📊 Anthropic tops OpenAI in paid enterprise adoption 📊
Ramp's AI Index reveals a historic replacement atop enterprise AI adoption rankings.

For the first time, Anthropic has overtaken OpenAI in paid business adoption, according to fintech firm Ramp's latest AI Index tracking corporate payments across 50,000+ U.S. businesses. Anthropic rose 3.8% in April to reach 34.4% adoption share, while OpenAI fell 2.9% to 32.3%, even as overall enterprise AI usage climbed to 50.6%. Claude Code anchored much of the swing, helping Anthropic expand beyond technical teams into finance, legal, and research workflows. Over the past year, Anthropic has quadrupled its business adoption while OpenAI grew just 0.3% — a clear sign that competitive pressure is overriding vendor lock-in. However, Anthropic still faces headwinds from recent Claude outages and rising costs relative to OpenAI and open-source alternatives.

🤖 Figure 03 humanoids run full 8-hour factory shift 🤖
Figure AI's autonomous humanoids hit a speed and integration milestone on the factory floor.

Figure AI livestreamed a full 8-hour autonomous factory shift using teams of its Figure 03 humanoid robots powered by the Helix-02 system. The robots sorted packages by detecting barcodes, picking them up, and placing them on conveyors — averaging one package every 3 seconds, roughly matching human speed. All processing runs entirely onboard using a single neural network for vision, movement, balance, and manipulation, with no cloud inference required. When battery levels drop, a robot requests a replacement; if a fault is detected, it can self-diagnose, walk to maintenance, and call another unit to take over. This marks a major step toward continuous 24/7 robotic operation in real industrial settings.

🛡️ Microsoft ships MDASH with 100+ AI security agents 🛡️
Microsoft's multi-agent security integration outperforms Anthropic's Mythos on real-world exploits.

Microsoft unveiled MDASH, a multi-agent AI security system that deploys more than 100 specialized AI agents working across multiple models to find real-world software vulnerabilities. The system operates in stages: one set of agents scans code for potential flaws, a second group debates whether each finding is real and exploitable, and a final stage constructs proof-of-concept attacks to confirm bugs exist. MDASH surpassed Anthropic's Mythos on the CyberGym benchmark, which measures AI's ability to reproduce real-world vulnerabilities, and caught 16 flaws across Windows. This could reshape how enterprises detect and patch threats at scale.

INTERESTING TO KNOW

💰 Cerebras raises $5.5B in year's largest IPO 💰

AI chipmaker Cerebras raised $5.5 billion in its U.S. IPO at a $40 billion valuation, confirming the pricing power of dedicated inference hardware. The deal was 20x oversubscribed, led by Morgan Stanley, Citigroup, Barclays, and UBS. Cerebras' WSE-3 chip features 44GB of on-chip SRAM at 21 PB/s — roughly 6,000x the memory bandwidth of an Nvidia H100 — positioning it for low-latency inference use cases like voice and AI wearables. Analysts note this signals a coming split between "answer inference" and "agentic inference" in the chip market.

🔎 Google readies new Gemini model for I/O 🔎

Google is preparing to roll out a new Gemini model at its annual I/O developer conference on Tuesday, with reports indicating the model will roughly match OpenAI's GPT-5.5 in capability. The rollout places Google squarely in the frontier model race alongside OpenAI and Anthropic at a moment when enterprise adoption is shifting fast. A competitive Gemini release could increase multi-model adoption pressure, as Google uses I/O to set the narrative for its AI platform and developer ecosystem for the rest of the year.