Nvidia GPUs Fuel $260M B300 Deployment and 30× MLPerf Leap
Axe Compute secured a $260 million contract for a 2,304-GPU B300 cluster delivering 4.8 MW of power from Q3 2026, highlighting rising enterprise AI demand for Nvidia technology. Oracle cancelled $1.1 billion in Super Micro GB300 rack orders and Nvidia’s GB200 NVL72 platform delivered 30× MLPerf gains, underscoring investor appetite.
1. Major Enterprise GPU Deal
Nvidia has been selected by Axe Compute for a $260 million, 36-month contract to deploy 2,304 B300 GPUs in a dedicated cluster with 4.8 megawatts of power from a Tier 3 U.S. data center, with options to renew beyond the initial term and a targeted start in Q3 2026.
2. Breakthrough MLPerf Benchmark
The GB200 NVL72 platform delivered up to 30× higher MLPerf inference throughput compared with the H200 configuration, while recent software improvements lifted Blackwell performance, reinforcing Nvidia’s cost-per-performance leadership in AI workloads.
3. Super Micro Rack Order Cancellation
Oracle cancelled 300–400 Super Micro GB300 rack orders valued at $1.1–1.4 billion after 100–200 units had been delivered, prompting a shift of enterprise GPU rack business to alternative suppliers and reducing immediate demand for Nvidia GB300 systems.