Weekly AI News: Microsoft Polaris, Anthropic's $965B Valuation & SoftBank's Mega-Investment

Welcome to the CribConnects AI News Update. Every morning we bring you the most important AI developments for leaders and professionals making strategic decisions. Here are the five biggest stories of the week (May 26 – June 2, 2026).

1. Microsoft launches Project Polaris at Build 2026

On June 2, Microsoft opened its Build 2026 developer event in San Francisco with the announcement of Project Polaris — an in-house mixture-of-experts coding model that will replace GPT-4 Turbo as the default engine behind GitHub Copilot. Microsoft says Polaris performs "at or above" Anthropic's Claude 3.7 Sonnet on SWE-bench Verified at significantly lower inference cost. From August 2026, Copilot subscribers will be automatically migrated, with an optional 3-month fallback to GPT-4.

What this means: Microsoft is severing the OpenAI dependency for Copilot. For enterprise customers on Copilot, this means lower costs and tighter Azure integration — but also a new vendor lock-in to evaluate deliberately.

2. Anthropic raises $65B at a $965B valuation

This week Anthropic closed a $65 billion funding round at a $965B valuation — overtaking OpenAI for the first time, which was valued at $852B in March. The company also filed confidential S-1 paperwork with the SEC for a potential IPO ahead of OpenAI.

What this means: The frontier model market is shifting from a one-horse race to a real duopoly at the top, with serious competition on both model quality and enterprise distribution. A good moment for procurement teams to renegotiate commercial terms.

3. SoftBank commits €75B to French AI data centers

At the 2026 Choose France summit, SoftBank announced an investment of up to €75 billion ($87.5 billion) to develop 5 gigawatts of AI data center capacity in France. The first phase commits €45B to deliver 3.1 GW of capacity in the Hauts-de-France region by 2031. This is Europe's largest single AI infrastructure investment announced to date.

What this means: European organizations gain serious in-region compute capacity, materially simplifying GDPR and NIS2 compliance work. European AI sovereignty shifts from policy document to physical infrastructure.

4. GitHub Copilot moves to token-based billing

As of June 1, 2026, GitHub Copilot has moved from flat subscription pricing to token-based billing. Usage is now charged per request, similar to direct API pricing. Light users remain cheaper off; teams running agentic Copilot heavily will see per-developer costs rise.

What this means: Time to actively monitor Copilot spend and set up internal governance — comparable to cloud cost management. Organizations without an AI FinOps practice will feel this within the quarter.

5. Nvidia unveils RTX Spark and Vera Rubin at Computex

CEO Jensen Huang opened Computex Taipei on June 1 with the message that "AI is now a profit generator." Nvidia unveiled a new RTX Spark processor for Windows laptops and announced expanded production of its Vera Rubin AI platform. The message to Wall Street: AI is no longer a capex story but a measured return category.

What this means: Local AI workloads on laptops become seriously viable for enterprise scenarios with sensitive data. For IT strategy: revisit your 2027 hardware refresh plan.

What does this mean for your organization?

This week marks a shift: AI infrastructure, model competition and commercial models are all maturing at the same time. The question for leaders is no longer whether AI should be deployed strategically, but how to align vendor choices, cost structure and governance with that maturity.

At CribConnects we help organizations translate developments like these into concrete roadmaps. Want to discuss what this week means for your context? Reach out at info@cribconnects.com.