Key Highlights

  • Anthropic’s lead over OpenAI in enterprise widens. Ramp’s latest AI Index puts Anthropic at 34.4% share of paid business users vs. OpenAI’s 32.3% — and shows Anthropic’s adoption up 4× since 2025 while OpenAI’s growth has plateaued. Much of the gap traces to Claude Code expanding beyond engineering into finance, legal, and research workflows.
  • NVIDIA and David Silver bet on “superlearners.” A new strategic engineering partnership between NVIDIA and Ineffable Intelligence — the London lab founded by the AlphaGo architect — targets RL infrastructure for systems that “learn continuously from experience.” Silver: researchers have largely solved “the easier problem of AI… how to build systems that know all the things humans already know.” The next frontier is discovery.
  • Video becomes searchable infrastructure. NVIDIA’s Metropolis VSS Blueprint v3 ships with a modular fusion-search architecture and agent-skill integration, letting coding agents like Claude Code and Codex deploy live-stream video analytics through chat prompts instead of manual microservice wiring.
  • Quantum-materials science gets a 1,000× speedup. Researchers compressed XFEL data analysis from nine months to under four hours on 32 NVIDIA GB200 Grace Blackwell Superchips — a concrete demonstration of how accelerated computing collapses experimental cycle times in materials physics.
  • Google Arts × Es Devlin launches a UK-wide AI portrait installation at the National Portrait Gallery, running through October 2026 — pairing Gemini Image with charcoal-and-chalk styling for a participatory live wall.

Analysis & Opinion

The enterprise shift OpenAI saw coming — Rundown

Two months after OpenAI leadership flagged Anthropic’s enterprise momentum as a strategic threat, Ramp’s AI Index — drawn from corporate-card and invoice data across 50,000+ U.S. businesses — confirms the inflection. Anthropic’s paid-customer share climbed to 34.4% in April, overtaking OpenAI’s 32.3%, while overall AI usage across companies in the index reached 50.6%. The shift is attributed largely to Claude Code, which has moved Anthropic beyond technical buyers into finance, legal, and research workflows. The piece reads alongside yesterday’s TechCrunch report on the same data, but adds OpenAI’s internal awareness as historical context — the company saw it coming, but couldn’t reverse the trajectory.

New Products & Tools

Es Devlin’s “A National Portrait” for the National Portrait Gallery — Google

A decade-long Devlin–Google Arts & Culture collaboration produces a participatory installation running through October 27, 2026. UK adults photograph themselves via a dedicated platform; Gemini Image transforms the photo into an animated charcoal-and-chalk portrait in Devlin’s style. Participants can redraw up to five times and opt to include their portrait in the gallery’s live collective display. Builds on prior Devlin–Google works Please Feed the Lions and Poem Portraits.

Check out the latest news from YouTube’s Brandcast 2026 — Google

YouTube unveiled AI-driven advertising tools at Brandcast 2026: Custom Sponsorships uses AI to surface videos aligned with brand-preferred moments; a new Masthead with Custom Content Shelf adds a secondary creative slot to the platform’s marquee placement; Buy with Google Pay brings two-click purchases to connected-TV viewing; and Affiliate Partnerships Boost lets brands amplify organic creator content that already features their products.

Transform Video Into Instantly Searchable, Actionable Intelligence with AI Agents and Skills — NVIDIA Developer

The Metropolis Blueprint for Video Search and Summarization (VSS) v3 ships with a modular architecture, enhanced fusion search, and an “agent skills” specification that makes the blueprint deployable through conversational agents — Codex, Claude Code, OpenClaw, NemoClaw — instead of manual microservice configuration. Targets real-time intelligence on millions of live streams or recorded hours: monitoring operations, identifying trends, automated reporting. Setup runs via NVIDIA’s Brev Launchable with VSCode remote attach.

Research

NVIDIA, Ineffable Intelligence Team Up to Build the Future of Reinforcement Learning Infrastructure — NVIDIA

Strategic engineering collaboration between NVIDIA and David Silver’s London lab Ineffable Intelligence to build RL infrastructure for what Jensen Huang calls “superlearners — systems that learn continuously from experience.” Silver’s framing: pretraining-era models know what humans already know; the next problem is discovering new knowledge through experience. Unlike fixed-dataset pretraining, RL pipelines generate their own data — placing unique demands on interconnect, memory bandwidth, and serving architecture that the partnership aims to address.

Accelerated X-Ray Analysis for Nanoscale Imaging (XANI) of Novel Materials — NVIDIA Developer

A breakthrough in XFEL data processing: 42 TB of data analyzed in under four hours on 32 GB200 Grace Blackwell Superchips, down from nine months with conventional methods, while maintaining data precision. Demonstrated on quantum-material characterization via phonon-dispersion reconstruction from ultrafast experiments. XFEL facilities like SwissFEL, SACLA, LCLS-II, European XFEL, and PAL operate at >1M pulses/sec with 35-megapixel detectors — the analysis bottleneck has been the limiting factor for fusion, semiconductor, battery, and catalyst research.

References

  1. The enterprise shift OpenAI saw coming — Rundown, 2026-05-14 [blog]
  2. Es Devlin’s “A National Portrait” for the National Portrait Gallery — Google, 2026-05-14 [blog]
  3. Check out the latest news from YouTube’s Brandcast 2026 — Google, 2026-05-13 [blog]
  4. NVIDIA, Ineffable Intelligence Team Up to Build the Future of Reinforcement Learning Infrastructure — NVIDIA, 2026-05-13 [blog]
  5. Transform Video Into Instantly Searchable, Actionable Intelligence with AI Agents and Skills — NVIDIA Developer, 2026-05-13 [blog]
  6. Accelerated X-Ray Analysis for Nanoscale Imaging (XANI) of Novel Materials — NVIDIA Developer, 2026-05-13 [blog]