A Blog by Jonathan Low

 

May 10, 2023

Investors Pour Billions Into Generative AI Startups, No Business Plan Required

Money is pouring into venture funds focused on generative AI startups.

Despite a still somnolent overall venture market, generative AI is creating a mania for AI investing. The excitement is based perceptions that this truly is the fabled 'new, new thing,' the next wave of tech innovation that will transform how everything is done, just as the PC and the cell phone did before it. JL 

Deepa Seetharaman and Berber Jin report in the Wall Street Journal:

While most of Silicon Valley’s venture-capital ecosystem remains in the doldrums, investors this year have been pouring funds into companies specializing in generative AI. Many of the companies getting backing are new and unproven. Venture investment in generative AI companies will be several times last year’s level of $4.5 billion. The market for AI applications in enterprise technology alone will rise to $98 billion in 2026. An open-sourced library for software developer tools with no revenue, recently closed a $10 million round at a $50 million valuation. Weeks later, Sequoia Capital quadrupled the valuation to $200 million.

Before their startup had customers, a business plan or even a formal name, former Google AI researchers Niki Parmar and Ashish Vaswani were fielding interest from investors eager to back the next big thing in artificial intelligence.

At Google, Ms. Parmar and Mr. Vaswani were among the co-authors of a seminal 2017 paper that helped pave the way for the boom in so-called generative AI. Earlier this year, only weeks after striking out on their own, they raised funds that valued their fledgling company—now called Essential AI—at around $50 million, people familiar with the company said.

While most of Silicon Valley’s venture-capital ecosystem remains in the doldrums, investors this year have been pouring funds into companies like Essential specializing in generative AI systems that can create humanlike conversation, imagery and computer code. Many of the companies getting backing are new and unproven.

Analysts at research firm PitchBook predict that venture investment in generative AI companies will easily be several times last year’s level of $4.5 billion. That is driven in part by Microsoft’s $10 billion investment in January into OpenAI, the startup behind the wildly popular ChatGPT bot. In comparison, such investment totaled $408 million in 2018, the year OpenAI released the initial version of the language model powering ChatGPT.

Entrepreneurs and their backers are hoping generative AI will change business activities from movie production to customer service to grocery delivery. PitchBook estimates the market for such AI applications in enterprise technology alone will rise to $98 billion in 2026 from nearly $43 billion this year.

As with the recently ended bull run of broader startup investing, though, investors often are jumping into AI startups even when it isn’t clear how they will make a profit—especially since the computational power required to train AI services can sometimes amount to tens of millions of dollars a year or more. The sudden influx of capital is also encouraging many AI researchers, some without management or operations experience, to start their own companies, adding to competition.

“It is undeniably a major inflection point, and great products and companies are going to be built,” said Matt Turck, an investor specializing in AI at FirstMark, a New York-based venture-capital firm. But “as in prior hype cycles, a lot of this will not end well. The market cannot sustain, all of a sudden, a million different companies with half-baked ideas. It feels like the gold rush.”

 

The current AI boom accelerated after OpenAI’s late November launch of ChatGPT, which became one of the fastest-growing consumer apps in technology history. OpenAI is one of several capital-intensive startups building enormous algorithms called large language models that can produce original text. 

Earlier this year, Google invested more than $300 million into OpenAI rival Anthropic. The company expects to raise up to $5 billion within the next two years to advance its large language model, which powers the startup’s ChatGPT rival called Claude, according to an investor presentation reviewed by The Wall Street Journal. 

Overall, U.S. venture funding plunged 55% to $37 billion in the first quarter, according to PitchBook. But AI-related events still boast a boom-time vibe. Prominent venture capitalist Ron Conway held two days of festivities in March for portfolio companies backed by his fund, SV Angel, many of which are incorporating AI in some way. The event’s speakers included former President Barack Obama and OpenAI Chief Executive Sam Altman, and attendees including the musician and entrepreneur will.i.am mingled with founders over drinks, according to people who attended.

To court talent, venture capitalists are also hosting bar nights and events in the chic San Francisco neighborhood of Hayes Valley, sometimes called Cerebral Valley, which has emerged as an AI hub. NFX, a venture firm that backs several generative AI startups, outfitted its Hayes Valley offices with a red-walled speakeasy and turned its garage into an event space with a dance floor.

 

Sizable deals are coming fast and furious. Eleven Labs, a voice AI startup, recently raised money from Andreessen Horowitz that valued the company around $100 million, according to people familiar with the matter. Humane, a startup created by two ex-Apple executives, recently raised $100 million from investors, including Mr. Altman, to make AI-powered wearable devices, the Journal reported in March.

Character.AI, an AI startup founded by two other former Google employees less than two years ago, raised $150 million in a recent round that valued it at $1 billion, the company said. Toronto-based generative AI startup Cohere has raised $250 million, which would value the company at $2 billion, according to people familiar with the matter. The Financial Times earlier reported Cohere’s fundraising.

One of Character’s founders, Noam Shazeer, co-wrote the 2017 Google paper with Mr. Vaswani and Ms. Parmer that invented a more efficient and accurate way for AI systems to analyze blocks of text. Cohere co-founder Aidan Gomez was also a co-author on the paper.

“It’s pretty vicious right now,” said Scott Beechuk, a partner at Norwest Venture Partners, which has backed companies including AI-powered search engine You.com. “It’s a mad swarm of VCs. Everyone wants in on those companies.”

 

The competition is enough that firms are willing to back well-respected researchers without so much as a business plan, according to several investors who have reviewed such startups, many of which ended up getting funding. 

“You can have a brand-name researcher launching something, in some cases, at absurd valuations,” said Jordan Jacobs, managing partner and co-founder of Radical Ventures, a five-year-old Toronto-based fund that focuses entirely on AI investments. 

Ms. Parmar and Mr. Vaswani raised funds from VC Thrive Capital and others for Essential in January after leaving another AI startup they co-founded a year earlier, according to people familiar with the matter. Essential’s founders are still fleshing out its plan, people familiar with their plans said. For now, they have opted to create a service that would help companies use their own data with multiple language models, one of the people said. 

Adept, the pair’s previous company, is developing tools that complete tasks when prompted by a human. Adept said in March it had raised $350 million from investors including General Catalyst and Spark Capital.

The heat around AI deals has intensified so much that many investors are worried the startups will falter due to the pressure to deliver. Several investors told the Journal that startups raising seed rounds are pushing for valuations of tens of millions of dollars or more. Some companies are raising back-to-back rounds of funding. 

Mr. Beechuk said he invested in an AI-powered software sales startup called Lavender as part of its Series A funding round, which took place weeks after the startup raised money as part of its initial seed round. “I don’t think that deal would be doable now,” he said. “There would be so much competition.” 

LangChain, an open-sourced library for software developer tools with no revenue, recently closed a $10 million round at a $50 million valuation, according to people familiar with the matter. A few weeks later, venture-capital firm Sequoia Capital swooped in with more funding and quadrupled the valuation to around $200 million, they said.

0 comments:

Post a Comment