Google Cloud’s NEXT Big Moment

Google’s rapid AI improvement has lifted its once-beleagured cloud division, with startups and enterprises now clamoring to build on top of the newly competitive Gemini. And this week, the company is set for its next big moment, as it brings together its stakeholders at the fast-growing Google Cloud Next event in Las Vegas.

Though Google Cloud Platform still trails its rivals in terms of market share, the business is growing fast, with Q4 cloud revenue increasing 48% to $17.7 billion. Meanwhile, Google’s cloud revenue backlog more than doubled in 2025, growing to $240 billion by the end of last year. Google’s also said AI customers on average use 1.8x as many Google products as non-AI customers.

Cloud Next, which has long outgrown San Francisco’s Moscone Center, will be a moment for the company to mark its progress in the cloud wars, a competition where it was once left for dead.

Given the AI race’s speed, it won’t be safe for Google to take a breath though. With the rise of Claude Code and OpenAI’s Codex, the company’s now under pressure to bolster its own agentic coding play. Google’s reportedly created a “strike team” to improve its AI coding models. And in the meantime, it will focus on helping its customers build agents of their own, a focus of its Vegas event.

Unlike some of the more messianic-style messaging you’d hear from Google’s competitors, the company is taking a more grounded approach this year. Bottlenecks, for instance, will be a key theme at Cloud Next, with more than a dozen sessions citing the topic. One session, “The human bottleneck: Why great tech fails and how to drive value in AI,” features World Labs co-founder Fei-Fei Li.

The bottleneck theme is pertinent given the ‘capability overhang’ that so many AI labs talk about. By this, they essentially mean that AI’s abilities have outpaced humans’ capability to implement them. If Google is betting that further growth will come from helping customers solve these problems, the payoff could be further gains against Amazon Web Services and Microsoft’s Azure.

Here’s a sample of this year’s Cloud Next talks:

• NVIDIA Omniverse For Physical AI: From Simulation to Reality

• The journey to agentic applications at Spotify

• Securing global innovation: How industry leaders scale defense across complex environments

• From algorithms to agents: How Gen Z, Gen Alpha, and AI are rewiring commerce

• Inside the race to build AI factories (Alex is participating)

The Intelligence Report

• The National Security Agency is reportedly using Anthropic’s new Mythos model despite the government blacklist.

• Google added new AI features for Chrome, including new search tools and “skills” powered by Gemini.

• Meta is expected to lay off around 10% of staff next month. Meanwhile, it’s also reportedly building an AI clone of Mark Zuckerberg to attend meetings and talk with staff.

• OpenAI had several major departures, including Kevin Weil (former chief product officer), Srinivas Narayanan (CTO of B2B Applications) and Bill Peebles (head of Sora).

• Publisher efforts to block AI scraping are creating serious worries for the Internet Archive, which operates the Wayback Machine.

• Amazon debuted a new agentic AI tool for drug discovery via AWS.

• Microsoft is reportedly testing a new always-on agent for Copilot similar to OpenClaw.

• A humanoid robot won a half marathon race in Beijing, beating the human race record by several minutes. (It was one of 300 robots that competed in Sunday’s race.)

• Emil Michael, Under Secretary of War for Research and Engineering at the Pentagon, joined the Big Technology podcast to talk about how AI is shaping modern warfare, his decision to label Anthropic as a supply chain risk, and what he thinks about the Pentagon Pizza Index.

Super Apps Vision Come Into Focus

Last week, OpenAI debuted a new Codex super app for desktops, with coding tools and other features for computer control and an in-app browser. The updates for Codex also include more than 90 extra plugins that combine skills, app integrations and MCP servers. OpenAI also released a new reasoning model called GPT-Rosalind to support life sciences research including biology, drug discovery and translational medicine.

Other AI model updates announced:

• Anthropic released a new AI tool called Claude Design that lets users create prototypes from text prompts. It’s powered by Claude Opus 4.7, which was also released last week.

• Perplexity released a new “Personal Computer” AI assistant for Mac users, with multi-model orchestration through new computer agents for working with files, apps, websites and other connectors.

• Google debuted a new Gemini desktop app for macOS, featuring a way to analyze content in any shared window. Google also released a new Gemini desktop app for Windows desktops.

7 Stats From Stanford’s 2026 AI Index

Stanford’s 2026 AI Index provides yet another annual deep-dive into AI’s technical progress, economic influence and societal impact. With more than 400 pages, the annual report from the Stanford Institute for Human-Centered AI features far more than we can fit into this section. But here are a few quick stats and other insights that stood out.

• AI is getting better at performing in professional domains like tax evaluations, mortgage processing, corporate finance and legal reasoning. (Accuracy for the top 15 models ranged between 60% and 90%.)

• The performance gap between U.S. and Chinese AI models is closer than ever, with the top U.S. models now only ahead by 2.7%

• Video models like DeepMind Veo 3 are getting better at generating content with realistic object behaviors without being trained on them. Two examples: simulating buoyancy and solving mazes.

• The number of AI researchers moving to the U.S. has slowed drastically, falling by 80% in just the last year.

• The number of AI safety incidents jumped to 362 in 2025, up from 233 in 2024.

• Open-source AI projects continue scaling, with 5.6 million projects on GitHub and Hugging Face uploading tripling in three years.