Berlin The start-up industry has been struggling with financing problems for months – but if money is still flowing into start-ups, then it is most likely into those related to artificial intelligence (AI). Since the beginning of the year, only three new “unicorns” have emerged in Europe, two are active in the field of AI. First, the Cologne-based translation service DeepL achieved the coveted status of a young company in a new round of financing, which is valued by its investors at more than one billion dollars, followed shortly afterwards by the London-based data analysis company Quantexa.
It seems that at least AI start-ups are unaffected by the acute funding crisis that is putting pressure on all other start-ups around the world and has already triggered a large wave of layoffs.
“Artificial intelligence is the topic par excellence. Investors are rushing into this space for fear of missing out,” said Tristan Post, head of the AI Founders program, which helps founders build AI startups. Accordingly, a whole new start-up generation is emerging.
So far, most of the news about AI has mainly come from the USA, where Microsoft partner OpenAI has now launched ChatGPT-4, the fourth, very capable language model, and Google is now following suit with Bard.
Bitkom President: “The race for AI is not over yet”
However, things are now also moving in Europe. It seems the enthusiasm radiating from Silicon Valley is contagious. “The race for artificial intelligence is not over yet,” says the new president of the tech industry association Bitkom, Ralf Wintergerst: “ChatGPT was a wake-up call.”
France’s President recently visited the tech fair Vivatech – a sign for young companies too.
(Photo: IMAGO/ABACAPRESS)
The current market data supports his thesis. An analysis by the venture capitalist Atomico shows that generative AI in particular – which can produce entire texts, program codes, images or videos – is now attracting investors.
35 percent of AI financing in Europe now flows into this specific sector, significantly more than last year, when the share was five percent. Investors have relatively little money when it comes to language models because they expect high returns from sales and IPOs due to the hype. McKinsey experts assume that generative AI could boost the global economy by several trillions of dollars.
There are 23 AI unicorns in Europe and 142 in the US
The data service Pitchbook determined exclusively for the Handelsblatt that there are now 23 active AI unicorns across Europe, and in the USA there are even 142. In Europe, the software AI unicorn Contentsquare from Paris is at the top of the ranking list, followed from Europe’s most valuable chip start-up Graphcore from Bristol and the tutoring platform Gostudent from Vienna. Five British start-ups are in the top ten: Graphcore, Cera, Builder, Exscientia and Quantexa.
Guillaume Lample, Arthur Mensch and Timothée Lacroix want to develop a language model like ChatGPT with their start-up – and have collected more than 100 million euros in just a few weeks.
(Photo: David Atlante)
In the UK, AI start-ups are receiving massive government support. 100 million pounds are available for the development of language models alone. And the number one, Contentsquare from France, is also getting tailwind from politics. Just recently, French President Emmanuel Macron announced at the Vivatech tech fair that he intended to spend 500 million euros to promote future AI champions.
>> Read more here: Europe’s most valuable education start-up acted anti-competitive
The federal government’s AI strategy, on the other hand, dates from 2018. In the ranking of European AI start-ups that have collected the most funding over the past few years, the first German company only appears in ninth place. It is the robot technology developer Agile Robots from Munich.
105 million euros after four weeks
The AI hype fuels the desire to found a company, which is apparently only topped by the desire of financiers to invest in this area. “Today there are two start-up areas for which there is more money than good companies: artificial intelligence and sustainability including cleantech or climate tech,” says Julian Riedlbauer, partner at the consulting and investment company GP Bullhound.
The start-up Mistral, which wants to build an AI language model for European companies, was formed in France just a few weeks ago. Already in the seed round, i.e. right at the beginning, Mistral collected 105 million euros from investors. The German fund Headline was also among the donors.
The founder of the financial start-up Merantix warns of the danger of a bubble.
(Photo: Merantix)
For comparison: since it was founded five years ago, Agile Robots, the most valuable AI start-up in Germany, has only earned three times the amount that Mistral came up with within four weeks. But even Mistral is still a long way from the US dimensions: OpenAI from Silicon Valley has now received a hundred times the Mistral financing.
Mistral does not yet have a product on offer, but headline boss Christian Miele sees enormous potential: “Mistral could become the AI infrastructure for the European economy and society.” Such language models promise a particularly high return. The large round of financing was necessary because training the model requires massive processor performance, says Miele, who is currently also head of the German start-up association – and that is expensive.
According to Riedlbauer, “the usual evaluation patterns of software companies” do not apply to AI: “Many evaluations are even decoupled from the classic key figures, in some cases you can see recognizable exaggerations.”
>> Read more here: SAP loses AI boss – Feiyu Xu founds start-up
The previous AI boss at SAP, Feiyu Xu, also wants to benefit from the trend. She turns her back on the Dax group to found the start-up Nyonic.
Nyonic also wants to develop a basic model for AI. This should be suitable for special applications in business and support several European languages. Similar to Mistral, Nyonic has yet to find customers. However, experts see a market for large customers from the economy.
Is there already an AI bubble?
It is now undisputed that generative AI will have a significant impact on how we work and interact in the future – and that is why investors are willing to take high risks. In view of the plentiful flow of funds, the first experts are already wondering whether a bubble is emerging. Many are reminded of the rapid rise of the quick commerce market, fueled by plenty of cheap money, with start-ups like Gorillas, Flink and Getir.
This text is part of the large Handelsblatt special on artificial intelligence. Are you interested in this topic? All texts that have already appeared as part of our theme week
You will find here.
Meanwhile, consolidation with many layoffs is in full swing there, and many venture capitalists (VCs) are at risk of losing money. Newer players such as alpacas or yababa have long since disappeared from the scene.
Bubbles are notoriously difficult to spot, especially once they begin to inflate. The founder and head of the Berlin AI investment platform Merantix, Adrian Locher, sees at least the danger of a bubble: “90 percent of the current VC-based AI start-ups will not work. A lot is being financed at the moment, but I think very few things will make a big difference in the end.”
Cream cake alone is useless
Philipp Hartmann from the European AI initiative “Applied AI” advises taking a close look at the business model: “If you buy someone else’s cream cake, put a cherry on top and then sell cream cake with a cherry, that’s probably not good positioning.” Hartmann also gives examples. If your start-up is only based on the ChatGPT language model, you always have to fear that OpenAI itself will enter the field. And if OpenAI decides to raise prices, for example, you can quickly come under pressure.
Radek Zaleski from the Polish consulting company Netguru points out that so far only a handful of companies, including Microsoft, have presented a more or less scalable business model, “while the rest continue to work on the basis of vague ideas”.
>> Read more about it here: Investment bankers – an endangered species
Naturally, the Quantexa from London sees things very differently. The British hopeful raised $129 million from investors in March and is now valued at $1.8 billion.
Quantexa wants to break into the data analytics market. The company calls its real-time AI solution “decision intelligence” and has already won Europe’s largest bank HSBC as a customer. The financial institution uses Quantexa to screen more than 689 million transactions for signs of financial crime every month. The start-up emphasizes that the business is growing rapidly.
However, an IPO is not currently in the pipeline. Only going onto the trading floor can provide the liquidity that makes investments particularly attractive.
Only 15 percent of companies in Germany use AI
Locher from the investor Merantix therefore states the goal: “We want to build AI start-ups that have the chance of going public.” However, the start-ups need customers early on for this. It cannot be the case that a start-up only finds out after a few years whether there is a demand for what it offers – there is no right to exist from the outset.
But that’s still difficult at the moment. According to a Bitkom survey, around three quarters of the companies surveyed assume that AI is of great importance for the future competitiveness of the German economy. But so far only 15 percent use AI in their own company. “The more AI applications come onto the market, the more companies should and – this is my hope – they will take advantage of them,” says Bitkom President Wintergerst.
Here you will find more texts on artificial intelligence
In the end they end up with Google and Co.
The companies need the right specialists for this – but according to a comprehensive survey by the venture capitalist Sequoia, there should be no shortage. According to this, the proportion of AI experts in Europe as a percentage of the population is 30 percent higher than in the USA – and almost three times as high as in China. A total of almost 200,000 engineers have already gained experience in the field of artificial intelligence, and around 43,000 are even proven specialists with extensive AI knowledge.
However, these AI experts often work for the US corporations Google, Microsoft, Meta and Amazon. According to the Sequoia study, Amazon and Google in particular currently have AI expert teams in several European cities. Compared to start-ups, no matter how well financed they are, these large companies usually pay higher salaries – and ensure a lot of reputation on the CV.
According to Sequoia’s talent manager Zoe Hewitt, AI experts are still increasingly opting for a start-up as an employer. This is because they can be at the forefront of new visions in young companies. And that’s worth more to many workers in this golden age of innovation.
A kind of gold rush mood prevails not only among founders and investors, but also among AI specialists.
More: Energy start-up 1Komma5 Grad becomes a unicorn
Collaboration: Arno Schütze, Dennis Schwarz and Peter Köhler