Microsoft goes solo, Suno's $400M bet, and AI's runaway budgets
Microsoft declares AI independence from OpenAI, Suno doubles its valuation despite lawsuits, and Uber learns the hard way that 'use AI freely' is an expensive policy.
Subscribe free →Microsoft and OpenAI Broke Up — Now They're Ready to Fight
At Build this week, Microsoft unveiled a full stack of in-house AI capabilities: its own reasoning models, AI agents, a "super app," and a cybersecurity tool — all without leaning on OpenAI. For years Microsoft was essentially an OpenAI reseller; that era is over. The Scout personal assistant, portable agent policy files, and a new open-source eval framework (ASSERT) dropped in the same 48-hour window.
Why it matters: Microsoft isn't just hedging against OpenAI — it's now a direct competitor, and this Build was its coming-out party.
Suno Raises $400M at $5.4B Valuation — Lawsuits Be Damned
The AI music generator just more than doubled its valuation in seven months, going from $2.45B to $5.4B on a fresh $400M round. This is happening while Suno is still actively fighting copyright infringement lawsuits from major record labels. Investors are clearly betting the legal risk is priced in — or irrelevant.
Why it matters: If Suno can raise at this scale mid-lawsuit, it signals that copyright liability is no longer a dealbreaker for AI content startups.
Uber and Walmart Both Hit the Same AI Budget Wall
Uber blew through its entire annual employee AI tool budget in four months — after actively encouraging staff to use AI as much as possible — and has now capped spending. Separately, Walmart quietly throttled its internal AI coding assistant, Code Puppy, after LLM demand exceeded projections. Two massive enterprises, same lesson: "use AI freely" is not a financial strategy.
Why it matters: Corporate AI rollouts are smashing into real cost ceilings, and the unbounded-access era for enterprise AI is ending fast.
UK Regulators Force Google to Let Publishers Opt Out of AI Search
The UK's Competition and Markets Authority is requiring Google to build a tool that lets website publishers exclude their content from generative AI search features — separately from existing web crawl opt-outs. The tool will be piloted in the UK first, then rolled out globally. Publishers have been furious about AI Overviews eating their traffic; this is the first regulatory mechanism that gives them real leverage.
Why it matters: This sets a global precedent — if publishers can opt out of AI search in the UK, pressure will mount on every market to follow.
Google's Gemini Spark Knew Things Users Never Told It
The Verge's hands-on with Google's new Gemini agent Spark surfaced something unsettling: the assistant accurately recalled one tester's dog's name and another's wife's first name — details neither had explicitly shared with Google. It's piecing this together from Gmail, Photos, Calendar, and other Google data. The capability is impressive. The privacy implications are uncomfortable.
Why it matters: As AI agents get better at being helpful, the tradeoff between utility and surveillance becomes impossible to ignore — and Google's data moat makes it the most capable player in this space.
Coralogix Raises $200M to Be the Watchdog for AI Agents
Coralogix just closed a $200M round on the thesis that as AI agents flood enterprise production environments, someone needs to monitor what they're actually doing. The observability firm is building tools to track agent behavior, catch failures, and generate the operational data needed to keep autonomous systems running reliably. It's infrastructure for the infrastructure.
Why it matters: AI agent monitoring is becoming a category in its own right — and the race to own that layer is now funded at serious scale.
Quick Hits
- →Anthropic has filed for an IPO, marking what analysts are calling the shift of generative AI from research venture to enterprise utility. AI News
- →Meta's WhatsApp Business AI agent is now globally available, billed to businesses by token usage. TechCrunch
- →Google is rolling out AI deepfake call detection to combat impersonation scams — as fraudsters increasingly spoof trusted numbers with synthetic voices. TechCrunch
- →GitLab is cutting 14% of staff and exiting 22 countries to redirect resources toward AI workload infrastructure. TechCrunch
- →Martin Scorsese is using AI — but only for storyboarding, making him Hollywood's most reluctant AI adopter on record. TechCrunch
AI TLDR