Issue 25

20 June 2026

Hi there,

SpaceX agreed to buy Cursor in a $60 billion stock deal, while Hetzner changed prices for new server orders. This week also brought a useful GitHub Actions safety default and clearer Copilot usage figures.

Industry

1. SpaceX agrees to acquire Cursor for $60 billion

TechCrunch reported on 16 June that SpaceX agreed to acquire Anysphere, the company behind Cursor, in an all-stock deal valued at $60 billion. The deal was expected to close in the third quarter, subject to the usual review. An agreement is not yet a completed acquisition.

Artificial Intelligence

2. Anthropic contests an order to remove two models

In its 12 June statement, Anthropic said it was complying with a US government directive to remove access to Fable 5 and Mythos 5 while contesting its basis. Other Anthropic models were unaffected. Access and model capability are distinct questions here.

Infrastructure

3. Hetzner changes prices and standardises server products

Hetzner’s notice, updated 15 June applies the revised prices to new orders and rescales; currently rented servers retain their terms. The company also reduced setup fees on most dedicated servers and standardised product designations. Existing deployments and new capacity requests therefore face different cost calculations.

Microsoft

4. GitHub adds per-user AI-credit figures to Copilot metrics

Since 19 June, enterprise and organisation administrators can retrieve an ai_credits_used field in daily and 28-day user-level reports. It measures total credits used per person, not per prompt. That makes a usage-versus-cost conversation less speculative without pretending every token can be attributed to a business outcome.

Development

5. Typst 0.15 brings variable fonts and richer export

Typst’s 15 June release post lists variable fonts, MathML and multi-file HTML export, multiple bibliographies, and a printed documentation build. The migration guide matters to anyone using Typst to generate client-facing PDFs; the changes are not only typographic polish.

6. Elixir 1.20 is released with annotation-free gradual type checking

The 3 June official announcement describes type inference and gradual checking without annotations, including detection of verified bugs. This release predates this week, but its stable status matters to anyone deciding whether to adopt the new type checker.

Information Security

7. GitHub closes a dangerous Actions checkout pattern

GitHub’s 18 June changelog says actions/checkout v7 refuses to fetch forked pull-request code in specified privileged pull_request_target and workflow_run contexts. The trap is straightforward: the workflow has access to the base repository’s token, secrets and cache, then executes a fork’s unreviewed code. Audit workflows that deliberately use that combination rather than assuming a version bump alone settles every permission question.

Coming up

July (estimated): npm v12 is expected to change install-script, Git and remote-dependency defaults. Check the warnings in npm 11.16.0 before upgrading.

Reply if anything here changes how you’re thinking about your stack.

Cheers, Julian

How this is made

Throughout the week I stumble across a mildly unreasonable number of interesting things, and I forward them instantly to the friend or colleague I think might care – sometimes to their delight, sometimes to their annoyance, and often with no context at all. Heartbeat is the attempt to do that a little better.

Every Friday a small agent I built, Honoka, looks through the places where those links tend to leak out: my private email, work email, Matrix, Mastodon, WhatsApp, Apple Messages, Signal, and the faint imprints on the platen of my Olympia typewriter (still not an API, tragically). It sorts, filters, groups and summarises the week, then hands me a draft.

Honoka is guided by a private corpus of things I have written over the last fifteen years, so it can get closer to how I sound in more-or-less official emails and public notes. I still take a pass by hand: remove things, change sentences, check links, argue with the judgement. Whether that is enough is, frankly, the experiment. Every issue has one item written entirely by hand. If you can reliably spot it, hit reply and judge.