# Thursday Thoughts: Why Anthropic Is the Next AWS, but Potentially Worse > Source: > Published: 2026-07-09 14:41:40+00:00 A few weeks ago I wrote about how [AI-native mirrors cloud-native](https://dev.to/posts/thursday-thoughts-how-ai-native-mirrors-cloud-native) — lift-and-shift now, real architectural rethink later. The same pattern enterprises went through with the cloud. That post was about workflows and org design. This one pulls on a different, less comfortable thread of the same analogy: Anthropic is starting to look a lot like AWS did a decade ago. Same core move. Much faster clock speed. And I'm not convinced it (or we, startups) survives the difference. Quick side rant, then I'll get to the point. Anthropic and the other frontier labs are obviously more than infrastructure companies. They do real research, real alignment work, real science. But strip away the mission statements and look at the balance sheet: the core asset is a giant pile of GPUs with a business model of renting time on them. It's just denominated in tokens instead of instance-hours. An EC2 instance is general-purpose compute you point at whatever workload you want. A Claude API call is the same idea, just pre-loaded with one very specific, very valuable workload already running. That's not a knock. Ok, rant over. The interesting comparison isn't the compute. It's what companies do when they sit atop these platforms. A thriving ecosystem of startups and open source projects built directly on top of AWS for most of the 2010s. And AWS had a front-row seat to all of it. That vantage point let AWS watch which capabilities were working. Decide which ones were worth turning into first-party, cloud-native services. And then executing, quickly. The pattern is well documented at this point. AWS forked Elasticsearch into Open Distro in 2015 rather than pay Elastic for a managed offering. It shipped a MongoDB-compatible DocumentDB service that pushed MongoDB from AGPL to SSPL in 2018. It offered a managed Cassandra service. Redis followed the same arc in 2024, moving off BSD in response to AWS's ElastiCache and Azure's competing cache products, before reversing back to AGPLv3 in 2025 once the community fork (Valkey) had absorbed enough of the commodity pressure that Redis could go back to competing on product instead of licensing. HashiCorp did the same thing with Terraform's move to BSL in 2023, which is arguably the case that did the most lasting community damage, and which triggered the OpenTofu fork within weeks. The pattern is pretty clear: successful infrastructure project gains traction, a hyperscaler ships a managed, forked, or compatible version without meaningfully contributing back, the original company changes its license to defend the business, and the community reacts, sometimes forking around the license itself. It's now a repeating, named cycle in the open source world. AWS was the proximate cause of every iteration of it. What's interesting to me is that this dance didn't kill the ecosystem. Elastic, Redis, and HashiCorp are all still around. Redis' own CEO has since said publicly that pushing AWS onto its own fork put both companies on a level playing field where they compete on product rather than fighting over a shared codebase. AWS drew blood. Then eventually gave the ecosystem room to differentiate around it. Now look at Claude. A thriving ecosystem of companies has built directly on top of it. They have the same privileged vantage point AWS had. So naturally, Anthropic is doing exactly what AWS did: watch what's working, then ship a first-party version. The case that kicked up the most dust was Claude Design. Anthropic's chief product officer quietly resigned from Figma's board three days before the launch. Figma's stock dropped roughly 7% on launch day alone. And Clauyde Design landed as a direct, conversational alternative to opening Figma in the first place. To make matters worse, it offered a preferential export path into Canva, a company Anthropic had been partnering with for two years. Figma and Adobe, both long-standing Anthropic partners, told reporters afterward they'd had essentially no advance notice of what was coming. Design wasn't a one-off. Claude for Legal expanded into a full suite of plugins and MCP connectors aimed at law firms, directly overlapping with venture-backed legal AI startups like Harvey and Legora. Claude Science launched as a dedicated research workbench connecting to more than 60 scientific databases. Anthropic explicitly framed this as extending Claude's reach from "just a model provider" into owning the operating layer for an entire industry. Sound familiar? It's the same language used about what Claude Code did for software development. Anthropic acquired Coefficient Bio to bring first-party pharmaceutical-planning capability in-house. Claude for Word pushed directly at Microsoft's own productivity suite. And a round of Claude Cowork plugins aimed at legal and sales workflows was blunt enough to trigger what multiple outlets started calling a "SaaSpocalypse" sell-off across data analytics and professional services stocks. That's design, law, science, productivity software, and enterprise workflow tooling. All absorbed or threatened inside about the same number of months it took AWS to ship one Elasticsearch fork. Remember, Claude Code is even 18 months old yet. Here's the part that actually worries me. It's not the individual moves. It's the tempo. AWS was famous in its era for a relentless release cadence. It took years to work through its ecosystem: Elasticsearch in 2015, the Cassandra-compatible service later, the MongoDB and Redis license fights stretching from 2018 to 2024. That gave the ecosystem time to see the pattern coming and build defenses into it Change licenses. Fork a community version. Compete on genuine product differentiation. Anthropic movies at a pace that makes AWS look patient. If AWS was supersonic, Anthropic is hypersonic. The risk with hypersonic isn't just speed, it's that the ecosystem doesn't have time to adjust before the next shockwave. There's a second difference that matters just as much: AWS mostly stayed at the infrastructure layer. It rarely competed directly against the applications that ran core business processes. Anthropic is doing the opposite. It's competing directly at the application layer, against the actual products entrepreneurs built on top of Claude. That's a meaningfully higher-stakes threat to the founders building in this ecosystem when your coopetition is for the operating the same business process. Now let me argue against myself, which I'm quite good at. This is not a done deal. AWS earned its dominance the hard way. It built a decade of infrastructure expertise service by service. Anthropic's one genuinely deep, first-party domain expertise is software engineering. Claude Code is the proof: it went head-to-head with Cursor, a company built entirely on top of Claude models. Anthropic won convincingly, leveraging it's deep understanding of coding at the level required to out-execute a specialist. Does Anthropic have that same depth in legal? Finance? Science? Arguably not, at least not yet. Which means that Claude's first-party verticals end up being "batteries included." Competent-enough defaults that get users started, and that power users graduate out of toward best-of-breed tools once their needs outgrow the generic version. It's the same way plenty of Redis and Elastic customers stuck with the originals even after AWS shipped a managed alternative. Although, the Coefficient Bio acquisition could be an interesting indicator that it will buy domain expertise and develop Claude-native services to maintain hypersonic at high fidelity. Only time will tell whether Claude for Legal is Anthropic's Elasticsearch fork or its actual Elasticsearch. There's something both ironic and instructive happening here. A company that started with explicitly altruistic, safety-first goals is turning into a genuinely shrewd capitalist machine. I don't think "predatory" is the right word for it any more than it was for AWS 10 years ago. This is just what a company with a privileged, ecosystem-wide vantage point rationally does. But perception is reality. Right now Anthropic's biggest gap isn't judgment. It's that it doesn't have a startup incubation muscle. AWS methodically built an entire motion around cultivating the ecosystem it was also competing with — credits programs, an investment arm, a startup showcase built into re:Invent itself. Anthropic has early, thin versions of the same idea. I see the Menlo Anthology Fund, small business credits program with Workday, a handful of CDFIs. That's a start. It's not yet the kind of counterweight that keeps new ideas alive long enough to prove themselves before Anthropic eats them. Software is eating the world. Anthropic is eating software. Where does that leave the world?