Sponsored by

✦

Thursday’s operating memo: the growth loop is officially cosmic. SpaceX is quietly workshopping a $1.75T IPO, because apparently the new fundraising deck is ā€˜rockets + satellites + AI compute’—while Anthropic accidentally ā€˜open-sources’ half a million lines of Claude Code by shipping the wrong artifact.

Elsewhere in the org, venture funding just posted a record quarter, nuclear startups are pitching bespoke reactors as the next data-center amenity, Meta’s AI center shows up with ten natural gas plants like it’s a normal line item, and 200 robotaxis in Wuhan hard-freeze mid-ride—an unplanned reminder that autonomy still has downtime. Consider this week’s theme: velocity is great until release engineering, energy, or reality becomes the bottleneck.

-šŸ•¶ļø

✦

Seven bullets of updates

  1. āš–ļø Swedish legal AI startup surges to $100M ARR in just 18 months, showing how quickly AI is transforming legal workĀ worldwide.

  2. šŸ¤– Shares leapt 30% after a 132% revenue boost, reflecting fast-rising global AI ambitionsĀ despite net losses.

  3. āš”ļø Backed by $450M, a new wave of nuclear tech could power AI data centers and ease looming energy crunchesĀ with smaller, custom reactors.

  4. šŸš€ Global venture funding hits a record $297B in Q1 2026, driven by an AI investment surgeĀ reshaping startup capital.

  5. šŸ”„ Meta’s new AI center will run on 10 natural gas plants—enough to power 400,000 homes—raising energy questions in South Dakota.

  6. šŸš• Over 200 Baidu robotaxis suddenly froze mid-ride in Wuhan, leaving riders stranded and city roads jammed.

✦

One bad npm publish, 512k lines of ā€œcompetitive advantageā€

Photo by Fahrul Razi on Unsplash

In AI, the moat isn’t just the model — it’s the scaffolding around it. Anthropic just sprayed a lot of that scaffolding across GitHub via an exposed source map file in the Claude Code npm package, revealing ~2,000 TypeScript files and 512k+ lines powering its CLI.

Once the archive landed on GitHub, it spread fast, with tens of thousands forks and researchers already mapping out Claude Code’s memory architecture and orchestration patterns. That’s effectively a public blueprint for one of the more successful AI dev tools on the market.

Anthropic called it a human release packaging error with no customer data exposed. The real damage is strategic: it shows how much IP sits in product glue code, and how a single misconfigured build step can compress years of competitive advantage into one npm publish.

✦

The problem with equal equity

✦

The ops hire that onboards in 30 seconds.

Viktor is an AI coworker that lives in Slack, right where your team already works.

Message Viktor like a teammate: "pull last quarter's revenue by channel," or "build a dashboard for our board meeting."

Viktor connects to your tools, does the work, and delivers the actual report, spreadsheet, or dashboard. Not a summary. The real thing.

There’s no new software to adopt and no one to train.

Most teams start with one task. Within a week, Viktor is handling half of their ops.

✦

AI Pitch Deck Reviewer

Slidebean has been helping startups craft pitch decks for over 10 years. We recently built an AI Pitch Deck Review tool, that processes the text and visuals on your presentation, and provides actionable feedback on the story, potential missing items, and recommendations on how to improve each slide.

✦

Project Apex: Musk’s $75B IPO swing

SpaceX is finally trading Mars dreams for Wall Street terms. The company has confidential IPO paperworkĀ in with the SEC, reportedly targeting a $1.75T valuation and a $75B raise — eclipsing Saudi Aramco’s 2019 record in one shot.

To pull this off, Musk has assembled an unusually large bank syndicateĀ of 21 firms under ā€œProject Apex.ā€ This isn’t just liquidity for rockets; it’s a public-market validation of ā€œfull-stack hard techā€ — rockets, global connectivity, social, and frontier AI — under a single cap table.

The IPO bankrolls Starship, Starlink replenishment, spectrum, and compute for xAI after February’s AI-space merger bet. If public investors accept a $1T+ multiple on an integrated space–AI–data center play, late-stage pricing for capital-intensive infra (launch, chips, power, cloud) will re-rate upward — and everyone else will be benchmarking against the Musk risk curve.

✦

Startup Events and Deadlines

  1. Google for Startups Accelerator: India | April 19 | Apply

  2. Curinos FinTech Incubator | April 20 | Apply

✦

How did we do?

Your feedback fuels us.

Login or Subscribe to participate

Keep Reading