Reports
AI-generated structured vendor updates
Coherent Expands InP Fab with $50M CHIPS Grant, AI's Connectivity Bottleneck Drives Photonics Arms Race
Coherent receives $50M CHIPS Act grant to expand its 6-inch InP fab in Texas, quadrupling capacity. NVIDIA's $2B strategic investment and CEO Jensen Huang's presence signal a shift from GPU compute scaling to optical interconnect as the new AI infrastructure bottleneck.
NVIDIA Acquires Groq LPU: Inference Architecture Shift from HBM to On-Chip SRAM
NVIDIA signs ~$20B licensing deal with Groq for LPU tech, featuring 230MB on-chip SRAM at 80TB/s bandwidth. This targets Transformer inference decode, replacing HBM bottlenecks with ultra-low latency on-chip storage, potentially reshaping the AI inference chip landscape.
NVIDIA Tops Data Center Ethernet Market: GPU Compute Dictates Network Architecture
IDC reports NVIDIA captured 21.5% of the data center Ethernet switch market in Q1 2026, with $2.1B revenue. This milestone, driven by the Spectrum-X platform using RoCE and NVLink, marks a control shift where GPU compute dictates network architecture, directly challenging Cisco and Arista.
Huawei Ascend 910C Trains 1.6T-Parameter MoE Model: First Full Pipeline on Domestic AI Chips
Huawei, in collaboration with research institutes, completed full-parameter post-training of DeepSeek-V4-Pro (1.6 trillion parameters, MoE) on an Ascend 910C cluster. Key metrics: stable 1,500 steps on 1,000 cards, 30% compute utilization, 14% operator efficiency gain, zero reliance on foreign GPUs. This marks the first end-to-end trillion-parameter training loop on domestic chips.
SK Hynix HBM4E Samples: 3nm Logic, 384GB/GPU, Igniting AI Memory Bandwidth Arms Race
SK Hynix has sampled its 12-layer HBM4E, featuring TSMC 3nm logic die and enhanced per-pin bandwidth, targeting Nvidia Rubin Ultra with 384GB per GPU. This marks the start of a sprint with Samsung in next-gen AI memory, where HBM BOM share has surged to 65-70%.
NVIDIA Absorbs Groq LPU: Feynman GPU to Integrate SRAM Inference Tile, Hybrid Architecture by 2028
NVIDIA secures Groq's LPU inference technology via a non-exclusive license and key hires, planning to integrate large SRAM tiles into its 2028 Feynman GPU using TSMC SoIC hybrid bonding. This enables deterministic scheduling and 80TB/s on-chip bandwidth, shifting NVIDIA from a pure GPU vendor to a hybrid inference/training platform.
NVIDIA Tops Data Center Ethernet Switch Market: AI Factory Reshapes Networking Landscape
IDC reports NVIDIA as the #1 data center Ethernet switch vendor in Q1 2026 with $2.1B revenue (+192.7% YoY). This is driven by the Spectrum-X platform, a vertically integrated ecosystem of Spectrum switches, BlueField DPUs, and LinkX cables, purpose-built for AI GPU clusters, signaling a fundamental shift from general-purpose to AI-optimized networking.
OpenAI Invests $150M to Certify 300K Enterprise AI Advisors, Shifts Ecosystem Control
OpenAI launches Partner Network with $150M investment to certify 300K enterprise AI advisors by end of 2026, partnering with McKinsey, Accenture, and others. This marks OpenAI's first independent certification and sales channel outside Microsoft, signaling a shift from model supremacy to deployment ecosystem warfare.
Palo Alto's $25B CyberArk Buy Shifts Security Control to Machine Identity
Palo Alto Networks acquires CyberArk for ~$25B to create a unified agent identity and privilege management platform. This shifts security control from network firewalls and EDR to machine identity lifecycle, addressing the AI agent explosion, and ties revenue to token-based consumption.
OpenAI and Anthropic Mandate Identity Verification: End of Anonymous AI Era
OpenAI and Anthropic simultaneously enforce identity verification (ID upload, 3D liveness detection). This marks the end of anonymous AI usage, shifting AI from a chat tool to a authenticated agent capable of real-world actions, setting a new industry-wide user access standard.
Intel Lands Google TPU Package Order: Foundry Pivot Gains Traction, TSMC Still Core
Intel secured a multi-million unit order for Google TPU packaging using its EMIB-T technology, marking its largest external AI chip deal. However, analysts caution the order is primarily for packaging, not wafer fabrication, with TSMC retaining the core manufacturing role.
Microsoft GitHub Leases AWS Capacity: AI Demand Forces Cross-Cloud Collaboration, Shattering Vendor Lock-In
Microsoft's GitHub, facing a 14x surge in AI-driven code commits, is renting compute capacity from rival AWS. This reveals that no single cloud provider can meet AI infrastructure demand, breaking traditional cloud competition and heralding cross-cloud hybrid deployment as the new norm.
Google Gemini 3.5 Flash Turns Search into AI-First Answer Engine, Shifting Control from Links to Summaries
Google transforms Search into an AI-first answer engine powered by Gemini 3.5 Flash, with redesigned search bar, AI-generated summary pages, and proactive monitoring. Model improvements include 1M context, 65K output tokens, and multi-agent orchestration via Antigravity, enabling complex task automation.
Z.ai GLM-5.2 Open-Source: 744B MoE, 1M Context, MIT License as Geopolitical Shield
Z.ai releases GLM-5.2: 744B MoE with 40B activated parameters, 1M input and 131K output context, under MIT license. Released one day after Anthropic Fable 5's government takedown, it offers a downloadable, unbanable alternative with Anthropic API compatibility for zero-code migration, giving enterprises a sovereign AI option.
SGLang 0.5.13 Delivers 25x MoE Inference Speedup via Predictive Routing and Sparse KV Cache
SGLang 0.5.13 introduces two-stage MoE routing prediction and sparse KV cache, achieving a 25x inference speedup on NVIDIA GB300 NVL72. Benchmarks on A100 show 65% throughput gain, 40% latency reduction, and 62% lower routing overhead. This optimization directly attacks the core bottleneck of MoE inference, potentially reshaping AI inference economics.
Google TurboQuant: 6x KV Cache Compression, AI Inference Memory Cost Inflection Point
Google releases TurboQuant, a two-stage KV cache compression algorithm (PolarQuant + QJL) achieving 6x memory reduction (3-bit quantization) and 8x attention speedup with no measurable accuracy loss. The announcement triggered a sell-off in memory stocks (Micron -3%, Western Digital -4.7%), signaling a potential structural shift in AI inference memory demand.
CrowdStrike Launches Continuous Identity for AI Agents via SPIFFE, Shifting Control from Static Credentials to Dynamic Risk Plane
CrowdStrike unveils Continuous Identity for AI Agents at Identiverse 2026, leveraging the SPIFFE open standard to assign cryptographically verifiable identities to each AI agent, replacing static API keys. It provides real-time risk-based authorization per operation, zero standing privileges, delegated context propagation, and integration with Falcon AIDR. Built on acquired SGNL technology, it aims to define a new category in AI agent identity governance.