Google took the click out of search

At I/O 2026, Google launched Gemini Spark, a persistent cloud agent that runs around the clock and surfaces what changed across the parts of the web you care about. No query, no scroll, no click.

Spark runs on Antigravity 2.0, a multi-agent framework Google demoed by spinning up 93 parallel sub-agents that built an operating system core in roughly twelve hours.

The same keynote made Gemini 3.5 Flash the default across Google's stack, four times faster than 3.1 Pro on coding and agentic benchmarks. Gemini Omni added native video generation.

The downstream economics are the real story. Programmatic ad inventory has been built around one assumption for twenty-five years: a human types a query, scrolls, clicks.

Spark removes all three steps. If background agents read the web for billions of users, the click-through rate that funds independent publishers stops being a unit of value.

The economics of the open web were organized around the human reader. That assumption is now a product choice. Read More ↗

At this week's Beijing summit, Treasury Secretary Scott Bessent announced a plan to establish an AI safety protocol.

The focus: "best practices" for frontier models. Specifically, keeping powerful AI out of the hands of nonstate actors, criminal networks, extremist groups, actors outside government control.

Nvidia CEO Jensen Huang joined the delegation as H200 chip sales to major Chinese firms remain in limbo.

A safety protocol won't solve Taiwan or rare earths. But it creates one narrow lane where both governments agree that capability surprises need a shared floor.

That lane matters more than it sounds when the UK just published data showing AI cyber capability doubles every 4.7 months, with the latest frontier models outpacing even that. Read More ↗

OpenAI is preparing a confidential SEC filing for a September 2026 public debut at an $852B valuation, led by Goldman Sachs and Morgan Stanley.

The path cleared on May 19, when Judge Yvonne Gonzalez Rogers dismissed Elon Musk's lawsuit attempting to block the restructuring. The ruling cited a three-year statute of limitations. Musk has vowed to appeal.

This would be the largest planned tech IPO on record, executed with one of the original co-founders legally locked out on a calendar technicality.

Post-IPO, OpenAI's incentives change. Quarterly public-market accountability has historically pulled long-horizon research labs toward shorter product cycles. Read More ↗

A line buried in SpaceX's new S-1 filing (Nasdaq ticker SPCX): Anthropic pays $1.25 billion per month through May 2029 for compute capacity at the Colossus and Colossus II data centers.

Anthropic's internal projections target $10.9B in Q2 revenue, up from $4.8B in Q1, and its first quarterly operating profit at $559M. Most of that growth is enterprise adoption of Claude Code.

On the same day the deal surfaced, Andrej Karpathy joined Anthropic's pretraining team to build a unit that uses Claude to accelerate Claude. The recursive-self-improvement framing now has hiring momentum.

The political optics are worth holding. Anthropic spent the past month publicly resisting Trump administration pressure on military and surveillance use, and the compute it relies on flows through a company run by the administration's most visible ally. Read More ↗

Pope Leo XIV will publish his first papal encyclical fully dedicated to AI and human dignity on May 25. Christopher Olah, Anthropic co-founder, will co-present at the Vatican press conference.

This lands inside an active fight. Anthropic has been under pressure from the Trump administration to loosen safeguards that block its models from autonomous lethal use and state-level surveillance.

The encyclical, Magnifica Humanitas, structurally echoes Rerum Novarum, the 1891 document on worker exploitation under industrial capitalism. The Holy See is treating uncontrolled AI deployment as the same category of structural threat. Read More ↗