Morning Digest, May 15, 2026
May 15, 2026 · 6 newsletters · 5 overlapping stories
Top Stories
OpenAI-Apple Alliance Frays, Legal Fight Looms
(3 newsletters)
Two years after their Siri integration deal, OpenAI is reportedly preparing legal action against Apple. The company has enlisted outside counsel to explore options including a breach-of-contract notice, citing Apple’s failure to make an “honest effort” within the partnership. OpenAI had projected billions in paid ChatGPT subscriber conversions through Apple’s ecosystem, but internal data showed users strongly prefer the standalone app. Tensions are compounded by Apple’s plans to open Siri to rival providers (including Anthropic and Google) at WWDC on June 8, and by OpenAI’s emerging hardware ambitions through its Jony Ive deal.
OpenAI’s Codex Escapes the Desktop
(3 newsletters)
OpenAI rolled out Codex inside the ChatGPT iOS app for all users, including the free tier, letting developers monitor, approve, and kick off long-running AI coding tasks from their phone while the agent works on a desktop or remote host. A secure relay layer keeps the user’s computer off the open internet. The timing is pointed: Anthropic introduced Remote Control in February and Dispatch in March, and OpenAI’s blog copy seemed to address Anthropic users directly. The days of leaving a laptop open in coffee shops to keep an agent running appear to be numbered.
Anthropic’s Billing Overhaul Triggers Dev Backlash
(2 newsletters)
Starting June 15, Anthropic will pull Agent SDK, GitHub Actions, and third-party app usage out of subscription limits and into a separate monthly credit pool billed at full API rates. Pro subscribers get $20/month in agent credits; Max 20x gets $200. The change reverses April’s outright ban on third-party agents but removes the subsidized compute that made plans attractive for power users. High-profile developers, including T3’s Theo Browne, publicly cancelled their subscriptions, and the backlash arrives exactly as OpenAI is raising Codex limits to lure switchers.
Cerebras IPO Nearly Doubles on Debut
(2 newsletters)
AI chipmaker Cerebras Systems raised $5.5 billion in its Nasdaq debut, opening at $350 per share after listing at $185 for a $100 billion valuation. It is the largest tech IPO of the year so far, fueled partly by a major cloud deal with OpenAI struck in January. The blockbuster debut may signal a clear runway for anticipated public offerings from OpenAI, Anthropic, and SpaceX later in 2026.
Netflix Quietly Builds a GenAI Animation Studio
(2 newsletters)
Netflix launched “INKubator” in March, an artist-led incubator for AI-native animated content. The studio is staffing up to produce shorts and specials using fully GenAI-native production pipelines, framing the effort as a tool for storytellers rather than a replacement for them.
Also Worth Knowing
-
Warsh Replaces Powell as Fed Chair. Kevin Warsh was confirmed as the 17th Federal Reserve chair by the slimmest margin in history, ending Powell’s 8-year run with inflation still at 3.8%. Warsh has drawn parallels between AI-driven productivity and the 1990s computing boom, suggesting it could support rate cuts without fueling inflation.
-
Anthropic and Gates Foundation Commit $200M. The two organizations will deploy AI across global health, education, and economic mobility programs over four years. The Gates Foundation brings grant funding and program design; Anthropic contributes Claude usage credits and technical staff.
-
Claude for Small Business and Legal Launches. Anthropic shipped 15 ready-to-run agentic workflows for small business owners, plugged into QuickBooks, PayPal, and DocuSign, alongside 12 one-click workflows tailored to legal professionals.
-
The Forward-Deployed Engineer Race. Anthropic, OpenAI, and Google are all competing to embed engineers directly inside enterprise customer orgs, following Palantir’s FDE playbook. Once an FDE team spends months integrating AI into a firm’s workflows, the stack becomes nearly impossible to remove.
Quick Hits
-
/goal is becoming a standard primitive for AI coding agents. Both Claude Code and Codex CLI now let you write a “done state” and have the agent work until it hits it.
-
xAI launched Grok Build, an agentic CLI in beta, currently limited to SuperGrok Heavy subscribers.
-
Notion opened its Developer Platform: custom Workers in a secure sandbox, free through August on Business and Enterprise plans.
-
AT&T, T-Mobile, and Verizon agreed to a joint venture to pool satellite deals and spectrum and eliminate wireless dead zones in rural and underserved areas.
-
Honda posted its first annual loss since 1957, a $2.7B hit from failed EV investments. It is ditching full-electrification targets and pivoting to 15 new hybrid models by 2030.
-
A Bitcoin trader recovered $400K using Claude after losing a wallet password 11 years ago. Claude found an old backup file and decrypted it.
-
Higgsfield launched “Supercomputer”, a chat-based agentic creative pipeline that takes a plain-language brief and produces a finished asset without prompts or tool-hopping.