRAM, storage, and validation services are higher, while also making it easier for enterprise buyers to standardize on one architecture for on-device AI workloads. The 192GB unified-memory ceiling and 160GB VRAM allocation are especially relevant for inference-heavy buyers who want to avoid discrete GPU BOMs, which could pressure entry workstation dGPU vendors and narrow the performance gap versus lower-end accelerator configs.

Second-order, this is more of a channel and design-win story for HPQ than a near-term unit-volume story for AMD. Commercial refresh cycles are slower, but once a platform is validated by HP/Lenovo and locked into enterprise procurement, it tends to persist for multiple quarters, creating a longer tail of revenue and reducing pricing volatility versus consumer Ryzen launches. The weaker NPU tiers on the mid-SKUs matter less for raw AI spend than for procurement differentiation: buyers may treat the top SKU as the only true “AI PC” spec, which could concentrate ASPs and leave the 490/485 as volume fillers with lower mix benefit.

The contrarian miss is that this may be a modest competitive win rather than a breakout catalyst. If the market is already assuming AMD’s AI PC traction, the real upside is in margin durability and ecosystem lock-in, not a sudden demand surge; the launch timing into 3Q26 also pushes monetization well out, leaving execution risk, OEM channel inventory, and any competitive response from Intel/NVIDIA plenty of time to intervene. For HPQ, the opportunity is more about maintaining premium enterprise share than expanding TAM, so any disappointment in attach rates or enterprise AI budgets would show up first in slower workstation order growth, not headline unit shipments.

Key Decisions for Investors

• Overweight AMD on pullbacks into weakness over the next 3-6 months: treat this as a medium-duration mix/margin catalyst rather than a near-term volume driver. Risk/reward is attractive if the market is underpricing premium commercial attach and memory-dense platform economics.
• Add a smaller tactical long in HPQ into evidence of design-win announcements or channel checks for 2H26 workstation refreshes. The setup is lower beta than AMD, but the leverage to enterprise premium mix is cleaner than the market typically assigns.
• Pair trade: long AMD / short a basket of legacy workstation-dGPU exposure on any rally, with a 6-12 month horizon. The thesis is substitution pressure from high-memory unified platforms reducing the need for discrete GPU entry configurations in selected commercial AI workflows.
• Use call spreads rather than outright longs for AMD if implied volatility rises into OEM launch chatter. The catalyst is real but delayed, so defined-risk exposure is better than paying for upside too early.
• Avoid chasing broad AI PC enthusiasm here; wait for evidence that 490/485 volumes do not cannibalize the premium 495 mix. If ASPs compress or OEM commentary skews to the lower SKUs, trim exposure quickly.