# Anthropic Eyes $900B Valuation, EU AI Act Omnibus Deal, and Enterprise AI's ROI Gap Widens

*Published Tuesday · May 12, 2026*

Tuesday, May 12, 2026. Your daily dose of what matters in AI, curated for business leaders.

The AI industry's valuation ceiling just broke through another floor of disbelief — Anthropic is in talks at $900 billion, Nvidia has crossed $40 billion in investment commitments this year alone, and Sierra just proved that vertical agentic AI is a multi-billion-dollar category. Meanwhile, a sobering survey finds only 23% of enterprises report significant ROI from the agents they've deployed. The money is flowing. The results are not keeping pace.

This edition covers thirteen stories spanning funding, policy, security, and the widening gap between AI ambition and AI impact. The throughline: we are deep into the deployment era, but the organizations winning are the ones treating AI as an operating model transformation — not a technology procurement exercise. Canada's regulatory vacuum, the EU's Omnibus recalibration, and Washington's internal AI power struggle all add jurisdictional complexity on top. Let's get into it.

## Today's Stories

- 💰 **[Anthropic Eyes $900B Valuation — Now Bigger Than OpenAI on Paper](https://www.cnbc.com/2026/04/29/anthropic-weighs-raising-funds-at-900b-valuation-topping-openai.html)** — Anthropic is in talks to raise at a $900 billion valuation, eclipsing OpenAI's $852B, on the back of $30 billion in annualized revenue — triple its 2025 full-year figure — driven largely by Claude Code adoption. The company is reportedly winning 70% of head-to-head enterprise sales against OpenAI among first-time buyers. For any enterprise advisor building a vendor shortlist, that win rate matters more than the valuation; for boards, the real question is whether a safety-first public benefit corporation can sustain that posture under trillion-dollar market pressure.

- 🏢 **[Enterprise Agentic AI Reality Check: 97% Deploy Agents, But Only 23% Report Significant ROI](https://writer.com/blog/enterprise-ai-adoption-2026/)** — A survey of 2,400 C-suite executives finds that while 97% of companies deployed AI agents in the past year, only 23% report significant ROI and 48% call adoption a "massive disappointment" — up from 34% last year. Three-quarters of executives admit their AI strategy is "more for show" than actual internal guidance, yet 69% plan AI-related layoffs anyway. This is the consulting opportunity hiding in plain sight: the gap between deployment spend and P&L impact is the exact space where structured transformation work creates value.

- 📜 **[EU AI Act "Omnibus" Deal Reached — High-Risk Rules Delayed, Core Obligations Intact](https://www.consilium.europa.eu/en/press/press-releases/2026/05/07/artificial-intelligence-council-and-parliament-agree-to-simplify-and-streamline-rules/)** — EU co-legislators reached a provisional deal at 4:30 a.m. on May 7, amending the AI Act before the August 2026 high-risk deadline. Sandbox deadlines shift to August 2027, the transparency grace period for AI-generated content tightens to December 2026, and a new provision explicitly bans AI-generated non-consensual intimate imagery. For any client with EU operations, treat "simplification" as regulatory spin: the substance of what's required for high-risk AI systems hasn't moved, and the compliance clock is still ticking.

- 🇨🇦 **[Canada's AI Regulation Gap Widens — Globe & Mail Calls for a "Third Path"](https://www.theglobeandmail.com/business/commentary/article-ai-artificial-intelligence-economy-regulation-eu-us/)** — Canada's original AI and Data Act died with Parliament's prorogation in January 2025 and has no legislative replacement, making Canada the most significant G7 economy without an operative AI regulatory framework. A Munk School fellow proposes a "buy, partner, build" approach designed for interoperability with allies. For Canadian enterprises deploying AI in banking, healthcare, or insurance, this vacuum isn't freedom — it's unmanaged risk, and boards need governance guidance now, not when legislation eventually arrives.

- 🔐 **[Google Thwarts Hacker Group's Plot to Use AI for "Mass Exploitation Event"](https://www.cnbc.com/2026/05/11/google-thwarts-effort-hacker-group-use-ai-mass-exploitation-event.html)** — Google disrupted a criminal group's plan to use the open-source OpenClaw framework for mass software vulnerability exploitation, following Anthropic's delay of Claude Mythos over fears it could discover decades-old zero-days. Cisco's AI security team previously found OpenClaw skill packages performing data exfiltration and prompt injection without user awareness. Any enterprise running open-source agent frameworks should be conducting a security review of installed skill and plugin packages immediately — AI-enabled cyber offense is no longer theoretical.

- 📜 **[US Intelligence Agencies Push for Power Over AI Regulation as White House Is Split](https://www.washingtonpost.com/politics/2026/05/11/trump-ai-regulation-commerce-intelligence/)** — The Trump administration is internally divided over giving spy agencies greater sway over AI regulation, with national security officials seeking expanded authority ahead of a China summit. The tension reflects a structural breakdown between Commerce-oriented AI promotion and intelligence-community risk posture. For multinationals operating in both the U.S. and Canada, this volatility creates real uncertainty in AI procurement — particularly for anything touching government or defense-adjacent workflows.

- 💰 **[Mira Murati's Thinking Machines Lab Locks In Frontier-Scale Compute With Google & Nvidia](https://techcrunch.com/2026/04/22/exclusive-google-deepens-thinking-machines-lab-ties-with-new-multi-billion-dollar-deal/)** — Thinking Machines Lab signed a multi-billion-dollar Google Cloud deal for Nvidia GB300 infrastructure, following Nvidia's commitment of at least one gigawatt of Vera Rubin compute — a threshold only the largest frontier labs approach. However, the lab has hemorrhaged talent: Meta poached five founding members (including co-founder Andrew Tulloch at a reported $1.5B package), while two co-founders returned to OpenAI. The compute commitments signal frontier-scale ambition, but the talent exodus is a real execution risk clients should weigh when asking "who are the next tier of frontier labs?"

- 🤖 **[Sierra Raises $950M at $15B+ Valuation — Claims 40%+ of Fortune 50 as Customers](https://techcrunch.com/2026/05/04/sierra-raises-950m-as-the-race-to-own-enterprise-ai-gets-serious/)** — Bret Taylor's AI customer experience platform closed a $950 million round, pushing past $15 billion in valuation, with agents now handling billions of interactions across mortgage refinancing, insurance claims, and retail returns. Sierra claims more than 40% of the Fortune 50 as customers. This is the clearest proof point that vertical agentic AI for customer operations is a multi-billion-dollar category — any enterprise running large-scale customer service needs a position on buy versus build for the agent layer.

- 💰 **[OpenAI Accelerates Acquisition Spree — Seven Acqui-Hires in 2026](https://aifundingtracker.com/ai-startup-funding-news-today/)** — OpenAI has completed seven acquisitions year-to-date, most recently Hiro Finance (backed by Ribbit and General Catalyst), alongside Astral (the team behind Python tools uv and Ruff), spanning finance, developer tooling, and personal agents. Revenue run-rate hit $19 billion as of February. OpenAI is executing a classic platform land-grab — enterprise buyers evaluating it as a long-term partner should factor in that the product surface area is expanding fast, which means more capability but also significantly more lock-in surface.

- 🇨🇦 **[Cohere Merges With Germany's Aleph Alpha at $20B Combined Valuation — Sovereign AI Play](https://medium.com/jonathans-musings/ai-startups-you-should-know-48d69d1ddf61)** — Canada's Cohere merged with Germany's Aleph Alpha at a combined $20 billion valuation — nearly triple Cohere's prior standalone number — positioning explicitly around sovereign AI for regulated industries. The deal combines Cohere's enterprise NLP track record with Aleph Alpha's European government and defense relationships. For PwC Canada clients in banking, healthcare, or public sector, this is the most relevant sovereign AI option in the market and a strategic positioning story worth tracking closely.

- ⚡ **[Nvidia Crosses $40B in 2026 Investment Commitments](https://llm-stats.com/ai-news)** — Nvidia has deployed over $40 billion in equity investments year-to-date, including a $30 billion stake in OpenAI, $3.2 billion in Corning, and $2.1 billion in IREN, while global AI venture funding hit $56 billion in April alone. Nvidia has transformed from a chip vendor into the AI industry's most powerful strategic investor — a company that now owns equity across the model, cloud, and infrastructure layers. Enterprise buyers should recognize that vendor alignment incentives in this market don't always run through the customer.

- 🤖 **[ServiceNow and Accenture Launch "Forward Deployed Engineering" to Scale Agentic AI](https://newsroom.accenture.com/news/2026/servicenow-and-accenture-launch-forward-deployed-engineering-program-to-scale-agentic-ai-across-the-enterprise)** — ServiceNow and Accenture are embedding engineers directly inside mutual customers' environments to build agentic AI workflows in production, with access to 300+ pre-built agent skills — a model borrowed from Palantir's playbook. Accenture's own research found only 32% of leaders report sustained, enterprise-wide AI impact. This is competitive intelligence for PwC: the firms that win the next wave of AI work will be in the customer's environment owning workflow redesign, not just presenting in the boardroom.

- 🏢 **[AI Is Reshaping the C-Suite — But Culture, Not Tech, Remains the #1 Barrier](https://www.cnbc.com/2026/05/11/heres-how-artificial-intelligence-is-changing-boardrooms.html)** — McKinsey calls AI "the largest organizational shift since the industrial and digital revolutions," IBM finds 59% of executives expect the CHRO's influence to grow, and 93.2% of respondents in Bean's 2026 AI & Data Leadership Survey cite cultural challenges as the principal barrier to adoption. Firms are increasingly creating Chief AI Officer roles to own governance, infrastructure, and integration. Clients without clear AI ownership at the executive level are making a de facto governance decision — just not a deliberate one.

## One Thing to Think About

Here is a number worth pinning to the wall: 97% of enterprises have deployed AI agents, yet only 23% report meaningful ROI, and three-quarters of executives admit their AI strategy exists primarily for optics. Meanwhile, Anthropic is raising at $900 billion, Sierra just proved agentic customer operations is a standalone category, and Accenture is physically embedding engineers inside client environments to close the delivery gap. The market has split into two distinct games. One game is the capital-and-capability race — valuations, compute commitments, model benchmarks. The other game is the one that actually shows up on a P&L: organizational redesign, change management, governance architecture, and the messy human work of making AI agents earn their keep in production workflows. Every dollar your client has spent on the first game without investing proportionally in the second is a dollar at risk of becoming a very expensive pilot. The firms that win this cycle won't be the ones with the most advanced model — they'll be the ones that close the gap between deployment and value fastest.

## Resources Worth Your Time

- **[Writer Enterprise AI Adoption Report 2026](https://writer.com/blog/enterprise-ai-adoption-2026/)** — The "48% call AI a massive disappointment" finding is the sharpest slide you can put in front of an executive sponsor to shift the conversation from technology procurement to transformation strategy.
- **[EU Council Press Release: AI Omnibus Provisional Agreement](https://www.consilium.europa.eu/en/press/press-releases/2026/05/07/artificial-intelligence-council-and-parliament-agree-to-simplify-and-streamline-rules/)** — Primary source on what actually changed (and what didn't) in the EU AI Act amendments; essential reading for any compliance team before August 2026.
- **[Globe and Mail: Canada Needs a "Third Path" on AI Regulation](https://www.theglobeandmail.com/business/commentary/article-ai-artificial-intelligence-economy-regulation-eu-us/)** — The most articulate framing yet of Canada's regulatory vacuum and what a proportionate, interoperable approach could look like; worth circulating to public sector and financial services clients.

*Curated by your AI briefing assistant for Chiel Hendriks.*