The Boardroom Brief

Tuesday, May 6, 2026  ·  Issue #19

OpenAI has restructured its landmark Microsoft partnership, ending the exclusivity arrangement that locked its models behind Azure. Microsoft retains right of first refusal and royalty-free access through 2032, but OpenAI is now free to deploy on AWS, Google Cloud, and other hyperscalers. GPT-5.5, launched this week, is already live on AWS — and its API revenue is growing 2× faster than any prior OpenAI release.

Exec angle: The walled-garden era of enterprise AI is cracking. If you standardized on Azure AI because it was the only door into OpenAI, that logic no longer holds. Multi-cloud AI is now the default posture — and your procurement and architecture teams need to re-evaluate vendor lock-in assumptions made 18 months ago.

Q1 2026 was the worst quarter for tech layoffs in over two years — 81,747 roles eliminated across the industry. Meta is cutting another 8,000 starting May 20. Microsoft is offering buyouts to 7–8% of its U.S. workforce. Amazon, Intel, and Salesforce have all reduced headcount significantly. Meanwhile, every one of these companies is reporting strong earnings. Big Tech's combined capex in 2026 is tracking toward $715 billion — nearly 3× what it was in 2024 — almost entirely directed at AI chips, data centers, and infrastructure. Meta alone is committing up to $145B in AI capex this year.

Exec angle: This isn't a signal that tech is struggling — it's the opposite. Headcount is being converted into compute. The firms doing this are reporting record revenues. Any board conversation about "AI investment" should include this benchmark: the world's largest operators are betting $715B this year alone. What's your AI infrastructure multiple against revenue?

On May 1, the Department of Defense signed classified AI contracts with SpaceX, OpenAI, Google, Nvidia, Reflection AI, Microsoft, and AWS — granting access to deploy AI models into its most secure classified networks (IL6/IL7) for "any lawful use." The DoD's GenAI.mil platform already has 1.3 million users running hundreds of thousands of AI agents, compressing workflows from months to days. The FY2026 defense AI budget stands at $33.7 billion, with an autonomous warfare unit receiving a budget increase described as 24,000% over prior levels. Anthropic was explicitly excluded after refusing to drop its "lawful-use" ethical standards.

Exec angle: The government is now one of the most aggressive AI deployers on earth. For enterprise leaders, this has two implications: (1) the vendors on this list now have classified-grade deployment experience, which will filter into commercial products fast; and (2) Anthropic's exclusion signals that "safety-forward" positioning has a cost in certain markets — a tension every AI procurement team will need to navigate.

Apple has quietly wound down Vision Pro development, disbanding parts of the hardware engineering team after M5 refresh sales underwhelmed. Resources are being redirected to AI-optimized Siri improvements and a lighter, more affordable glasses product to compete directly with Meta's Ray-Ban line. Simultaneously, Apple raised the Mac mini's base price from $599 to $799 by eliminating the 256GB model — citing surging demand for local AI workloads and chip supply constraints. Mac revenue hit $8.4B in Q2 2026, up 6%, driven by on-device AI use.

Exec angle: Apple's retreat from Vision Pro is a clear market signal: immersive spatial computing is not yet a mainstream enterprise interface. The real hardware battle is moving to AI-powered glasses at consumer price points — and the price hike on Mac mini confirms that on-device AI inference is driving a new hardware premium cycle. Plan for rising hardware costs in your AI deployment budgets.

Meta terminated over 1,100 AI data trainers — primarily contractors — after discovering they had leaked unreleased product details related to the Ray-Ban smart glasses program. The workers had access to confidential training data and product roadmaps as part of their annotation and RLHF work. The scale of the terminations — spanning multiple vendors — signals that Meta's internal investigation was broader than initially reported.

Exec angle: As AI training pipelines touch more proprietary data and roadmap material, the contractor and vendor workforce handling that data becomes a strategic security perimeter — not just an operational one. Your AI data governance policy needs to treat annotators and trainers with the same access controls as software engineers. Most companies haven't made that adjustment yet.

Elon Musk's lawsuit alleging OpenAI "betrayed its mission" by converting to a for-profit structure has entered active trial testimony. Musk testified for three-plus days this week, continuing his claim that OpenAI violated the founding nonprofit charter. In a notable twist, Musk's own attorneys acknowledged that xAI distilled models from OpenAI outputs — a common industry practice, but one that undercuts the moral authority of the betrayal argument. Meanwhile, OpenAI's CFO denied separate reports that the company is missing internal revenue and user targets, citing "a vertical wall of demand."

Exec angle: The trial's real value isn't the verdict — it's the discovery. Testimony and filings are producing the clearest public picture yet of how AI labs are actually governed, how board oversight works (or doesn't), and what "mission alignment" means when billions of dollars are on the table. Pay attention. These governance patterns will eventually become regulatory standards.

Click to subscribe to our weekly newsletter and additional content
for less than $1 / week

You’re receiving this because you signed up at theboardroombrief.news

Free Tier: Monthly Issues, up to four briefs