At the Google Cloud Next 2026 keynote today, Sundar Pichai disclosed that three-quarters of all new code written inside Google is now generated by AI and reviewed by human engineers — up from 50% last fall and just 25% in late 2024. Pichai framed this as evidence Google is shifting to "truly agentic workflows" across its engineering organization. The acceleration from 25% to 75% in roughly 18 months is the clearest public benchmark yet for how fast AI is absorbing production software development at scale.
OpenAI rolled out a substantial Codex desktop update introducing background computer use on Mac — agents can see, click, and type with their own cursor while you work in other apps, with multiple agents running in parallel. The update also adds an in-app browser for rendering and annotating local and public pages, image generation via gpt-image-1.5, and over 90 new plugins spanning JIRA, GitLab, CircleCI, and more. Codex now proactively suggests work based on project context and memory, pushing further into autonomous developer tooling.
Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, signed a multibillion-dollar infrastructure agreement with Google Cloud to access systems powered by Nvidia's GB300 chips for training and deploying its Tinker platform. The deal — valued in the single-digit billions — supports the startup's reinforcement-learning-heavy architecture. Thinking Machines has rocketed from a $12B seed-round valuation to nearly $50B, underscoring the insatiable compute appetite of frontier AI labs.
CEO Evan Spiegel announced the layoff of roughly 1,000 employees and closure of 300+ open roles, explicitly citing "rapid advancements in artificial intelligence" as the driver. AI agents at Snap now generate over 65% of new code, handle more than 1 million support queries monthly, and flag 7,500+ bugs via automated code review. Snap's stock jumped 11% on the news — Wall Street is pricing in the thesis that smaller, AI-augmented teams can run a major platform.
Also at Cloud Next, Google Cloud announced a $750 million fund to help consulting giants — McKinsey, Accenture, Deloitte — prototype, build, and deploy agentic AI for their clients. The fund covers AI value identification, agent prototyping, deployment, upskilling, and embedded Google engineers. Separately, Vista Equity Partners inked a multiyear deal to deploy Google's agentic AI across its 90+ portfolio software companies.
At its annual product launch event today, Horizon Robotics revealed Stellar, China's first AI chip that integrates both in-cabin and autonomous driving functions on a single piece of silicon. The chip consolidates what previously required separate processors, reducing cost and latency for automakers. It signals China's push to build a domestic AI chip stack for the automotive sector amid ongoing US export restrictions on advanced semiconductors.
