AI Infrastructure Pulse

Enterprise GPU System Expansions

• 🔥 Supermicro Adds Blackwell GPUs: Super Micro Computer announced new enterprise systems featuring NVIDIA RTX...

Rack-scale momentum builds as OEMs integrate NVIDIA RTX PRO Blackwell GPUs:

• Supermicro adds certified systems with up to 8 GPUs/node in 4U/5U...

OCI deepens NVDA integration with new Model Import for Nemotron 3 Super—the first NVIDIA model supported—enabling advanced reasoning on managed OCI...

CRDO's connectivity edge sharpens with 800G ZeroFlap transceivers that eliminate optical link flaps plaguing large-scale AI networks.

• Boosts...

• $1T Market Thesis: NVIDIA projects $1 trillion AI chip revenue opportunity through 2027.
• Heterogeneous Stack Shift: Vera Rubin architecture...

Nvidia-backed US startup and Korean conglomerate announced plans to build the largest AI data center in Seoul. Key signal for GPU demand in Asia-Pacific expansions.

Oracle's partnerships with AWS, Azure, and Google enable it to capture database workloads as enterprises shift to multi-cloud strategies.

Earnings & Guidance

• 🔥 Oracle Infrastructure Revenue: Oracle's infrastructure business revenue jumped 84% to $4.9 billion in the three months...

• Massive Scale: OCI Superclusters connect hundreds of thousands of GPUs for 17 zettaFLOPS peak, 131 Pb/s throughput
• Nvidia Rubin Power: Next-gen...

Nvidia advances Rubin roadmap with Rubin Ultra tray demo—key highlights:

• World's first AI GPU packing 1TB HBM4E memory per package with four compute...

Broadcom (AVGO) expects $22 billion in Q2 FY26 revenues, a 47% YoY growth, boosted by new optical and Ethernet tech for AI.

Oracle raised FY2026 capex guidance from $35 billion in its Q2 earnings report, brightening Cloud Infrastructure outlook for OCI expansion. Key signal for AI data center scaling investors.

Nvidia is integrating technology from AI chip startup Groq into its GPU chip systems, as expected—hinting at evolving needs in the AI stack.

Jensen Huang's GTC vision pivots NVDA from one-GPU-fits-all to specialized stacks, blending Rubin GPUs with Groq LPUs for AI factories.

• 35x...

AMD-Celestica partnership launches Helios, a rack-scale AI platform with MI450 GPUs targeting late-2026 hyperscale deployments.

• Integrates dozens...

Key angles from Oracle's Q3 FY26 earnings:

• OCI infrastructure revenue jumped 84% to $4.9B in the quarter ended Feb. 28, fueled by explosive AI cloud...

Broadcom (AVGO) dominates with 80% market share in high-end Ethernet switching, positioning it as the primary beneficiary and architect of the AI...

Major Partnerships & Deals

• 🔥 CoreWeave-BCE 300MW AI Data Center: BCE partners with CoreWeave and Cerebras for a 300 MW facility featuring...

Heterogeneous AI stack advances signal perf-per-dollar shifts:

• Rubin NVL8 taps Intel Xeon 6 host CPU for 2.3x memory bandwidth, orchestration...

• Reference architecture unveiled: Palantir Foundry integrates with NVIDIA Blackwell GPUs for on-premise AI systems
• Targets governments, defense,...