Good morning. It's Wednesday, April 1, and we're covering Sora’s sudden shutdown, new state-level AI guardrails in California, Europe’s race to build massive AI infrastructure, and more.

If you enjoy this email, share it with a friend. First time reading? Sign up here to keep up with the future of tech.

Nebius Plans Massive Europe AI Factory
Nebius plans a 310MW AI data center in Lappeenranta, Finland, targeting operations by 2027. The facility would rank among Europe’s largest as demand for compute surges. The move comes amid a wave of regional AI infrastructure investments. Europe faces hurdles including energy costs and grid access.

OpenAI Raises $122B, Nears $1T Valuation
OpenAI secured $122B in funding at an $852B valuation, accelerating its AI infrastructure push. Revenue has surged to $2B monthly, with ChatGPT nearing 1B weekly users. The company is expanding compute partnerships and building an AI “superapp,” with enterprise driving over 40% of revenue.

Claude Code Leak Exposes AI Playbook
Anthropic accidentally exposed Claude Code’s source via a public npm package, revealing ~512,000 lines of code. The leak outlines key systems like “self-healing” memory and autonomous background agents. No user data was compromised, but the IP exposure could aid competitors. The incident comes as Claude Code generates ~$2.5B ARR with strong enterprise adoption.

Salesforce Supercharges Slack With AI
Salesforce unveiled 30 new AI features for Slack, centering on an upgraded Slackbot agent. New “AI skills” let users automate tasks like budgeting, meeting setup, and data retrieval. Slackbot can now summarize meetings, connect with external tools, and act autonomously across workflows. The push aims to transform Slack into a core enterprise productivity platform.

Prove it. ↓ Real or fake. Answer below.

New research shows AI harnesses can self-improve, boosting model performance and enabling fully self-evolving software systems.

OpenAI shut down its video-generation tool Sora just 16 months after launch, citing low usage and unsustainable costs, according to a Wall Street Journal investigation. The app’s user base peaked at about 1 million before dropping below 500,000, while operating costs reportedly reached roughly $1 million per day due to compute-intensive video generation.

CEO Sam Altman redirected resources toward more commercially viable areas as competitors like Anthropic gained traction with enterprise tools. The shutdown also abruptly ended a major partnership with The Walt Disney Company, which had committed to a reported $1 billion partnership, underscoring the economic pressure shaping AI product decisions. → Read the full article here.

Gavin Newsom issued an executive order on March 30, 2026 requiring artificial intelligence companies contracting with the state to implement safety, privacy, and transparency safeguards. The order mandates vetting of companies’ policies on misuse, bias, surveillance, and harmful content, while also directing the state to independently assess vendors flagged by federal authorities.

It comes amid tensions with Donald Trump over federal resistance to state-level AI regulation. The policy also introduces watermarking for AI-generated state media to combat misinformation, signaling California’s effort to shape AI standards through procurement power. → Read the full article here. (Paywall)

Meta Platforms unveiled two new AI-powered Ray-Ban smart glasses on March 31, 2026, priced at $499 and designed for prescription users. The models—developed with EssilorLuxottica—expand Meta’s push into wearable AI, a category where it currently leads shipments. Available for pre-order in the U.S. with wider retail rollout starting April 14, the glasses include adjustable hardware features for personalized fit.

The launch comes as Meta ramps up investment in “personal superintelligence” devices and faces growing competition from Snap Inc. and Alphabet Inc., reinforcing its strategy to scale AI hardware adoption. → Read the full article here.

Young Workers Revive Skilled Trades: Report highlights rising interest in hands-on careers like diesel repair and forensics, as demand grows for non-automatable skills blending judgment, dexterity, and technical expertise.

New York Times Drops Freelancer: The paper cut freelancer Alex Preston after a January review reused Guardian phrasing via AI, breaching standards; he admitted error and apologized.

Alexa+ Adds Conversational Food Ordering: Amazon’s upgraded AI assistant now integrates Uber Eats and Grubhub, enabling natural voice ordering with real-time edits; rollout begins now on Echo Show devices.

Anthropic Expands AI Safety Partnership: The company will share economic and risk data with Australia under a new agreement, supporting research, policy planning, and AI adoption tracking.

Penguin Sues OpenAI Over Dragon Books: Penguin Random House filed a Munich lawsuit alleging ChatGPT mimicked its Coconut the Little Dragon series, potentially setting a precedent for AI copyright disputes.

The Image Is Real. Tune in next week for another Real or AI test.

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team