OpenAI’s release of ChatGPT and the immediate, subsequent buzz has put generative AI on the map for venture capitalists and tech firms eager to put money into a smart idea in a downturn where smart ideas feel scarce.
Generative AI is the catchall term for artificial intelligence that can create something new — text or images for example — from existing data, when given prompts. OpenAI’s generative AI tools can seemingly do anything from write poetry and code to producing frame-worthy art, triggering huge excitement in the space.
And venture capitalists and companies want to get in early on tech that has the potential to be transformative — especially at a time when most tech moonshots aren’t working out.
For example: Microsoft is weighing up whether to invest a further $10 billion in OpenAI at a $29 billion valuation, according to reports, having first invested $1 billion in 2019.
AI copywriter Jasper AI raised $125 million in Series A funding in October – high for an early-stage round at the tailend of dismal funding year for startups.
And in December, New York-based Runway, which uses AI tools to aid in video editing, raised $50 million at a $500 million valuation.
These are notably large individual deals, and this year looks set to see the funding rush continue.
But are we headed for an AI bubble?
—François Chollet (@fchollet) January 8, 2023
The wild revenue multiples are here
One sign of frothiness in private tech startups is whether their valuations seem out of whack with their business potential.
Generative AI startups are already securing sky-high valuations despite little evidence of commensurate revenue, let alone profit — reminiscent of the low-interest-rate boom times that saw firms with unproven business models fetch valuations worth billions of dollars despite having little to show.
OpenAI’s reported target valuation of $29 billion would be on revenues in the “tens of millions”, according to the Wall Street Journal. That could mean a revenue multiple ranging from around 290x to 2,900x. OpenAI is commercially promising thanks to its tie-ups with Microsoft, but that is a big show of faith.
Jasper AI secured a $1.5 billion valuation from its recent round, despite expecting to end 2022 with just $75 million in revenue, according to TechCrunch, a 20x revenue multiple.
By way of comparison, publicly listed enterprise AI firm C3 trades at around a 1.3x revenue multiple.
Another sign of a bubble is a cash rush. Pitchbook figures suggest VC investment in AI hit $1.37 billion in 2022 across 78 deals globally, representing almost the total invested over the past five years.
How generative AI generates revenue is unclear
All of this might be justifiable if there was a clear path for AI firms to make money. But the current state of AI is still expensive, and profits are theoretical.
—Kyle Harrison (@kwharrison13) November 21, 2022
For Dr. Richard Windsor, founder of research firm Radio Free Mobile, conclusions that ChatGPT is about to topple Google are premature. OpenAI, he said, is “a long way from challenging Google,” particularly as the firm subscribes to what he calls “the infinite monkey theorem of AI.”
In his view, OpenAI’s approach is one of: “If you throw enough data and enough compute at a problem then the answer will eventually pop out at the end.” Consequently, he said, OpenAI uses a lot of computational power.
In a research note, Morgan Stanley analysts compared ChatGPT’s cost-per-query to Google’s cost-per-search. At an estimated average of $0.02 per query, they estimated that OpenAI’s costs are about seven times higher than Google’s $0.003 per query.
The difference, the analysts note, is the significance of “the compute intensity of ChatGPT’s natural language models.”
They added that even if OpenAI has access to the lowest-priced tiers for Azure, the Microsoft cloud service that hosts ChatGPT, queries made on the AI chatbot would still be four times more expensive than an average Google search.
Some of the hype around generative AI may be justified and the technology feels genuinely exciting. But to avoid another bubble, investors would do well to quiz the fundamentals of generative AI businesses, before handing out hefty valuations.
