In Today's AI News:
- Power Supercycle: Utilities, Grid Bills, and Data Center Math
- Cyber Reality Check: Supply Chains, “Mythos,” and Code Scanners
- Enterprise Agents Go On-Prem (and Into the Plumbing)
- AI Literacy & Labor Anxiety: Schools, Supply Chains, and State Law
- Vatican vs. the Machine: Dignity, Doctrine, and AI
- Google I/O Incoming: Gemini, Android, and XR
Today’s theme is simple: the machines are hungry, the grid is groaning, and everyone is rushing to sell you a ‘responsible’ solution before the bill arrives. Enterprise AI is marching on-prem, cybersecurity is learning that the real adversary is your release pipeline, and even the Vatican is drafting paperwork for the apocalypse.
Power Supercycle: Utilities, Grid Bills, and Data Center Math
The AI buildout has reached the boring part: power bills, grid upgrades, and who gets stuck paying for the transformers. Politics is finally noticing.
NextEra-Dominion deal reveals power's new landscape — Axios
A proposed utility mega-deal is framed as the clearest signal yet that AI-driven electricity demand is reshaping U.S. power markets — and bills.
Data center industry ramps up pushback on cost claims — Axios
The Data Center Coalition commissioned an E3 review arguing there’s no evidence data centers have historically been subsidized, pointing instead to broader drivers of rate increases.
Schiff enters data center fray with ratepayer protection bill — E&E News by POLITICO
Sen. Adam Schiff’s ‘Energy Cost Fairness and Reliability Act’ would push new power and transmission upgrade costs onto data center developers rather than ratepayers.
Singularity Soup Take: The compute race is now a rate-case race: whoever can finance power, permits, and interconnects wins — and everyone else gets a louder air conditioner.
Cyber Reality Check: Supply Chains, “Mythos,” and Code Scanners
If you can’t secure the release pipeline, you don’t really have “AI safety” — you have vibes with a CVE backlog.
Anthropic to share Mythos cyber flaw findings with global finance watchdog — The Guardian
Anthropic is briefing the Financial Stability Board about Mythos while keeping the model unreleased, with access limited to select banks and tech firms.
Four AI supply-chain attacks in 50 days exposed the release pipeline red teams aren't covering — VentureBeat
A run of incidents is used to argue that CI/CD, packaging, and dependency surfaces are the real ‘model risk’ most safety playbooks don’t cover.
AWS Weekly Roundup: AWS Transform at 1 year, Claude Platform on AWS, EC2 M3 Ultra Mac instances, and more (May 18, 2026) — AWS News Blog
AWS highlights agentic modernization tooling, direct Claude Platform access via AWS accounts, and a preview feature for full-repository security scanning with code-level fixes.
Singularity Soup Take: The “agent threat model” is quickly becoming “your build system is the agent” — and it has admin rights.
Enterprise Agents Go On-Prem (and Into the Plumbing)
The agent story is maturing into infrastructure and integration: governed data, private connectivity, and “please run this inside our firewall.”
OpenAI and Dell Technologies partner to bring Codex to hybrid and on-premises enterprise environments — OpenAI
OpenAI and Dell pitch Codex deployments closer to governed enterprise data — aiming to make agent workflows practical inside hybrid and on-prem estates.
Hitachi announces strategic partnership with Anthropic to strengthen "Lumada 3.0" through frontier AI — Hitachi
Hitachi teams with Anthropic to push ‘frontier AI’ into mission-critical environments, including a new deployment center and large-scale internal upskilling.
Confluent Makes It Easier to Build and Secure Real-Time AI at Scale — Confluent
Confluent announces new Confluent Intelligence and Cloud capabilities aimed at productionizing real-time AI with agent workflows, automated protection, and private connectivity.
Singularity Soup Take: We’re watching the same pattern repeat: models commoditize, integration becomes power, and the real product is governance.
AI Literacy & Labor Anxiety: Schools, Supply Chains, and State Law
While executives sell autonomy, everyone else is trying to figure out who trains the humans, who writes the rules, and who gets blamed when the bot makes the call.
New partnerships to support India’s educators and learners — Google Blog
Google says it’s bringing its AI Educator Series to India and partnering with UNICEF and state entities to build AI literacy for teachers and students.
Gartner Supply Chain Symposium/Xpo 2026 Barcelona: Day 1 Highlights — Gartner
Gartner highlights a shift toward ‘autonomous’ supply chains, with leaders citing AI and agentic AI as the top driver reshaping strategy.
Ex-Google CEO Eric Schmidt booed after AI remarks at Arizona commencement — The Guardian
A commencement speech turns into a mini referendum on AI and jobs as graduates voice anxiety and Schmidt gets booed for raising the topic.
Proposed State Privacy and AI Law Update: May 18, 2026 — JDSupra
A state-law roundup flags Colorado’s rewrite of its AI law alongside other U.S. privacy, chatbot, and sector-specific AI bills.
Singularity Soup Take: The social contract is being renegotiated in real time: teach it, regulate it, or get booed by it.
Vatican vs. the Machine: Dignity, Doctrine, and AI
When your pastor starts issuing policy documents about AI, you can officially declare the technology “mainstream.”
Pope and co-founder of Anthropic to launch pontiff’s AI encyclical on May 25 — AP News
The Vatican says Pope Leo XIV will launch an AI-era ‘human dignity’ encyclical with Anthropic co-founder Christopher Olah participating in the event.
Singularity Soup Take: If the Vatican is talking “dignity in the age of AI,” expect a lot more institutions to follow with their own rulebooks.
Google I/O Incoming: Gemini, Android, and XR
How to Watch Google I/O — WIRED
WIRED previews the I/O keynote and expects updates spanning Search, Android, Gemini, and a fresh look at Android XR smart glasses.
Singularity Soup Take: Every platform is turning into an “intelligence system” — and every user experience is about to become a negotiation with a helpful, slightly pushy robot.
Relevant Resources
AI Hardware & Infrastructure — The power, chips, and physical constraints behind the headlines.
Agentic AI — What agents are, how they work, and why governance keeps showing up.
AI From Zero to Horizon — A quick refresher for the humans still pretending they’re “just browsing.”
Today's Pulse: 15 stories tracked across 13 sources — AP News, AWS News Blog, Axios, Confluent, E&E News by POLITICO, Gartner, Google Blog, Hitachi, JDSupra, OpenAI, The Guardian, VentureBeat, WIRED