SemiAnalysis rated Jensen Huang's ComputeX 2026 keynote 'F Tier,' citing zero new AI datacenter hardware. The analyst firm also revealed NVIDIA's ARM laptop chip is delayed 6-8 months with broken video output.
Key facts
- Keynote rated 'F Tier' by SemiAnalysis
- Zero new AI datacenter hardware announced
- NVIDIA ARM laptop chip delayed 6-8 months
- Video output broken by NVIDIA-MediaTek interference
- Laptop makers banned from running benchmarks
The analyst firm SemiAnalysis published a scathing assessment of NVIDIA CEO Jensen Huang's ComputeX keynote, calling it 'one of the worst keynotes he has done' [According to @SemiAnalysis_]. The critique centers on two failures: no new AI datacenter announcements and a troubled consumer ARM chip.
No AI Datacenter News at ComputeX
The keynote, typically a venue for major GPU or networking reveals, delivered nothing on NVIDIA's core AI infrastructure business. Competitors including AMD and Intel have used ComputeX to preview upcoming datacenter silicon; NVIDIA's absence of new AI hardware announcements stands out as a strategic miss, particularly as hyperscalers like Microsoft and Google accelerate custom chip development.
NVIDIA ARM Laptop Chip: Delayed and Broken
SemiAnalysis reported that NVIDIA's laptop ARM chip, developed in partnership with MediaTek, is delayed by 6 to 8 months from its original launch window. The high-speed connection between the NVIDIA and MediaTek components caused interference that completely broke video output during development. Laptop manufacturers have been told 'definitely not let anyone turn them on or run benchmarks,' which SemiAnalysis described as 'immature hardware' [According to @SemiAnalysis_].
Unique Take
The structural observation here is that NVIDIA's AI dominance on the datacenter side may be creating a resource allocation problem: the company is so focused on high-margin AI GPUs (Hopper, Blackwell) that it is neglecting the consumer and PC roadmap, leaving partners with delayed, broken silicon. This mirrors Intel's struggles during its own GPU push — a cautionary tale of a company's core business cannibalizing secondary efforts.
Prior Context
In March 2026, NVIDIA reported datacenter revenue of $35.6B for Q4 FY2026, up 120% year-over-year. The company's consumer GPU business, by contrast, is a fraction of that. The delayed ARM chip threatens NVIDIA's attempt to replicate Apple's successful transition from x86 to ARM, a move that would unlock the PC market — but only if the hardware works.
What to watch
Watch for NVIDIA's next major datacenter reveal — likely at GTC 2027 in March — and whether the ARM laptop chip ships in Q2 2027 as now expected. If delays persist, PC OEMs may defect to Qualcomm's Snapdragon X or AMD's Ryzen ARM efforts.
