Artificial Intelligence Fuels the Rise of Next-Gen Startups

The hottest AI companies today are no longer the foundational developers like OpenAI, but rather startups that package AI technology into user-friendly, easy-to-deploy applications.

A wave of new startups is drawing attention by building applications on top of AI models from OpenAI and other major developers – Photo: Bloomberg

Not long ago, Silicon Valley remained indifferent to startups like Harvey. While OpenAI focused on building cutting-edge artificial intelligence (AI) models capable of transforming nearly every industry, Harvey took a humbler approach: creating software that makes OpenAI’s technology more useful for lawyers.

Winston Weinberg, CEO of Harvey, once remarked, “The market saw companies like ours as just ‘GPT wrappers.’” The term was a jab at startups that seemingly did nothing more than repackage OpenAI models. He added, “If investors want to put their money somewhere, they’ll choose OpenAI or Anthropic.”

But now, these so-called “AI wrappers” have become the center of attention. Walk into any venture capital (VC) office in Silicon Valley, and you’ll hear buzz about startups offering AI chatbots, research tools, and applications for coding, healthcare, and customer service - all built on large language models (LLMs) from top AI developers.

These startups are witnessing explosive growth in revenue and valuations, all while operating with just a fraction of the cost spent by foundational model developers on chips, data centers, and talent. Harvey, founded in 2022, surpassed $50 million in annual recurring revenue by December, according to Weinberg. Similarly, Anysphere - the company behind the popular AI-powered code editor Cursor - is said to have reached $100 million in annual recurring revenue, according to anonymous sources.

Michael Mignano, a partner at Lightspeed Venture Partners, likens today’s AI wave to the mobile app boom nearly two decades ago. “Just like when the iPhone launched and millions of new mobile apps appeared,” he said. Mignano, an investor in Granola - an AI note-taking app using OpenAI and Anthropic technology - believes, “Now, with AI and LLMs, we’ll see millions of new AI products.”

INVESTORS FLOCK TO AI APPLICATION STARTUPS

Investors are eagerly pouring money into these services. Harvey raised $300 million earlier this year in a round led by Sequoia, pushing its valuation to $3 billion. Anysphere followed suit, raising $105 million in a January round led by Thrive Capital and Andreessen Horowitz, valuing the company at $2.5 billion. Investor demand is so high that Varun Mohan, founder of Codeium - a company offering AI tools for developers - said the traditional fundraising process is “no longer something we have to do.”

Codeium is currently raising funds at nearly a $3 billion valuation in a round led by Kleiner Perkins.

Of course, giants like OpenAI and Anthropic continue to attract massive funding from VCs in large-scale rounds. However, investing in these high-valuation companies is becoming too costly for some funds. There’s also growing skepticism about companies spending billions to build models, especially after Chinese startup DeepSeek claimed it had developed a competitive model for just $5.6 million. Furthermore, some prominent model developers have either dropped out of the AI arms race or been partially acquired by Big Tech, raising concerns about consolidation in the industry.

Jesse Zhang, CEO of Decagon - a startup building customer support agents powered by OpenAI and Anthropic - observed, “Right now, AI applications are clearly the best place to invest because that’s where the revenue and customers are. The models will keep getting better and cheaper, and the applications will benefit the most from those improvements.”

A NEW GENERATION OF RISING STARTUPS

No company illustrates the promise of the AI application era better than Anysphere. Founded in 2022 by four friends from MIT with only internship-level experience, Anysphere has seen explosive growth thanks to its AI-driven code editor, Cursor.

Anysphere’s investors tout it as the fastest-growing software startup of all time, surpassing the previous record held by cloud security company Wiz, which reached $100 million in annual recurring revenue in 18 months. Anysphere did it in just 12 months.

By combining LLMs from various providers, Anysphere built a developer tool that saves engineers time and has gained widespread popularity among both software developers and the tech elite willing to pay for it. Brian Armstrong, CEO of Coinbase Global Inc., posted on X that 100% of the programmers at his company use Cursor.

Miles Grimshaw, a partner at Thrive Capital and investor in Anysphere, commented, “There’s a new generation of talented engineers who know how to use this computing model to reshape product experiences.”

GROWTH BRINGS CHALLENGES

Despite rapid growth, Anysphere currently has only about 40 employees - reminiscent of early mobile app success stories like Instagram, which also began with a lean team. Meanwhile, Harvey has grown to 260 staff and plans to double its headcount this year using part of its new funding.

Rapid expansion, however, comes with challenges. Like Anysphere’s founders, Harvey’s Weinberg had little experience prior to launching the company, having only briefly worked as a lawyer. “I’ve never done this before, I won’t lie,” he admitted. “The challenge is how to scale - not just the company, but yourself.”

The abundance of capital also comes with potential pitfalls. Shiv Rao, CEO of Abridge AI Inc. - a platform for AI-powered medical conversations that raised $400 million in the past year - warned, “I don’t think raising money just because it’s available is a good idea. That can lead to trouble quickly if startups aren’t disciplined.”

But perhaps the biggest concern is whether OpenAI or another major LLM developer might eventually compete directly with these services - similar to how Apple has copied popular apps on the App Store, or how Meta has disrupted smaller apps with feature rollouts.

For now, investors remain optimistic. Mignano of Lightspeed believes it would be “naïve” to think the major LLM providers will build tailored applications for every industry. Instead, the task of bringing AI to specific sectors - from law to healthcare - may fall to a legion of small but high-potential startups.

Source: Thanh Minh / VnEconomy.vn
View original article here.