Nvidia said its latest chip designs are now reaching customers, a move the company argues will solidify its dominant position in the artificial intelligence market even as its largest buyers develop their own silicon.
New Chip Rollout · Nvidia
Nvidia's Newest Silicon Reaches Customers — Betting on the AI-Agent Era
The Vera CPU is now in full production and shipping to early adopters including Anthropic, OpenAI and xAI, while a new RTX Spark superchip targets a fresh "AI PC" category — all built on the ramping Blackwell architecture.
Performance leaps on Blackwell
How the new generation compares against what came before — same metric, drawn to scale.
MoE model speedup — NVL72 rack
AI tasks — GeForce RTX 50 series
RTX 50 delivers ~4.5× the AI throughput at roughly 30% lower power consumption.
3
Early Vera CPU adopters: Anthropic, OpenAI & xAI
6
PC makers to ship RTX Spark: Lenovo, HP, Dell, Surface, Asus, MSI
2026
RTX Spark Windows PCs arrive from autumn
12–15
Months to high-volume US output at TSMC Arizona & Texas
Welcomed
Faster training & inference for large models on Blackwell
PC makers back RTX Spark as the seed of a local "AI PC" category
Still Unclear
Independent Vera CPU benchmarks remain limited in public data
Real-world gains for early adopters not yet demonstrated
The tension
Nvidia's biggest buyers are becoming competitors — building custom silicon of their own.
Google TPU
In-house AI accelerators
Amazon Trainium
Custom training chips
Nvidia's answer
New Vera, Spark & Blackwell designs to defend the lead
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…