The AI Land Grab Looks Familiar

In May of 2013, Yahoo announced plans to buy Tumblr for 1.1 billion dollars. Yahoo’s CEO, Marissa Mayer, stood in front of the press and said, “We promise not to screw it up”.

Tumblr users responded with exactly the kind of dramatic sass you’d expect: an angry petition that topped 168,000 signatures, threats to delete their blogs, and endless reaction GIFs. It turns out, they were right to do all that.

Note: This post is a blog-style-adaption of the video above.

Fast forward a couple of years, and Tumblr’s founder David Karp was stepping down, the advertising revenue Yahoo had promised had not materialized, and by 2016 Yahoo had already slashed 712 million dollars off Tumblr’s valuation.

The nail in the coffin? In December 2018, Tumblr banned all adult content, which happened to be what a surprising number of users were there for. Traffic dropped thirty percent over the next 3 months. CNBC later reported that daily Tumblr posts had fallen from at least 84 million in 2014 to just 30 million by 2018, more than a 50% drop.

And by 2019, Verizon, which had absorbed Yahoo, sold Tumblr to Automattic, the creators of WordPress (yes, the dramatic open source company), for about three million dollars.

Three million dollars. For a website that once sold for 1.1 billion.

Unfortunately, in the tech world, this tale is as old as time. Acquisition leads to enshittification. The acquired products that you love so often become neglected, or just lose their identity as they cater to a new enterprise audience. Products you used to be able to use for free become paid services, or are locked into whatever ecosystem the new owner wants you to adopt.

So when Anthropic announced in December 2025 that they were acquiring Bun, the JavaScript runtime that so many developers, myself included, have been migrating to for a faster experience, you’d expect the dev community to riot.

But, weirdly, they didn’t. Many even cheered. But more on that in a minute.

And then in March 2026, just a few months later, OpenAI announced they were acquiring Astral, the team behind some of the fastest-growing Python tools in the ecosystem. And for that acquisition? Well, the community panicked.

Two AI giants. Two beloved open-source projects. Two wildly different reactions. What’s going on here?

To answer that, we need to look at one more apples-to-apples historical comparison.

Oracle, Sun, and the fear that open source gets dismantled

See, way back in 2009, Oracle agreed to buy Sun Microsystems in a deal valued at about 7.4 billion dollars. Sun was one of the great open-source companies of its time. It maintained four massive projects: Java, MySQL, OpenSolaris, and OpenOffice. Now, if you’ve been around in tech, you know these weren’t just niche tools. They were the foundations people built entire careers and even companies on.

Within months of the acquisition, James Gosling, the creator of Java, resigned.

So, the community decided to take matters into their own hands. MySQL’s original creator, Michael Widenius, launched a petition to pressure regulators over Oracle’s takeover of Sun and MySQL. The OpenOffice community formed The Document Foundation and forked LibreOffice.

And Oracle’s response? They sued Google over Java’s use in Android, a case that got dragged all the way to the Supreme Court. And by 2017, iTWire reported that Oracle had laid off much of its Solaris core development staff.

To say that Sun’s open-source ecosystem was neglected after the acquisition would be quite the understatement. It was systematically dismantled.

So the fear is justified.

But here’s the question I come back to: what’s happening now? Are Anthropic and OpenAI pulling an Oracle? Or is something genuinely different this time?

We scraped the comments on the Hacker News posts from both the Bun acquisition and the Astral acquisition and used Claude Opus 4.6 to do some qualitative sentiment analysis.

Comments on the Bun acquisition were 32% positive, 33% neutral, 35% negative.

Comments on the Astral acquisition were 13% positive, 22% neutral, and a staggering 65% negative.

It’s weird because it feels like the same playbook, AI giant buys beloved open-source developer tool, but the community reaction is night-and-day different.

There’s something different about these deals. We’re going to dig in, in just a second.

Why the Bun deal felt different

Let’s start with Bun, because that’s where this story really begins, and remember, it’s the one a surprising number of developers actually cheered for.

Jarred Sumner began working on Bun in 2021, then made its public debut in 2022. At first, the pitch was about a blazing-fast JavaScript runtime written in Zig, with Bun positioning itself as dramatically faster than Node.js. Over time it grew into something broader: a bundler, a package manager, a test runner, so much of what you need for typical dev work all in one toolkit. It was a big deal then, and is still considered one of the most successful and beloved open-source projects of the last 5 years.

What makes Bun’s story interesting, though, isn’t just the product, but the founder.

Jarred Sumner had already built a reputation as a very public, very online programmer who shipped in the open and talked directly to his users. In an industry that sometimes feels like it’s run by people who knew the right people, that made him feel like a scrappy underdog.

And it wasn’t long until Bun had over 7 million monthly downloads and 82,000 stars on GitHub. And it had something else that’s harder to measure: a community that genuinely cared whether it survived.

But here’s the thing that made Anthropic’s interest in Bun make sense: Claude Code, Anthropic’s AI coding product, depends on Bun. Not Node, and not Deno. As Jarred himself put it in the announcement: “Claude Code ships as a Bun executable to millions of users. If Bun breaks, Claude Code breaks”.

So think about what that means from Anthropic’s perspective. Most AI companies had been vertically integrating into compute, buying GPU clusters, negotiating cloud capacity deals, and trying to optimize inference costs. This Anthropic acquisition wasn’t in any of those categories. It was a free-to-use JavaScript runtime. Why?

Well, it’s not just the models, but the software the models’ products run on. The worry of some, is that just generating code isn’t actually where all the high margin value is created. Maybe, you’ve got to own the toolchain that ships the tool that generates the code. It’s kind of like Apple owning the iPhone, the OS, the App Store, and the developer tools.

The acquisition, then, wasn’t just a random land grab but a vertical integration. And the incentive alignment was really good. Anthropic has every reason to keep Bun excellent, because if Bun degrades, Claude Code degrades.

Another big factor regarding community acceptance was that Anthropic already had a better reputation with developers than OpenAI did, and that only strengthened in early 2026 when Anthropic publicly clashed with the Trump administration over the Pentagon’s access to Claude. The cited concerns were about mass surveillance and weapons using AI to make kill decisions - something most people were proud of Anthropic for saying no to. OpenAI then immediately stepped in with its own Pentagon deal, which, as you can imagine, did not earn them the same goodwill.

So the community reaction made sense. A founder they liked, joining a company that seems to be aligned with the preservation of the product, and the fact that the acquiring company was in everyone’s good graces.

So what the hell happened with Astral and OpenAI?

Why the Astral deal caused panic

Astral was founded in 2022 by Charlie Marsh, and what they built in the three years that followed is honestly kind of astonishing. As a Python user it really snuck up on me all at once. Their flagship tools, uv and Ruff, have become some of the most prominent modern tools in Python development, in a world where pip and conda ruled for what feels like decades.

Uv is positioned as a single tool to replace pip, pip-tools, pipx, poetry, pyenv, and virtualenv, among others. Ruff is a linter and formatter with drop-in parity for Flake8, isort, and Black-style formatting.

These aren’t really optional utilities. You’ve got to use something. They’re in the default stack of every modern Python project I’ve seen in the last year. When you acquire that, you’re not buying a developer tool, you’re acquiring switching costs at scale across a meaningful share of the Python ecosystem.

But, Codex doesn’t ship on Python, the way Claude Code ships on Bun. So what’s going on?

Python is not a random target. As OpenAI itself put it, Python has become one of the most important languages in modern software development, powering everything from AI and data science to backend systems and developer infrastructure. Control the Python toolchain, and you likely have a presence in a huge share of serious AI development workflows.

Needless to say, OpenAI knew exactly what they were getting. By the time OpenAI announced the acquisition in March 2026, Astral’s tools were seeing hundreds of millions of downloads per month.

So why did the community react so differently to this than to Bun?

Well, I think some of it has to do with Astral’s lore. Astral didn’t have the same cult following, the same benevolent-dictator-for-life social media presence, or the same underdog energy around its founder. When Anthropic bought Bun, people said, “congrats to Jarred.” When OpenAI announced Astral, people felt like a company had sold out.

But most of it, I think, is about OpenAI.

The Hacker News thread on the Astral acquisition hit 1,489 points and 901 comments last I checked. Which, for Hacker News, is not indifference. It’s front-page news. And you can understand why when you consider what OpenAI has looked like to the open-source community over the past few years.

Let’s see: controversies around AI alignment, the organizational upheaval, the pivot away from the nonprofit mission that was supposed to be the whole point of the company. Each of these events is a reason to wonder: what happens to uv when OpenAI needs a competitive advantage that conflicts with maintaining it as a neutral community resource?

OpenAI’s own announcement leaned into Codex, their AI coding assistant, which had tripled in user growth since the start of 2026 and passed two million weekly active users. The stated goal was to make Codex capable of participating in the entire development workflow, not just generating code, but managing environments, running linters, catching type errors, everything Astral’s tools do. By integrating Astral’s toolchain directly, OpenAI puts Codex inside the Python workflow in a way that competitors can’t easily match.

How you feel about that depends on how you feel about OpenAI. It is a compelling product vision perhaps, but it’s also significant leverage over an ecosystem that a lot of people share.

These companies are buying narratives

Here’s what I think is actually going on with both deals. These companies aren’t just buying tools, instead they’re buying narratives.

When Anthropic bought Bun, they also bought the brand of Jarred Sumner. And they knew what they were doing. Jarred, just by who he is, is in a fairly public-facing role in Claude Code development. The goodwill that the community had built up around him, the years of rooting for a scrappy, honest, non-pedigreed founder… That goodwill is now associated with Anthropic. They acquired credibility. And developers are a community that values credibility and transparency more than most.

OpenAI went in a different direction; they ALSO just bought TBPN.

If you haven’t heard of it: TBPN is Technology Business Programming Network, a daily livestreamed tech talk show hosted by John Coogan and Jordi Hays. It’s kind of like a SportsCenter, but for Silicon Valley news.

OpenAI acquired them in April 2026. The show will maintain editorial independence, or so they’ve said. It will continue to cover whatever it covers, but it now reports into OpenAI’s strategy organization, and its founders will also advise on messaging and marketing strategy.

And now we’re getting to the heart of the matter, and it’s an entirely different kind of acquisition. Not a developer tool. Not a runtime or a linter. TBPN is a straight-up media property. An audience with a channel and a brand and a voice. Kind of like us here at Boot.dev.

Sam Altman posted on X that he doesn’t expect TBPN to go easy on OpenAI after the acquisition, adding that he’ll do his part to help enable that with occasional stupid decisions. Which is either a remarkably self-aware joke or precisely the kind of thing you say when you want to seem remarkably self-aware. I couldn’t say.

What’s interesting is that OpenAI’s stated rationale was almost candid about it. Their head of AGI deployment wrote that the standard communications playbook just doesn’t apply to a company like OpenAI, and that rather than rebuild that capability from scratch, it made sense to acquire it. Buy the thing. Don’t build it. Same logic as Astral. Same logic as Bun.

And I just have to add, because it’s hilarious, that as we were doing research for this video, OpenAI acquired Hiro Finance and Cirrus Labs… so nothing seems to be slowing down.

These labs are acquiring companies like crazy. Developer tooling and narrative and community trust. One acquisition at a time, they’re trying to own the entire context in which software gets built, and in which they get talked about. So much money is at stake in the so-called “AI bubble”, that the biggest risk in their minds at the moment simply seems to be the runner-up.

Are Bun and Astral doomed?

Now for the part you really care about. Does any of this mean Bun and Astral are doomed?

I don’t think so, I certainly don’t hope so, as I use both tools.

Open-source licensing is a real constraint. When a project is released under an MIT license, the code that has already been released can still be copied, modified, and redistributed under that license. Even if Anthropic or OpenAI decided tomorrow to take their newly acquired tools closed-source, which would be an unbelievable act of self-sabotage, the open-source community would still have the most recent MIT-licensed version. That’s the basis for a fork.

Remember Oracle? That’s what happened with MySQL and MariaDB. The community has done this before, and they’d do it again.

But if I’m being honest: I don’t think we’re heading down a quick road of malicious enshittification. It’s likely a bit more subtle.

Perhaps the roadmaps of these tools start to serve the acquirer’s priorities over the community’s. Maybe Bun’s development accelerates in exactly the areas where Claude Code benefits, and slows down in the areas it doesn’t. Maybe uv gets deep Codex integration while general community feature requests wait in the queue. That’s not instantly obvious villainy, it’s just what happens when a product’s maintainers have a new primary stakeholder with different incentives.

The real risk isn’t closure - it’s stagnation on features that don’t serve the acquirer’s products. That’s the thing to watch out for. Not whether the MIT license survives, it will, but whether the products grow in ways that serve developers that use them, or just in ways that serve the parent company.

And underneath all of this is a financial reality that I don’t really want to talk about, but I think we have to talk about.

The money being spent right now, by these frontier labs and by the investors feeding them capital, is beyond anything we’ve seen before, even in the dotcom era. Anthropic’s revenue run rate was 1 billion dollars in January 2025 and had grown to 7 billion dollars by October of the same year. The company has raised billions in private funding, with the latest round valuing it at 183 billion dollars.

The numbers speak for themselves. They are the numbers of an arms race to gain decisive advantages before the other side does. Acquisitions are the way to buy capability when you don’t have the time to build it.

The question hanging over all of this, is what happens if the music stops. AI is generating crazy revenue growth and even crazier valuations. But there is a real and serious argument that significant parts of this could be a bubble, inflated by capex spending from major corporations that are betting enormously on AI outcomes that, while I do believe AI is here to stay, the margins, the leverage, and the business models are still too soon to call. If that bubble deflates, not even pops, just deflates, the maintenance and development of tools like uv, Ruff, and Bun suddenly look less certain.

Anthropic has a direct financial incentive to keep Bun excellent now, but financial incentives can change. People can leave. So who knows.

Where I land

So, here’s where I land on all of this, at least for now.

I don’t think Bun is the next Tumblr. I don’t think Astral is the new OpenSolaris. The incentive structure is different, and these acquisitions, unlike Yahoo’s purchase of Tumblr or Oracle’s absorption of Sun, are driven by genuine technical dependency rather than wishful thinking of direct monetization.

Anthropic literally cannot afford for Bun to degrade. That alignment isn’t a guarantee of anything, but it’s not nothing, either.

And the developers who built these tools earned what they got! I see a tendency in the community to be suspicious of founders who sell. They treat a payout as a betrayal, but I don’t think that’s always fair. Jarred Sumner built something people genuinely love, made a decision he thought hard about, and landed somewhere with more resources to do more work. He doesn’t owe anyone anything, open-source is a gift with no strings attached.

Charlie Marsh built a Python toolchain that went from zero to hundreds of millions of downloads per month in three years. I’ll be very sad if the projects don’t continue to thrive, but I can’t blame the creators for simply not donating even more time and energy to the projects that I use for free.

All that said, I’m also not going to pretend the picture is 100% clean. OpenAI now owns some of the most important Python tools that AI developers use. Anthropic owns the JavaScript runtime their coding product runs on. They’re acquiring media and narrative infrastructure alongside technical infrastructure, trying to own the context in which AI gets built and talked about.

Remember the famous words of Marissa Mayer in 2013? “We promise not to screw it up?” The problem with Tumblr was never that Yahoo was super evil. It was that Yahoo’s incentives and Tumblr’s needs diverged the moment the acquisition closed.

What makes Bun and Astral potentially different is that the incentives, for now, seem to be pointing in the same direction. The tools need to be good for the community to use them, and they need the community to use them for the products to work. For now.

Bibliography

1. “Yahoo to Buy Tumblr for $1.1 Billion.” The New York Times, May 19, 2013. https://www.nytimes.com/2013/05/20/technology/yahoo-to-buy-tumblr-for-1-1-billion.html
2. Marissa Mayer, “Tumblr + Yahoo! It’s Officially Official.” Tumblr post, May 20, 2013. https://www.tumblr.com/@yahoo/53441093826/tumblr-yahoo-its-officially-official?source=share
3. Glenn Chapman, “Tumblr fans get sassy on Yahoo! takeover news.” Yahoo News Singapore / AFP, May 21, 2013. https://sg.news.yahoo.com/tumblr-fans-sassy-yahoo-takeover-news-212230231.html
4. Liana B. Baker and Heather Somerville, “Tumblr founder David Karp to step down.” Reuters, November 27, 2017. https://www.reuters.com/article/business/tumblr-founder-david-karp-to-step-down-idUSKBN1DR2HW/
5. “Yahoo cuts Tumblr’s value by $712m.” BBC News, July 21, 2016. https://www.bbc.com/news/technology-36843375
6. Shannon Liao, “Tumblr will ban all adult content on December 17th.” The Verge, December 3, 2018. https://www.theverge.com/2018/12/3/18123752/tumblr-adult-content-porn-ban-date-explicit-changes-why-nsfw
7. Shannon Liao, “After the porn ban, Tumblr users have ditched the platform as promised.” The Verge, March 14, 2019. https://www.theverge.com/2019/3/14/18266013/tumblr-porn-ban-lost-users-traffic-similarweb
8. Leah Collins, “How Tumblr went from a $1 billion Yahoo payday to a $3 million fire sale.” CNBC, September 15, 2022. https://www.cnbc.com/2022/09/15/how-tumblr-went-from-1-billion-yahoo-payday-to-3-million-fire-sale.html
9. Jarred Sumner, “Bun is joining Anthropic.” Bun Blog, December 2, 2025. https://bun.com/blog/bun-joins-anthropic
10. “Anthropic acquires Bun as Claude Code reaches $1B milestone.” Anthropic, December 3, 2025. https://www.anthropic.com/news/anthropic-acquires-bun-as-claude-code-reaches-usd1b-milestone
11. Charlie Marsh, “Astral to join OpenAI.” Astral Blog, March 19, 2026. https://astral.sh/blog/openai
12. “OpenAI to acquire Astral.” OpenAI, March 19, 2026. https://openai.com/index/openai-to-acquire-astral/
13. “Oracle Buys Sun.” Oracle Press Release, April 20, 2009. https://www.oracle.com/corporate/pressrelease/oracle-buys-sun-042009.html
14. James Gosling, “Time to move on…” personal blog, April 9, 2010. https://nighthacks.com/jag/blog/324/index.html
15. “Sign the Petition.” Save MySQL! Internet Archive capture, February 8, 2010. https://web.archive.org/web/20100208205431/http://helpmysql.org/en/petition
16. “LibreOffice Timeline.” The Document Foundation. Accessed April 14, 2026. https://www.libreoffice.org/about-us/libreoffice-timeline/
17. Google LLC v. Oracle America, Inc., 593 U.S. ___ (2021). U.S. Supreme Court, decided April 5, 2021. https://www.supremecourt.gov/opinions/20pdf/18-956_new_3fb4.pdf
18. Sam Varghese, “Bye, bye Solaris, it was a nice ride while it lasted.” iTWire, September 4, 2017. https://www.itwire.com/open-sauce/79738-bye,-bye-solaris,-it-was-a-nice-ride-while-it-lasted.html
19. “Bun: A fast all-in-one JavaScript runtime.” Bun. Accessed April 14, 2026. https://bun.com/
20. Jarred Sumner, “Bun is joining Anthropic.” Bun Blog, December 2, 2025. https://bun.com/blog/bun-joins-anthropic
21. Adam Gabbatt and Dara Kerr, “OpenAI to work with Pentagon after Anthropic dropped by Trump over company’s ethics concerns.” The Guardian, February 28, 2026. https://www.theguardian.com/technology/2026/feb/28/openai-us-military-anthropic
22. “uv.” Astral Docs. Accessed April 14, 2026. https://docs.astral.sh/uv/
23. “Ruff.” Astral Docs. Accessed April 14, 2026. https://docs.astral.sh/ruff/
24. “Astral to Join OpenAI.” Hacker News, item 47438723, March 19, 2026. https://news.ycombinator.com/item?id=47438723
25. Fidji Simo, “OpenAI acquires TBPN.” OpenAI, April 2, 2026. https://openai.com/index/openai-acquires-tbpn/
26. Rebecca Bellan, “OpenAI acquires TBPN, the buzzy founder-led business talk show.” TechCrunch, April 2, 2026. https://techcrunch.com/2026/04/02/openai-acquires-tbpn-the-buzzy-founder-led-business-talk-show/
27. Julie Bort, “OpenAI has bought AI personal finance startup Hiro.” TechCrunch, April 13, 2026. https://techcrunch.com/2026/04/13/openai-has-bought-ai-personal-finance-startup-hiro/
28. Fedor Korotkov, “Cirrus Labs to join OpenAI.” Cirrus Labs, April 7, 2026. https://cirruslabs.org/
29. “The MIT License.” Open Source Initiative. Accessed April 14, 2026. https://opensourceorg.github.io/osi_website_archive/archive.opensource.org/licenses/MIT.html
30. Dario Amodei, “A statement from Dario Amodei on Anthropic’s commitment to American AI leadership.” Anthropic, October 21, 2025. https://www.anthropic.com/news/statement-dario-amodei-american-ai-leadership
31. “Anthropic Explores IPO as It Makes First Acquisition.” AIwire, December 4, 2025. https://www.hpcwire.com/aiwire/2025/12/04/anthropic-explores-ipo-as-it-makes-first-acquisition/