Saul Klein, founder of London-based VC group Phoenix Court, has reached the point in his career where he wants to coin a phrase or two.
Klein, who has raised £1.4bn across dozens of funds and invested in companies like Cleo, TravelPerk, Motorway, Monzo and Wise, has now invented a new way of assessing highly-promising startups. Out with “unicorns” and the obsession with hitting a $1bn valuation, he’s decreed, and in with “thoroughbreds” and “colts” — companies with annual revenues of $100m+, and $25m-$100m, respectively.
“Focusing on revenue is a really good evolution for the innovation economy if we want it to be taken seriously — and we need it to be taken seriously,” he tells Sifted over a video call.
In a new report on Europe’s best performing investors, based on the number of unicorns, thoroughbreds and colts in their portfolio, published in collaboration with data company Dealroom last week, Phoenix Court comes out top, followed by Index Ventures and Point Nine.
It’s convenient timing for Klein, who’s set to raise a new set of funds “this year or next year”, he confirms to Sifted. (The firm submitted an SEC filing for ‘LocalGlobe XIII’ and ‘Latitude IV’ in December.)
Colts and thoroughbreds
After trying (and only somewhat succeeding) in making the phrase “New Palo Alto” (an area encompassing London, Paris, Amsterdam and Brussels) catch on, I ask Klein what led him to invent two new terms.
“We’ve spent over a decade with the hope and promise of unicorns, even though, as we all know, private market valuations are an exercise in shared subjectivity.
“In the last two or three years it became self-evident that the innovation economy is becoming so fundamental to national and regional growth that we needed a better way to talk about the seeds we plant and what they grow up into,” he says, speaking from the firm’s HQ, Phoenix Court, an aesthetically-pleasing office/events venue/coworking space near London’s Kings Cross.
“But if we want policymakers, large corporates etc. to back — buy from, invest in — the innovation economy, we need to be selling more than hopes and dreams and promises. It’s hard to ask people outside of our bubble to give us credence unless we’re prepared to put up some real numbers.”
In Europe, the Middle East and Africa, there are now over 600 “thoroughbreds”, according to Dealroom’s report — and the top 100 made a collective $168bn in revenue last year.
AI ARR hype
But do the ARR numbers we’re seeing bandied around on LinkedIn by companies like Lovable count as “real numbers”, given they’re forward projections of revenue based on relatively few months of user activity, I ask?
“$100m ARR — however you want to get into the accounting of it — is a much stronger predictor of sustainability than a $100m valuation or even a $1bn valuation. I do think it’s an incredibly important threshold” says Klein.
“Although I do agree that until you’ve had two or three years of trading, it’s hard to say. Churn is the killer: you can pour 100 people into the top of the funnel, but if 80 of them leave, you’re just burning money to feed the top of the funnel.”
Still, says Klein: “It’s an incredible signal of the maturation of the ecosystem that we’re having this debate.”
When will the AI bubble burst?
The giddy speed at which a company like Lovable is onboarding new customers and seeing its valuation shoot upwards (funding offers, some at a $4bn valuation, were reported to be coming in thick and fast the day following Klein and I’s conversation) is reminiscent of the trajectory of some pandemic-fuelled startups like online events business Hopin.
Is investment into AI also a bubble, soon to burst? “There’s no question in my mind that there will be some breakouts,” says Klein. “But I think we’ll also have 18-24 more months of more ‘Hopins’, for want of a better word: ones that rise high, but don’t figure out that churn metric.
“There’s no question that the underlying infrastructure AI is bringing to software in general is fundamental, transformational, enormous… and there’s also no question that we’re seeing unprecedented levels of capex from all the big tech companies. And when you see governments focusing the way they are, investors focusing the way they are, there’s going to be a lot of waste.”
Across early-stage fund LocalGlobe and Series B fund Latitude, Phoenix Court has backed AI companies in everything from applied AI (Cosine Therapeutics, Gradient Labs, Dexory) to semiconductors (Graphcore) to foundational models (Mistral).
To figure out which AI companies have real potential, Klein says there are some key questions to ask. “At the application layer, churn is an incredibly important question, but you can’t really assess predictable churn for 12, 18, 24 months.”
“I’m personally cautious of these companies that just spike,” he says — adding that it’s a debate currently going on at LocalGlobe HQ. “If your price point is relatively low, the quality of your revenue can be very unpredictable.”
Other questions to ask, he adds, include: “What are your costs of acquisition? What’s your net promoter score? How close are you to an area the platform or foundational models might get into? Obviously as companies like OpenAI and Anthropic realise there’s real value in their system, they’re going to start crowding out companies in those areas.”
It might not be long, he hints, before we see “platform” AI companies — like OpenAI and Anthropic — move into “adjacent” applications, and perhaps start limiting which third-party applications can exist in their ecosystem.
Productivity software and vibe coding are obvious first areas for the platform companies to move into, he says. “I’m a lot less worried about a vertical AI — like legal — built on top of deep domain expertise, possibly unique and/or proprietary data sets that are so far down the list for the platform companies that maybe they’ll get to it three to five years from now. The core software tools they’ll get to sooner.”
The scaleup funding gap
If some US tech giants move in to acquire — or edge out — smaller European competitors, it could further exacerbate the growth funding gap between the US and Europe.
(One of LocalGlobe’s portfolio companies, enterprise AI platform Humanloop, saw its three cofounders and several of its engineers hired by Anthropic in August, in a ‘reverse acquihire’ that some industry observers think could become much more common.)
That scaleup funding gap currently sits at an estimated $30bn, according to the Dealroom report; an increase on the previous year. Is it showing signs of closing again anytime soon?
“If we’re being generous, look at the UK: in the last 12 months, the chancellor encouraged the Mansion House group [of large UK pension providers] to double the size of their commits [to venture] by 2030. The amount that the Treasury has given to the British Business Bank has also significantly increased, and the chancellor and the pensions minister are nudging local authority pension funds to consolidate to create bigger pools of capital. The plumbing from a policy perspective is coming into place.
“The big question is alright, it’s easy enough to make a commitment and say there’s £150bn, £200bn coming over the life of this parliament, but until we see the capital flowing it’s just a commitment. At best we have a trickle right now.”
When UK companies succeed, who gets the benefit? A teacher in Canada and a member of the Saudi Royal family. That’s not good for economic sovereignty.’
In the UK, 79% of capital invested into tech companies at the scaleup stage is non-domestic, Klein adds. In the Bay Area, it’s reversed, with 82% being domestic.
“The UK is the third best innovation economy on the planet, but only 20% of the capital going into our most promising companies is domestic, so when those companies succeed, who gets the benefit?
“A chimney sweep in Bavaria, a nurse in Denmark, a local authority employee in South Korea, a teacher in Canada and a member of the Saudi Royal family. That’s not good for economic sovereignty.”
Read the orginal article: https://sifted.eu/articles/saul-klein-big-interview-arr/
