Tech-Reader AI Digest

Friday, May 1, 2026

#AI #TechNews #Digest

Story 1: The Pentagon Signs Classified AI Deals With Eight Companies — The Vendor Lock-In Hedge, and Anthropic Conspicuously Absent

What happened: Seven leading artificial intelligence companies have reached deals to deploy their technology in classified Pentagon computer networks, the Defense Department said Friday. The agreements — which include Microsoft, Amazon Web Services, OpenAI, Google, Nvidia, SpaceX, and the startup Reflection AI — will give those firms' AI systems access to the military's most classified network environments: Impact Level 6 and Impact Level 7. IL7 represents the highest tier of cloud security, covering Top Secret data — a massive milestone for OpenAI and Google. Oracle was subsequently added, bringing the total to eight companies. (Source: Washington Post / GeekWire / TechCrunch / Breaking Defense / CNN / Nextgov)

The Pentagon says the effort is already well underway. More than 1.3 million Defense Department personnel have used GenAI.mil, the military's official AI platform, generating tens of millions of prompts and deploying hundreds of thousands of agents in just five months. Officials say the technology has cut some tasks from months to days.

The Pentagon's own statement explains the strategic logic explicitly — this is a Vendor Lock-In Hedge: "The Department will continue to build an architecture that prevents AI vendor lock-in and ensures long-term flexibility for the Joint Force." By diversifying across eight providers simultaneously, the Pentagon is physically constructing an infrastructure that makes any single company's ethical red lines irrelevant to their operations.

Pentagon CTO Emil Michael took a thinly veiled shot at Anthropic on CNBC: "What we've learned since we started this effort at the Department of War is that it's irresponsible to be reliant on any one partner, and we learned that that one partner didn't really want to work with us in the way we wanted to work with them. So we went out and made sure that we had multiple different providers." Defense Secretary Pete Hegseth called Anthropic CEO Dario Amodei an "ideological lunatic" during a Senate Armed Services Committee hearing Thursday. Hundreds of Google employees simultaneously sent a letter to company leadership urging them to refuse Pentagon use of its AI on classified data.

The timing is layered: the White House is simultaneously drafting an Administrative Offramp to bring Anthropic back into the federal fold — reported yesterday — while the Pentagon announces eight classified AI deals that pointedly exclude Anthropic today. The two branches of the executive are pulling in different directions simultaneously. (Source: Washington Post / GeekWire / TechCrunch / Breaking Defense / CNN / Nextgov)

Why it matters: Until recently, Anthropic's Claude was the only AI model available in the Pentagon's classified network. The eight-company Vendor Lock-In Hedge changes that permanently. Even if the White House Administrative Offramp succeeds in bringing Anthropic back, the Pentagon has already built the infrastructure to operate without them. Anthropic is missing out on substantial classified contract revenue that eight competitors now have access to.

Aaron's take — "Ideological lunatic." That's the Defense Secretary calling the CEO of the company the NSA is already using for cybersecurity. The Pentagon and the White House are running contradictory Anthropic strategies simultaneously — one branch is building an offramp, the other is calling Dario names in a Senate hearing. That's not a unified government position. That's a negotiation in public with the volume turned up.

Story 2: Apple Posts Its Best March Quarter Ever — $111.2 Billion, Capital-Light AI, and a $100 Billion Buyback

What happened: Apple announced financial results for its fiscal Q2 2026 ended March 28. The company posted quarterly revenue of $111.2 billion, up 17% year over year, with diluted EPS of $2.01, up 22%. "Today Apple is proud to report our best March quarter ever, with revenue of $111.2 billion and double-digit growth across every geographic segment," said Tim Cook. (Source: Apple SEC Form 8-K / CNBC / 9to5Mac / MacRumors)

iPhone revenue reached $56.99 billion — a March quarter record — with sales jumping 22% year over year for the second consecutive quarter of 20%+ growth, driven by extraordinary iPhone 17 demand. Apple's Services business set a new all-time record at nearly $31 billion, growing 16.3%, beating Wall Street expectations. Gross margin expanded to 49.3% versus the expected 48.4%. The board authorized an additional $100 billion in stock repurchases and declared a cash dividend of 27 cents per share, up 4%. Q3 guidance of 14-17% growth crushed the analyst expectation of 9.5%. Apple stock rose approximately 5% Friday.

The strategic contrast with peers is striking. Apple spent only $4.3 billion in capex in the first half of fiscal 2026 — while Microsoft is running at $190 billion annually, Amazon at $200 billion, and Google similarly. Apple is practicing Capital-Light AI: achieving record profits and AI delivery through on-device silicon and ecosystem services, without the massive speculative infrastructure bets of its peers. On the earnings call, Cook said the collaboration with Google on Apple Intelligence "is going well." (Source: Apple SEC Form 8-K / CNBC / 9to5Mac / StockTitan / TidBITS)

Why it matters: Apple doesn't need to win the cloud AI race to benefit enormously from it. The iPhone is the on-device AI delivery mechanism for hundreds of millions of people. The Capital-Light AI strategy — $100 billion buyback instead of $100 billion capex — is a fundamentally different bet on where AI value accrues: at the device and ecosystem layer, not the infrastructure layer. This week's earnings confirmed both strategies are working simultaneously.

Aaron's take — Every major tech company that reported earnings this week beat estimates. Microsoft. Amazon. Alphabet. Apple. Meta. The AI infrastructure buildout is generating real revenue across the entire stack — chips, cloud, devices, services, advertising. The "show me the ROI" crowd has had a very bad week.

Story 3: Trial Week 1 Closes — The Shivon Zilis Revelation Changes Everything

What happened: As the first week of the Musk v. Altman trial concludes — with no proceedings Friday and the jury under strict instructions not to discuss or research the case over the long weekend — the most consequential revelation of the week didn't come from Musk's four days of testimony. It came from the documents introduced around Shivon Zilis.

According to testimonies and evidence presented in court, Zilis's most crucial role in the context of the lawsuit is that of a covert liaison between OpenAI and Musk, particularly after he stepped down from the nonprofit's board in February 2018. A telling text message from Zilis to Musk on February 16, 2018, illustrates this duality: "Do you prefer I stay close and friendly to OpenAI to keep info flowing or begin to disassociate?" Musk responded, emphasizing the importance of staying "close and friendly," while also expressing a desire to move OpenAI personnel to Tesla.

When queried about his relationship with Zilis in court, Musk provided multiple descriptions — "chief of staff," then "close adviser," then "we live together, and she's the mother of four of my children." Zilis clarified in a deposition that Musk is more of a frequent visitor and maintains his own residence.

The legal implication is double-edged. The Zilis texts demonstrate Musk had a trusted informant inside OpenAI for years after his official departure — undermining his claim that he was kept in the dark about the for-profit conversion. They also demonstrate he was actively working to poach OpenAI talent for Tesla using Zilis as cover — supporting OpenAI's claim that Musk wanted control, not fidelity to the nonprofit mission.

Week 2 opens Monday with Greg Brockman and UC Berkeley AI safety professor Stuart Russell on the witness list. (Source: Wired / CNBC / Altagic / TechBuzz AI)

Why it matters: The Zilis texts are the most damaging evidence of week one — not for OpenAI, but for Musk. You cannot simultaneously be the insider who was getting active information flows and the outsider who was kept in the dark. The jury will spend the weekend with that contradiction.

Aaron's take — "Do you prefer I stay close and friendly to OpenAI to keep info flowing?" That text was written the same week Musk left the OpenAI board. He didn't walk away. He left a channel in place. OpenAI's lawyers introduced that text deliberately and carefully. It will be the last thing the jury thinks about before Monday morning.

Quick Hits — The Rest of Today's AI World

Anthropic / Claude

Anthropic absent from today's Pentagon classified AI deals — see Story 1. White House Administrative Offramp still in development. Anthropic $900B valuation talks ongoing. October IPO trajectory unchanged. (Source: TechCrunch / Axios)

Gemini (Google)

Google signed Pentagon classified network deal today — IL7 Top Secret access confirmed. Gemini 3.1 Pro already rolled out on GenAI.mil in late April. Hundreds of Google employees sent internal letter opposing the Pentagon deal. (Source: Nextgov / Washington Post)

VS Code / GitHub Copilot

No new announcements. Usage-based billing June 1 transition remains standing news. (Source: GitHub)

Replit

No new announcements. (Source: Replit)

Perplexity

No new announcements today. (Source: Perplexity)

Microsoft Copilot

Microsoft signed Pentagon classified network deal today — IL7 Top Secret access. Q3 earnings from Wednesday remain standing news. AI at $37B run rate. (Source: Breaking Defense / Microsoft)

xAI / Grok

SpaceX signed Pentagon classified network deal today — xAI capabilities now accessible in classified DOD environments at IL7. Musk v. OpenAI trial resumes Monday with Brockman and Stuart Russell expected to testify. (Source: Breaking Defense / CNBC)

Nvidia

Nvidia signed Pentagon classified network deal today — see Story 1. Reflection AI, a Nvidia-backed startup, also included — see below. (Source: TechCrunch / Breaking Defense)

Reflection AI

Reflection AI — a newer AI startup backed by Nvidia — was included in today's Pentagon classified network deals alongside the industry giants. Its Asimov coding agent is being positioned by the Pentagon as a government-preferred alternative to Claude Code for classified software development environments. The inclusion of a startup in a deal alongside Microsoft, Google, and OpenAI is an extraordinary endorsement — and a direct signal that the government is actively building Claude Code alternatives in environments Anthropic cannot currently access. (Source: Breaking Defense / TechCrunch)

OpenAI

OpenAI signed Pentagon classified network deal today at IL7 Top Secret level. GPT-5.5 Cyber Trusted Access program from yesterday remains active. OpenAI smartphone development per analyst Ming-Chi Kuo — targeting Q1 2027 specs, 2028 mass production. (Source: GeekWire / NeuralBuddies)

DeepSeek

DeepSeek V4-Pro and V4-Flash have been live in preview since April 24 — available via the official website, API, and mobile apps. The web and mobile chatbot defaults to V4 with Expert Mode (V4-Pro) and Instant Mode (V4-Flash). Legacy deepseek-chat and deepseek-reasoner endpoints retire July 24, 2026. Standing context: 1.6 trillion parameters, 1M token context, runs natively on Huawei Ascend 950PR — Sovereign Parity confirmed. (Source: Simon Willison / ChinaTalk / WaveSpeed / Gizchina)

Huawei / China AI

Huawei expects AI chip revenue to hit ~$12 billion in 2026, up 60% from $7.5 billion in 2025, as orders for its Ascend 950PR surge. This is Inference Localization in action — Chinese firms are no longer waiting for Nvidia export controls to ease. They are building a sovereign compute market from the ground up, with DeepSeek V4 as its flagship proof of concept. (Source: Financial Times)

Alibaba / Qwen / Z.ai

No new announcements today.

Inflection Pi / Mistral

No major news today.

That's your AI world for Friday, May 1. Welcome to May. Trial resumes Monday. Back then.

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.

Search This Blog

Tech-Reader.blog