Governments moved to the center of the AI story this week: Washington struck a deal to bring Anthropic’s most powerful model back online for a chosen few and asked OpenAI to vet customers for its newest release. Underneath the politics, the business reality kept hardening — Oracle tied 21,000 job cuts to AI, cheaper Chinese models pulled in US customers, and OpenAI leaned fully into selling ads inside ChatGPT.
1. US government to vet who can use OpenAI’s and Anthropic’s most powerful AI models
The Trump administration struck a deal to bring Anthropic’s Mythos 5 back online for a small set of “trusted partners” and asked OpenAI to stagger the release of its new GPT-5.6 “Sol” model — with both companies now required to get government approval for each new customer during a cybersecurity review. For businesses, access to the most capable American AI is no longer just a purchasing decision; it now runs through Washington.
2. Oracle cuts 21,000 jobs and points to AI
Oracle disclosed it shed roughly 21,000 employees — about 13% of its workforce — over the past year, taking $1.8 billion in restructuring costs and warning in its annual filing that AI adoption “may continue to result” in further reductions. It is the clearest sign yet that a major enterprise vendor is reshaping its own headcount around the technology it sells.
3. Cheap Chinese AI models gain US customers as buyers move to cut their AI bills
A wave of powerful, low-cost Chinese AI models is winning over US customers, while large enterprises increasingly mix cheaper options into their stacks to rein in soaring Anthropic and OpenAI bills. With Washington tightening control over frontier American models, price and availability are pushing businesses toward a genuinely multi-vendor world.
4. OpenAI declares itself ‘in the advertising business’ at Cannes
At its first-ever Cannes Lions, OpenAI pitched brands on ChatGPT ads, revealing that about 20% of queries carry direct commercial intent and that it is expanding the four-month-old ad business into new markets as part of a reported $100-billion bet. For marketers, the AI assistant is fast becoming an ad channel — and a new place to compete for customer attention.
5. Qualcomm nears a $4-billion deal for AI chip startup Modular
Qualcomm is in advanced talks to acquire Modular, an AI software-and-chip startup, in a deal valuing it at roughly $4 billion, according to Bloomberg. The move would deepen Qualcomm’s push beyond phones into the AI data-center stack, where it is racing Nvidia, AMD, and a field of custom-silicon rivals.
6. AI data-center spending jumped 60% to over $300 billion and could reach $1 trillion
The world spent more than $300 billion on AI data centers in 2025 — a 60% jump over the prior year — and annual spending could approach $1 trillion within a decade, according to Intersect360 Research. The figures put hard numbers on the infrastructure build-out that underpins, and increasingly constrains, every company’s AI plans.
7. Perplexity launches Computer for Counsel, an AI agent built for legal teams
Perplexity released Computer for Counsel, an agentic, multi-model system that automates research and document-heavy workflows for legal teams, extending its broader Perplexity Computer product. It is the latest example of AI vendors moving past general chatbots to purpose-built agents aimed at a single high-value profession.
8. Canadian news and culture groups press Ottawa for copyright clarity on AI
After the federal government released its national AI strategy, Canadian news and cultural organizations are demanding clear rules on how their work can be used to train AI systems, warning that the strategy left copyright unresolved. The fight previews a broader question every Canadian organization will face: who owns the data that fuels AI.
9. AI chipmaker Groq confirms a $650-million raise after Nvidia’s $20-billion talent deal
Groq confirmed a $650-million funding round and is re-staffing its executive ranks after Nvidia’s roughly $20-billion “not-acqui-hire” deal earlier this year, leaning into its neocloud business serving fast AI inference. The raise shows investors still backing challengers betting they can run AI workloads cheaper than the incumbent GPU giants.
10. Startup Engram raises $98 million at a $600-million valuation with just 13 employees
Engram, an American AI startup with only 13 employees, raised $98 million at a $600-million valuation, underscoring how small, technical teams can now command large rounds in the AI market. The deal is a reminder that AI is rewriting the economics of company-building, where a handful of researchers can attract capital once reserved for far larger firms.
