The AI Daily

China's Moonshot AI Kimi K3 is the latest Chinese model to meaningfully close the benchmark gap with US frontier labs, intensifying the open-weight model arms race and forcing fresh scrutiny of whether American dominance is as durable as assumed.

The compute scramble intensifies on multiple fronts: Anthropic is negotiating a blockbuster $10B lease of Meta's data center capacity, SpaceX is in Pentagon compute talks, and a secondary market for used Nvidia GPUs just opened — all while chip stocks sold off sharply this week.

The AI funding machine keeps churning: Databricks is finalizing a round at a $188B valuation, inference cloud General Compute raises $400M in debt, and Apple's trade-secrets lawsuit against OpenAI is casting a shadow over OpenAI's IPO timeline.

Enterprises are wrestling with the messy reality of deploying AI agents on legacy infrastructure — from Intuit rebuilding its agent architecture twice in four months to LinkedIn, Walmart, and Zendesk surfacing latency as the core bottleneck — while Google expands AI Mode with third-party app integrations and EY proposes multimodal knowledge graphs to improve RAG accuracy.

AI is embedding itself deeper into consumer and enterprise applications — Netflix is using generative AI on 300+ titles in production, Meta is adding parental safety AI for teens, Capital One is open-sourcing an AI security scanner, and BofA is appointing senior leaders to drive AI adoption across global markets.

The Trump White House is quietly asserting control over who can access frontier AI models — a dramatic power shift away from the labs themselves — while ASML navigates an increasingly treacherous US-China export-control tightrope with a fifth of its revenue at stake.

The AI boom is hitting India's smartphone market in a tangible and painful way — a memory crunch driven by on-device AI requirements is pushing up prices and slowing demand — while Khosla-backed upliance.ai is bringing AI-native appliances to Indian consumers.

Intuit scrapped its own AI agent architecture twice in four months. At VB Transform 2026, its AI VP called that the fast path — While most AI coverage focuses on model capabilities or funding rounds, Intuit's account of deliberately blowing up its agentic architecture twice in four months offers a rare, honest look at what enterprise AI deployment actually costs in organizational pain — and why speed of iteration beats correctness of initial design. For any senior strategist planning a large-scale agentic rollout, this is the operational playbook no one else is writing publicly. Read →