Tech-Reader AI Digest: Weekly Recap for May 4-8, 2026
.jpeg)
Tech-Reader AI Digest: Weekly Recap for May 4-8, 2026
Saturday, May 9, 2026
#AI #TechNews #Digest
This was the week a federal courthouse in Oakland started rewriting the history of modern AI — and the industry kept moving anyway.
Monday Greg Brockman took the stand. His stake in OpenAI: nearly $30 billion. His investment in OpenAI: $0. His own journal called the nonprofit commitment "a lie." Musk's attorney asked the same question twelve ways: "It takes $30 billion to get you out of bed in the morning, but $1 billion doesn't?" The answer "that's not what I'm saying" will be read back in closing arguments.
Before the session started, Musk had texted Brockman two days prior to gauge settlement interest. Brockman counter-proposed dropping claims against individuals. Musk's reply: "By the end of this week, you and Sam will be the most hated men in America." The exchange was ruled inadmissible. OpenAI's lawyers filed it publicly anyway.
Anthropic launched its enterprise joint venture the same morning — Blackstone, Goldman Sachs, and Hellman & Friedman as founding partners, valued at $1.5 billion, built on the Forward-Deployed Engineering model. Hours later, Bloomberg reported OpenAI was building The Deployment Company at $10 billion with 19 investors. Same logic, larger scale. The copycat pattern, now fully institutionalized.
Cerebras filed its IPO paperwork — $3.5 billion raise, $26.6 billion valuation — with OpenAI's $20 billion compute commitment as its anchor. Brockman holds a stake in Cerebras. The federal courthouse felt very small on Monday.
Tuesday Brockman returned to the stand and went on offense. Musk demanded $80 billion and full control of OpenAI — control he said he would decide when to relinquish. Brockman called it the Mars Motivator: Musk needed majority authority over OpenAI to fund a city on Mars. He also testified that Musk had OpenAI employees performing secret work on Tesla's self-driving technology while still sitting on the board — a direct counter-accusation of fiduciary breach.
The dramatic peak of the week arrived Tuesday: an August 2017 meeting that began with Ilya Sutskever's painted Tesla portrait as a gift. Musk grew angry about equity, said "I decline," stood up — and Brockman braced for impact. Instead, Musk grabbed the painting and stormed out. The jury will remember that sentence in deliberations.
Anthropic released ten pre-built AI agent templates for Wall Street — KYC screening, pitchbook creation, financial modeling. Direct output of the Blackstone/Goldman venture announced Monday.
Palantir posted its strongest quarter in company history: $1.63 billion in revenue, 85% growth year-over-year, net income quadrupled to $870 million, $1.5 million revenue per employee. The forward-deployed engineering model in its most mature form — and the scorecard Anthropic and OpenAI are now chasing.
Wednesday Mira Murati delivered the trial's most important third-party testimony. She had no stake in Musk winning. She was fired alongside Altman, returned when he was reinstated, then left voluntarily. Under oath, she described Altman "creating chaos" — saying one thing to one person, the opposite to another — and specifically testified that Altman misled her about safety protocol clearances on an unreleased model. Legal analysts are calling it the Candidness Gap: the bridge that connects the "not consistently candid" language the board used in 2023 to actual documented executive deception.
The November 2023 firing — dismissed by the world as a failed coup — is starting to look like a governance body that had been building a case for months.
Anthropic announced it had signed an agreement with SpaceX to use all compute capacity at Colossus 1 in Memphis — 300 megawatts, 220,000 GPUs. Claude Code's five-hour rate limits doubled the same day. The irony was complete: Musk, who had called Anthropic "misanthropic" in February, posted "No one set off my evil detector" and confirmed xAI would dissolve and rebrand as SpaceXAI. Analysts called the shift Exascale Diplomacy — business logic overriding personal rivalry at the exact moment SpaceX needs a marquee AI customer for its IPO.
Microsoft reported Q3 earnings: AI business at $37 billion annual run rate, up 123% year-over-year, Copilot at 20 million paid seats. Copilot Cowork went mobile, with Claude Opus 4.7 selectable inside it. Microsoft designed Cowork in close collaboration with Anthropic. The alliances, the rivalries, the infrastructure — all of it restructured on a single Wednesday in May.
Thursday three independent witnesses arrived in sequence — a former safety researcher, a former board member, and a nonprofit law expert — and all three described the same pattern.
Rosie Campbell testified her AGI Readiness team was disbanded in early 2024, the same period the Super Alignment team was shut down. Tasha McCauley testified to "dozens of pages" of documented examples of executive dysfunction compiled by Ilya Sutskever — now being called the Sutskever Dossier. She also confirmed the board was not informed prior to the public launch of ChatGPT. David Schizer testified that Microsoft tested a version of GPT-4 without going through OpenAI's safety review — directly echoing the Scott Memo from March 2018.
Anthropic committed $200 billion to Google Cloud over five years starting 2027 — representing more than 40% of Google Cloud's total revenue backlog. Contracts involving Anthropic and OpenAI now account for more than 50% of the $2 trillion in backlogs across AWS, Azure, and GCP. Alphabet's market cap reached $4.8 trillion, closing the gap with Nvidia to under $300 billion. The Vertical Extraction strategy — owning the chips, the cloud, and the investor stake simultaneously — was the architecture driving it.
Friday Claude entered Microsoft 365. Add-ins for Word, Excel, and PowerPoint went generally available. Outlook entered public beta. The headline feature: cross-app context continuity — one continuous session across four applications, no re-prompting, no copy-paste.
Copilot is a capable product with 18 months of enterprise momentum behind it. Claude's entry into the same four applications doesn't erase that — it raises the bar. The knowledge worker AI race just got more interesting.
The same day, TCI Fund Management — led by Sir Christopher Hohn, a nine-year Microsoft bull — cut its Microsoft position from 10% of its portfolio to 1%, liquidating nearly $8 billion in stock. In a letter to investors, Hohn cited AI disruption to the Office productivity franchise and risk to Azure. TCI simultaneously raised its Alphabet stake to 5% — now its largest technology position. Duquesne and Tiger Global made similar moves.
The hedge fund's exit thesis and the product announcement arrived simultaneously, from opposite directions, saying the same thing.
No court Friday. The trial resumes Monday with two weeks of sworn testimony now permanently part of OpenAI's public record — independent of whatever the jury decides.
The week closes with the industry reshaped along lines that weren't visible seven days ago. The circular infrastructure economy is real. The forward-deployed engineering model is now the consensus enterprise strategy. The knowledge worker AI race has a second horse. And a federal courthouse in Oakland has produced a documentary record that will be read by IPO investors, due diligence teams, and historians long after the verdict is announced.
The OpenAI trial resumes on Monday. See you then. — Aaron
Aaron Rose is a software engineer and technology writer at tech-reader.blog.
Catch up on the latest explainer videos, podcasts, and industry discussions below.
