Argument and analysis. Strong takes on the day's news, written by humans, signed by name.
98 stories · sorted by most recent
Stratechery outlines Meta's need for Walmart to endure Amazon competition, Expedia's effort to retain its intermediary status, and uncertainty around Google's next moves.
Amazon blocked Muse in a move that treats warehouses and logistics networks as core AI assets, though the two sides still have space to negotiate.
Stratechery notes that deliberate pacing at the AI frontier may serve both stated caution and the practical need for labs to close gaps created by fast model gains.
WorkOS is using a Daring Fireball sponsorship to direct developers toward its service for adding single sign-on instead of coding SAML flows themselves.
The update gives developers initial tools to resize and lay out apps for the dual-screen device.
Apple placed the iOS SDK and simulator support for the iPhone Duo in a future Xcode 27.1 release rather than the current 27.2 update, leaving developers without tools to begin adaptation.
OpenAI's ad tests show results while Walmart ends its holdout against Apple Pay.
Stratechery contends that the Anthropic CEO’s call to slow frontier development is unrealistic and functions mainly as a bid for political oversight of the technology.
Paul Ford argues that AI coding tools have made bad software easier to produce without delivering the promised replacement for skilled developers.
OpenAI delivers an impressive mathematical result with narrow reach, while Meta's new personal agent points toward wider daily use.
WorkOS Relay stores access tokens at the provider rather than handing them to agents, limiting exposure if a session is compromised.
Adobe has appointed its head of customer experience orchestration and field operations to succeed Shantanu Narayen as chief executive on December 1.
OpenAI has begun a limited release of GPT-6 Astra to select organizations, with access expanding over the coming days to ChatGPT subscribers and API customers at the same token pricing as Claude Fable 5 and 5.1.
Stratechery conversation with OpenAI’s president covers the company’s founding decisions, the Astra project, and the practical weight of steering toward safe advanced systems.
Apple’s new cellular feature pairs a primary iPhone with a companion unit while preserving a single phone number and centralizing all account controls on the main device.
Nvidia reported results that combined clear market dominance with predictable outcomes, all while steering toward an open AI supply chain.
A California court ruled the Defense Department’s designation of the AI company as a national security risk exceeded statutory authority.
An analysis of Claude-generated pull request descriptions shows eight fixed templates now dominate output, with one rising from 1 percent to 45 percent of the corpus in eighteen months.
Apple updates the Mac Mini and Mac Studio while OpenAI introduces a new AI computer, with both moves applying direct pressure on Nvidia.
European regulators accept revised Apple business terms but leave open whether the changes satisfy the Digital Markets Act.
Apple will continue issuing new Hide My Email addresses under icloud.com while moving future Sign in with Apple relay addresses to private.icloud.com.
WorkOS now lets apps publish an auth.md file so autonomous agents can complete registration and receive scoped credentials instead of abandoning human login flows.
OneDrive now lets users start a new Markdown file from the same menu that already offered Word, Excel, and PowerPoint.
Beta 6 brings more detailed buttons that echo older Mac OS X designs, part of a broader shift in interface direction.
Apple insists the July ruling granting access to federal procurement records must stand in the U.S. v. Apple antitrust matter.
Nvidia extends financing to an OpenAI facility, Anthropic maintains strong revenue growth, and Google purchases data assets from Spirit Airlines.
Anthropic is implementing steganographic changes to token selection across all Claude models to mark AI-generated text, drawing sharp criticism for distorting natural writing to meet EU rules.
Commerce Secretary Howard Lutnick says the U.S. government has directly informed Apple that sourcing RAM from Chinese makers is not acceptable.
Stratechery's weekly roundup frames persistent capital spending as the central constraint shaping AI development.
Anthropic's watermarking plan, introduced to satisfy European rules, exposes fundamental conflicts over how AI systems should represent their own outputs.
Nvidia is arranging fresh ways for its buyers to fund purchases, shifting more of the AI buildout risk onto the broader market.
Apple delivers the next preview of its 2026 iPhone software with targeted design revisions and broader on-device adjustments for AI voices.
WorkOS says REST handles human developers while MCP addresses what agents actually need, with the two working as stacked layers rather than alternatives.
John Gruber withdraws an entire post on an App Store rejection after determining its core premise was incorrect, and publishes the original text for public review.
Leaked audio shows the consulting giant confronting runaway costs from non-technical staff using AI for mundane chores like PDF conversions, undermining claims that elite engineering drives the spending surge.
Demis Hassabis is relinquishing day-to-day control of Google DeepMind to become its chair and Alphabet’s chief scientist, citing the near-term arrival of AGI.
Developers report inconsistent approval times, with some apps cleared quickly while others linger, amid claims of an AI-fueled increase in submissions.
OpenAI submits a direct court filing that attacks the substance of Apple’s complaint rather than procedural points.
Stratechery analysis finds Microsoft’s results reflect focused execution and efficiency gains that stand out against Meta’s heavier spending.
The Register asserts the AI bubble is already deflating, citing strange earnings patterns discussed only on its podcast.
John Gruber’s archived post argues that Electron apps erode Mac quality even as the platform gains users who do not prize native experience.
John Gruber’s 2018 post, resurfaced in the Daring Fireball archive, argues that Electron apps erode the Mac’s core advantages even as the platform’s user base expands.
Anthropic released a desktop Claude app built on Electron that fails to match the polish of native Mac software like ChatGPT.
Anthropic’s first desktop application for Claude arrived as an Electron package that ignored the frameworks its own model recommended for Mac development.
Ookla’s RootMetrics and Speedtest data record measurable growth in millimeter-wave 5G deployments across American networks while most countries continue to prioritize mid-band spectrum.
Meta’s combination of keystroke tracking, token counts, and a planned 10 percent cut is steering its engineers toward maximizing visible AI output instead of shipping code.
A flaw in the iCloud privacy feature can reveal the underlying email address it is supposed to conceal, and Apple has not fixed it despite a report filed over a year ago.
Apple has not fixed a flaw in its iCloud Hide My Email service that lets third parties recover the underlying address tied to an alias.
An app meant to simplify entry to Spanish cannabis clubs left users’ passports, addresses, and consumption records open to anyone who looked.
Effective August 1, 2026, Microsoft will increase prices on its remaining Xbox console models and remove the 2 TB variant from sale.
U.S. senators from both parties warned intelligence officials that Apple’s potential purchase of chips from a Chinese state-owned firm would create supply-chain security risks.
WebKit enables the Copy menu item in Safari, Mail, and third-party apps even when nothing is selected on the focused page or message.
The three leading memory chip companies risk creating stronger competitors in China, while Microsoft has clear commercial reasons to adopt Chinese AI models.
Apple is increasing product prices while declining to ship its Siri-based AI tools to users in the European Union.
Apple's shift to a single domain for its privacy aliases removes the technical friction that once made broad blocks costly for services.
Apple’s move to a single new domain for both services removes the previous friction that kept services from blocking iCloud aliases without also cutting off regular iCloud users.
Anthropic altered its Claude models to enforce stricter limits on surveillance and weapons applications, even after the Department of War sought broader legal access.
Anthropic altered its models to enforce stricter limits on surveillance and weapons applications after clashing with the Department of War, then extended similar restrictions to competitors building frontier systems.
Anthropic changed its models after a dispute with the Department of War, then used the changes to restrict access for competitors and government users alike.
Anthropic disables two models worldwide after an export control directive bars any foreign national from using them.
Access to Apple’s frontier models on Private Cloud Compute requires enrollment in the App Store Small Business Program and fewer than two million first-time downloads.
John Gruber hosted a special live recording of The Talk Show with Joanna Stern and Nilay Patel to break down Apple’s announcements.
Stratechery's June 8, 2026 edition notes Apple's Intelligence rollout, Anthropic's fable, and questions around European industry.
The European Commission states that Apple chose not to launch its AI-enhanced Siri in the EU after requesting an 18-month exemption from DMA interoperability rules, a request the regulator rejected.
Apple omitted the keynote from downloadable sessions in its Developer app, directing viewers to YouTube for an offline copy.
Apple says EU rules would force it to hand any AI system broad, autonomous access to user devices without ongoing controls.
Ben Bajarin spoke with Stratechery about WWDC and the current state of the AI compute industry.
The developer conference keynote ran shorter than recent editions and closed with an unannounced music video.
The presentation covered expected updates to iOS 27 and macOS 27 along with AI features and finished with an Easter egg music video.
Stratechery's latest piece argues that Siri remains adequate for most users even if it trails current AI standards.
Mux Robots apply automated AI tasks to new video files for summarization, translation, and moderation.
Stratechery's June 2026 weekly edition flags Google and Microsoft as central to ongoing industry realignments.
The annual audience episode returns to the California Theatre on June 9 at 7pm PT, with tickets priced at $45.
The transaction underscores rising demand for capital as the decisive input in technology markets.
Meta is now offering monthly subscriptions on its three largest consumer apps, starting at $2.99, with extra profile tools and reactions as the initial draw.
Stratechery concludes that no credible valuation supports taking SpaceX public, yet the technical possibility of orbital data centers may still draw capital.
Nvidia will now break out revenue from the largest cloud providers separately from all other customers, drawing a line between segments where its chips face commoditization and segments where it still controls the full stack.
WorkOS now offers pre-built connectors that handle OAuth, token refresh, and API calls so agents can read records from the services users already rely on.
The Ninth Circuit opinion that Apple now wants the Supreme Court to overturn addressed whether an injunction on App Store commissions should cover only Epic Games or extend to all U.S. developers.
Apple filed a petition asking the justices to decide whether a lower court properly held it in contempt and whether the injunction against the App Store was drawn too broadly.
Andrej Karpathy announced he has joined Anthropic to focus on core research after leaving his AI education startup.
Andrej Karpathy has left his AI education startup to join Anthropic, citing the next few years of large language model development as especially formative.
Andrej Karpathy has joined Anthropic to resume hands-on research on large language models after founding an education startup last year.
A California jury found Elon Musk waited past the three-year limit to sue Sam Altman and OpenAI over the company’s shift to a for-profit model.
Local resistance to new data centers stems from real burdens, and direct compensation offers the clearest way to resolve it.
arXiv will suspend authors for a full year if their submissions contain clear evidence of unverified generative AI output, according to the platform’s computer science section chair.
The preprint server will suspend authors who include LLM output without verification, treating such papers as untrustworthy by default.
OpenAI has ended the interim arrangement and given cofounder Greg Brockman formal control of product strategy while it continues to merge its separate offerings.
Apple executives told John Gruber that the company ships experiences, not raw technology, pushing back on calls for a standalone “killer AI” product.
arXiv will suspend authors for twelve months when their submissions show clear signs of unverified large-language-model output.
OpenAI has made Greg Brockman’s oversight of product strategy official as the company reorganizes to combine its main offerings under one structure.
OpenAI has made Greg Brockman’s leadership of product strategy official, ending his interim stint and folding it into his existing infrastructure duties.
John Gruber revisits his 2009 talk on obsession and voice in software, using developer David Smith's six-year Pedometer++ project as a modern example of indie creation that prioritizes passion over quick profits.
Adolescent boys are rushing Scientology's Hollywood headquarters in 'speed runs' for TikTok videos that garner millions of views, leading to user-created building blueprints from the footage.
Apple announces Tim Cook stepping down as CEO to become executive chairman, with John Ternus taking the role, as discussed on Daring Fireball's The Talk Show podcast.
Apple CEO Tim Cook pledges to reinvest any tariff refunds into U.S. innovation during an earnings call, blending financial recovery with political savvy.
Meta cancels a Kenyan contract for reviewing sensitive AI glasses footage after reports revealed workers viewing users' private moments, highlighting privacy risks in AI training.
Tim Cook pledges to reinvest any Apple tariff refunds into U.S. innovation, a move that dodges political backlash while leveraging vague prior commitments for flexibility.