Earlier this year, I had a brilliant idea.
For my job, I have to remember a lot of AI companies, so I thought of drawing a competition map for my desk. I would divide a sheet into columns and list the major Which part of the AI supply chain are companies based on?
Labs that build models and chatbots – OpenAI, Anthropic, Google DeepMind etc. There were AI coding platforms — Cursor, Cognitive, and Replication — that created coding assistants driven by those models. Then, there were startups building highly-specific apps: agents that live in your email, help you implement a marketing strategy, or automate boring tasks like payroll.
I realized that the map would get messy and fast as the AI companies mercilessly encroached on each other’s turf.
Ever-increasing valuations mean companies need to find new sources of revenue. As the models accelerate and commoditize with big-ticket IPOs on the loom, becoming a full-stack AI company is critical to revenue. Here’s how the terrain develops, too fast for my paper map:
Also vip coders
Companies that start with a narrow AI capability — boundary models, AI agents, vibration-coding tools — quickly expand into one or more other areas.
Last year, Anthropic and OpenAI launched Cloud Code and Codex, AI coding platforms to compete with Cursor and Cognitive. Now, screenshots on X for every user — Anthropic won’t confirm it — the company may be working on. The app builder for non-techies competes with vip-coding stars Loveable, Rippled, and Emergent.
SoftBank and Lightspeed-backed Emergent, for one, say they saw Anthropic coming.
“It’s not a surprise. We’ve been anticipating this for a while, thinking internally and preparing for it,” CEO Mukund Jha told me in an April interview.
Anthropic’s entry into the wipe-coding market isn’t game over for the year-old startup, he said.
“Coding is relatively like 20%-30% work. The hard work is actually taking the application to the last mile,” Jha said.
Building secure and production-grade applications, especially for non-technical users, is a difficult problem that cannot be solved by “thinly spread” organizations.
Nail rush
In the AI turf war, OpenAI makes agentive moves.
In February, the lab announced the hiring of Peter Steinberger, creator of OpenClaw, an AI assistant builder that gained popularity earlier this year.
The move adds OpenAI to the list of companies built in the agent space, such as former Meta Tech head Brett Taylor’s Sierra and Salesforce’s AI agent platform AgentForce. Now, Codex has evolved from a coding assistant to a virtual AI agent that can parse and respond to emails, manage files and schedule meetings.
Small players are also excited about this place. Last month, Emergent, which started as a vip-coding platform, expanded into the personal agent space.
Other examples of emerging overlap include Anthropic’s entry into the design market and graphic design firm Canva’s entry into the broader manufacturing AI and productivity suite business.
‘Google wants to touch everything’
If it sounds like you’ve heard this story before, you have too. But instead of OpenAI, Anthropic and Lovable, there were the letters Google, Amazon and Microsoft.
Michael Gotting, a partner at European venture firm Northzone, said there was a time when FAANG companies had their fingers in all the pockets.
Gotting, who co-founded e-commerce site Shopping.com, said Google was a big concern.
“I remember when I was building my first company 25 years ago, Google wanted to touch everything,” he told Business Insider in April. “For us at Shopping.com, we launched Google Froogle, and that’s what we were doing. And we were like, “Oh, we’re dead.”
He added: “But it turned out to be a side project. They made so much money in their core business, so how hard would they go? Well, the answer is they didn’t.”
Tom Sheridan, vice president of early lovable investor RTP Global, concedes that a so-called “super app” — an app that rules them all — is unlikely.
“Super app talk is mostly noise settled by the IPO calendar. Now we’re seeing foundation model players in the throws of a game of B&L chicken,” he said. “Once these companies go public, the money will stop being burned and channeled into categories where you’re good-but-not-great.”
“Google Graveyard,” an unofficial online directory, tracks 305 projects that have been sunsetted by the search giant over the years.
Apple is also famous for “Sherlocking” — introducing a new feature that makes a third-party tool irrelevant — but they don’t always stick. In 2023, Apple introduced PayLater, which rivaled Klarna and Assertion. It was discontinued in 2024.
Gotting said the same thing is happening at OpenAI and Anthropic, which founders fear for the day when they ship an application to startups. Worked for months.
It’s more profitable for Anthropic to keep building better models so it can charge more for its core service, he said. But if Chinese players or other labs turn out equally well, the samples change As a commodity, anthropology can be difficult in these services.
Besides getting “Sherlocked,” startups face another big risk: bias.
Startups build billion-dollar businesses on top of APIs controlled by companies that may eventually compete with them. For example, Cursor relies on anthropic models to power its features, but the two compete as indexing assistance providers.
Short-term success for customers
More players play more The freebies are a win for individual builders and small businesses — but only in the short term, Sheridan said.
“Foundation model companies can ship a shippable version of almost anything, but if the bundled tool isn’t as good as the specialized tool I’m already using, I’ll return in one sitting,” the VC said. “Product expansion that looks for blocking bumps risks worsening UX and users know it.”
As big labs like OpenAI and Google sprawl in every direction, companies like Reddit or LinkedIn provide tons of data and shut down scrapers. That’s bad news for small startups like sales tech tools or meeting summaries that want to build their services from data.
These changes bring an opportunity for founders to know what users want from their data.
“Today’s foundational models may see briefly shared meeting transcripts, but they don’t know what folders they should be filed under, what a team really cares about, or what follow-up actions are needed,” Sheridan said. “This is the gap that startups can create.”
The market is also ripe for consolidation.
“I expect one of the major consumer AI breakouts in the next 24 months to be by Google,” he said. “Google has a cost-absorbing consumer advertising business and is structurally very pessimistic about consumer AI talent.”
Sheridan said the first company to buy gets the best price.
“You don’t want to be the last consumer AI play when every major buyer is only taking one shot,” he said.