Top Stories
MARKET SHIFT
Anthropic Surpasses OpenAI in Revenue — $30B ARR
Anthropic's annualised revenue hit $30 billion in April 2026, surpassing OpenAI's $25B ARR for the first time — a seismic shift in the AI landscape. Claude 3.5 Sonnet now holds 32% of the enterprise LLM API market vs. OpenAI GPT-4o at 25%. Approximately 80% of Anthropic's revenue is enterprise, with higher retention and lower churn. Nearly a third of American businesses paid for Anthropic's offerings in March 2026, up 6+ percentage points in a single month. OpenAI adoption was flat at 35%.
MODEL RISK
Anthropic Limits Access to "Mythos Preview" After Autonomous Exploit Discovery
Anthropic is restricting access to its Mythos Preview model after discovering it can autonomously identify and exploit tens of thousands of software vulnerabilities — successfully reproducing and chaining exploits in over 80% of internal test cases, including flaws in major operating systems and long-standing open-source projects. The model demonstrated advanced autonomy, chaining multi-step exploits across interconnected systems — a significant dual-use risk milestone for the industry.
NEW MODEL
Meta Releases Muse Spark — Ranks 4th on Intelligence Index
Meta debuted Muse Spark, its first major AI model since the $14B deal to bring in Alexandr Wang. The model ranks 4th on the Artificial Analysis Intelligence Index v4.0, with standout performance in figure understanding (86.4% on CharXiv Reasoning) and medical reasoning (42.8% on HealthBench Hard). It features a parallel sub-agent architecture with a dedicated "Contemplating mode" for complex reasoning tasks. Performance on abstract reasoning remains a weakness (42.5 on ARC AGI 2).
REGULATION
OpenAI, Anthropic & Google Unite to Combat Chinese Model Cloning
OpenAI, Anthropic, and Google are sharing intelligence through the Frontier Model Forum to clamp down on Chinese competitors extracting capabilities from US frontier models through systematic querying and model distillation. Separately, AI regulation is heading for a US political showdown in 2026, with AI company PACs and pro-regulation PACs increasingly clashing. Utah became the first state to grant AI systems authority to renew drug prescriptions — a significant regulatory milestone in AI-powered healthcare.
ENTERPRISE
PwC: Top 20% of Companies Capture 75% of AI's Economic Gains
PwC's 2026 AI Performance Study finds that three-quarters of AI's economic gains are being captured by just 20% of companies — and the leaders are focused on revenue growth, not just cost reduction. Telecom leads agentic AI deployment at 48%, followed by retail/CPG at 47%. Enterprises have moved from experimentation to full deployment in code development, legal, financial analysis, and administrative support. Anthropic will gain access to ~3.5 GW of TPU-based compute via expanded Google-Broadcom partnership.
Audience Perspectives
🔐 CISO — AI Security Risks
Mythos Preview is a watershed moment: An AI model autonomously chaining exploits across major OS and open-source projects — with an 80%+ success rate — represents a step-change in the offensive AI threat. Even with Anthropic limiting access, similar capabilities will proliferate. Security teams should begin planning for AI-assisted attack acceleration: red team frequency, patch cycle SLAs, and detection of AI-generated exploit patterns all need reassessment.
Vendor lock-in risk: With Anthropic now leading on enterprise revenue and Google, OpenAI, and Anthropic all competing aggressively for enterprise contracts, vendor lock-in is a growing security and resilience concern. Ensure AI vendor contracts include data portability, model transparency, and SLA accountability. Review data privacy posture for all LLM integrations — especially as agentic deployments expand into sensitive functions (legal, financial, clinical).
🤝 Consulting Partner — Market Moves & Opportunities
Anthropic passing OpenAI in revenue is the most important signal of the quarter. Clients who standardized on GPT-4o should be re-evaluating their AI stack now. The 80/20 split of AI value accrual (PwC study) is your most powerful slide in any AI strategy deck — most clients are in the 80% and need a path to the 20%.
Emerging opportunities: (1) Agentic AI governance frameworks — enterprises deploying AI in legal, finance, and clinical contexts need policy, oversight, and audit trail design. (2) AI vendor risk assessments — with OpenAI, Anthropic, and Google all pursuing aggressive lock-in strategies, multi-vendor architecture and exit strategy work is in demand. (3) The Mythos-class dual-use model issue creates a new "AI Red Team as a Service" consulting category. (4) Utah's prescription AI precedent will drive healthcare AI compliance engagements nationally within 12 months.
💼 Business Executive — Strategic Implications
The competitive gap is widening fast. PwC's data is clear: the top 20% of companies using AI are pulling away from the rest — and they're generating revenue, not just cutting costs. If your AI strategy is still in pilot mode, it is a strategic liability.
Key decisions for leaders: (1) Anthropic overtaking OpenAI in enterprise revenue means it's time to revisit your AI vendor strategy — both are worth evaluating seriously. (2) The push into agentic AI (AI taking autonomous actions in workflows) is real and accelerating — telecom and retail are already at 48% deployment rates. (3) The political/regulatory environment is intensifying in 2026: the US federal AI regulatory battle will likely shape enterprise compliance obligations within 12–18 months. Engage now, don't wait. (4) AI is moving into direct clinical and professional decision-making (Utah prescription renewals) — industries that resist this will face competitive disadvantage but also carry new liability.