AMD Claims Zen 6 Venice Outperforms Nvidia Vera by 2.2x in Throughput

AMD is firing back at Nvidia's recent Vera CPU performance disclosures with SPEC CPU 2026 results for its next-generation Zen 6 'Venice' architecture, claiming a commanding lead in both throughput and per-core performance.

Throughput Claims

According to AMD, the EPYC 9996 Venice processor delivers 2.2x the throughput of Nvidia's Vera CPU in SPECrate 2026_int_base benchmarks. AMD's figures show Venice scoring 2070 compared to Vera's 925. The comparison pits AMD's 256-core chip at 600W TDP against Nvidia's 88-core Vera, which operates in a 250-450W TDP range.

Ravi Kuppuswamy, AMD's SVP of Server Product & Engineering, expressed satisfaction with Nvidia's published results. "I actually thought we were beating them by smaller numbers," Kuppuswamy said, according to Tom's Hardware.

Per-Core Performance

AMD also claims a 20% lead in per-core performance over Vera. This comparison uses a different Venice SKU: a 96-core high-frequency variant, likely the EPYC 9686F, rather than the 256-core EPYC 9996 used in the throughput benchmark.

The Venice architecture brings substantial specifications: up to 1024MB of L3 cache (1152MB for Venice-X variants), 16-channel DDR5 memory support, and clock speeds exceeding 5GHz.

Intel Comparison

In a separate benchmark, AMD's EPYC 9996 Venice claims up to a 3.4x performance jump over Intel's Xeon 6980P in data-heavy AI agent workloads. AMD also cites a 3.3x advantage for rack-level performance within a 100kW power constraint.

Context and Caveats

Nvidia released Vera performance data as part of its Vera Rubin platform, scheduled for late 2026. AMD's response aims to maintain competitive positioning in the high-performance AI and server CPU market, where tighter CPU-GPU integration is becoming critical for platforms like Nvidia's Vera Rubin and AMD's Helios AI.

The SPECrate scores cited by AMD represent company estimates and projections rather than independently verified third-party benchmarks. The disparity in core counts and TDP between the two architectures means that while AMD may lead in raw throughput, Nvidia's efficiency and workload-specific optimization remain key competitive factors.

The exchange underscores an intensifying battle between AMD and Nvidia in the CPU space, a market segment increasingly vital as AI platforms demand more sophisticated processor coordination.