Leaner, faster and powered by AI: A new breed of unicorn has arrived

When Anthropic CEO Dario Amodei was asked in May when the world would see a one-person billion-dollar company, he replied: “2026.” OpenAI CEO Sam Altman revealed last year that he’s part of a “little group chat” of tech CEOs who are placing their bets on when this moment will arrive.

Irrespective of if — or when — the tech world achieves this milestone, the question underscores a new reality: Billion-dollar private companies, also known as unicorns, are evolving to become a leaner, faster breed. A new generation of artificial intelligence startups are building tools to automate workflows across industries from customer service to sales to software engineering. The tools are built atop AI models like OpenAI’s ChatGPT or Google’s Gemini, meaning they’re unburdened by the data center costs that are sapping billions from Big Tech companies.

Driving down their expenses even further, and in keeping with the products they tout, these startups also tend to operate with AI tools over humans.

“[AI startups] are most aggressively pushing the boundaries of what’s possible because they don’t have much cash, and they don’t have public shareholders holding them accountable,” says Ash Barbour, who sold his AI-enabled sales software to Woodpecker.co in 2022, a publicly-traded Polish firm that automates cold emailing.

As a new generation of leaner startups gear up to reach ten-figure valuations at record speed, they’re rewriting the playbook of entrepreneurship. At the same time, whether such growth can be sustainable is less certain, leaving investors caught between capturing the rapid returns of modernity, and abiding by the guardrails of historical precedent.

Lovable, a Swedish AI-powered “vibe coding” platform, which allows non-tech savvy users to build software via natural language prompts, is Europe’s fastest-growing startup. Earlier in July, just eight months after launch, the 45-employee company became the region’s latest unicorn. Fellow vibe coding platform Bolt hit $20 million in annual recurring revenue (ARR) in just two months. Coding copilot Cursor confirmed in June that it’s hit $500 million ARR, a five-fold increase since January, achieved with fewer than 50 workers. And AI-enabled workflow automation startup Gumloop raised a $17 million Series A with just two full-time staff — and wants to reach a $1 billion valuation with just 10, a nod to a prediction from Altman that we’re going to see 10-person unicorns “pretty soon.” 

Safe Superintelligence, meanwhile, was launched one year ago by OpenAI’s former chief scientist Ilya Sutskever with the aim of building an AI agent capable of surpassing human intelligence. It’s already valued at $32 billion with just 20 employees.

These figures are already outpacing 2024, when AI unicorns, on average, had around 200 employees and reached billion-dollar valuation in just two years, according to CB Insights. Non-AI unicorns typically required nine years and nearly double the team size to do so. 

“I’m working towards a future scenario where we have zero operations and back-end employees. The only employees we have are people-facing,” says Barbour. Last year he founded Yes.inc, a service that pairs startups, including Lovable, with industry experts who can promote their product. Barbour runs the startup with just one employee, whose job is interfacing with their internal AI.

Artisan, the creator of an AI agent that does the work of a B2B sales rep, is just one of the companies building tools for consolidation. The company launched last year and soon raised a $25 million Series A. Artisan’s AI agent, Ava, comes with over 300 million contacts, can extract real-time data about potential buyers, and sends personalized emails from a rep’s inbox, producing the output of five to 10 reps. The “grunt work” of sales — like copy-pasting emails and prospecting on LinkedIn — is “where AI can, and should, step in,” says founder and CEO Jaspar Carmichael-Jack.

Artisan sparked outrage earlier this year when its marketing campaign went viral. Commuters spotted Manhattan billboards and ads plastered on the London underground that read: “Stop Hiring Humans.” In keeping with its motto, Artisan is run by a lean team of 35.

While the campaign may have been intended to turn heads, Carmichael-Jack doesn’t see human labor going extinct. “Human reps will still be key, but more like elite operators,” he says. Indeed, Artisan is looking to hire another 22 workers — including in sales.

Echoing this, Barbour also believes only the “really good” customer success reps and sales executives will remain. He expects them to become “highly sought-after,” and to receive increased salaries as a result.

So, it’s unsurprising that, according to PitchBook data, AI startups received 64% of all venture capital dollars in the U.S. during the first half of 2025. Illustrating this: Half the startups in the accelerator Y Combinator’s Spring cohort were companies building or designing tools to create AI agents.

The rate at which AI startups are hitting ARR milestones raises questions for investors. Before, if a startup was adding employees and expanding offices, it was assumed to be progressing, notes Right Side Capital, a pre-VC investment fund, in a recent blog post. But, when scaling costs drop dramatically, these markers become unreliable.

“Investors must now rethink traditional valuation frameworks,” the firm writes. “If a startup can reach $10 million ARR with a fraction of the burn compared to 2018, does it still justify the same valuation multiple? How does this change the size of early-stage funding rounds?”

At the same time, assigning AI startups high valuations based on the rate at which they’re growing revenue could be risky. Companies tend to experiment with many AI tools, Jamin Ball of Altimeter Capital, a VC firm, recently told The Economist. He calls this the “easy-come, easy-go” approach, meaning the companies aren’t committed to any one product. Ball suggests this creates an “experimental run rate,” rather than ARR.

This may explain why Blossom Capital, a tech-focused VC firm, has seen high-growth AI SaaS businesses with as much as a 10-20% monthly churn rate. (A “good” churn rate — the percentage of customers who leave in a given month — falls between 5-10% for B2B SaaS, LiveX AI notes.) On top of this, AI startups tend to use usage-based pricing rather than user-based, which can make revenue forecasting less predictable.

“AI is capable of doing all our jobs, my own included,” Klarna CEO Sebastian Siemiatkowski wrote in an X post in January. Echoing this, a survey by edX, a platform offering online courses on AI skills, found that about half of C-Suite executives believe “most” or “all” of the CEO role should be completely automated or replaced by AI — and 49% of CEOs agree.

But Barbour speculates that talk of AI CEOs has more to do with grabbing headlines than realistic forecasting.

“If [human executives] do a bad job, you can fire them, so they’re always going to be accountable for their decisions,” he says. An AI, though, can do “whatever it wants” without facing repercussions. That’s why Barbour foresees CEOs turning to AI assistants to help with decision-making.

“I find it extremely hard to believe that any company, especially public companies, would leave the final decision to AI,” he says.

Looking ahead, businesses are unabashedly announcing their intent to embed AI at the heart of their operations, gutting departments in the process. That may no longer be up for debate. But what remains unclear is how consolidated the AI landscape will become. Over 80% of AI projects will fail, according to research by the RAND Corporation, which is twice the failure rate for non-AI, tech startups. The challenge therefore may not lie in convincing investors that the future is automated, but rather, proving that running lean won’t sacrifice the weight of their products.