Rise of AI Summit 2020 · Berlin · Society & Strategy Day | Future Space

Challenges of the German AI Ecosystem

Vanessa Cann · Managing Director, KI Bundesverband e.V. (German AI Association)

At the Rise of AI Summit 2020, Vanessa Cann, Managing Director of the German AI Association (KI Bundesverband), presents findings from a study run with the German Startups Association, Telekom hubraum and the Institute for Applied Artificial Intelligence in Stuttgart. She argues Germany has strong preconditions, ranking second in AI research behind China and fourth in industrial GDP, but still fails to translate research into AI products, with only two German companies appearing on a widely used list of the top 100 global AI startups. She points to a large venture capital gap, a scarcity of AI startups compared with Israel, and weak connections between startups clustered in a few cities and the small and mid-sized companies that make up most of the economy. She proposes four steps toward AI leadership: clear and proportional regulation, more education on AI and entrepreneurship, direct state investment in young companies, and incentives for investors and established firms, then details her association's regulatory recommendations before taking audience questions on incentives, liability and why the well-known funding problem persists.

Key takeaways

  • Cann says only two German companies appear on a widely used list of the top 100 global AI startups.
  • She cites a 36.3 percent gap between the 57.5 percent of German AI startups wanting VC funding and the 21.1 percent that get it.
  • Israel, about the size of Hesse, has roughly 1,300 AI startups versus Germany's 300, with 30 times higher per-capita AI investment.
  • She proposes regulating AI through existing sector rules and measuring it against human performance rather than 100 percent accuracy.
  • She says SMEs, 90 percent of the economy, rarely meet AI startups because startups cluster in a few cities near VCs and tech universities.

Detailed notes

Strong assets, a missing product layer

Cann says Germany ranks second in AI research, just behind China and ahead of the US, and fourth in industrial performance measured by GDP, calling both genuine assets for its AI ecosystem. Despite this, she says Germany is not world class at building AI products, noting that ecosystem maps of the top 100 global AI startups tracked by CB Insights list only two German companies. She attributes part of the gap to how research spin-offs are funded: startups founded out of research institutions often have to share revenue or give up significant equity to the institution, which she says creates a weak foundation for raising later-stage venture capital.

The funding gap with Israel

Cann contrasts Germany, with around 300 AI startups, to Israel, a country about the size of the German state of Hesse with roughly 1,300 AI startups. She says German AI startups do receive more funding than the average German startup across all sectors, but 57.5 percent of them want venture capital funding while only 21.1 percent obtain it, a 36.3 percent gap concentrated in later-stage capital; per-capita investment in AI companies is around 30 times higher in Israel than in Germany. As a result, she says German AI companies tend to stay small and under-invest in marketing and investor visibility, unlike companies in Israel and the US that use funding to scale and internationalize.

Startups and SMEs rarely meet

Cann says three of four German AI startups already cooperate with established companies on a B2B basis, treating German industry as an asset, but that small and mid-sized companies, which make up 90 percent of the economy, rarely meet startups because startups cluster where venture capital and technical universities are, mainly Berlin, Munich, Hamburg and Karlsruhe. She says only about one in seven companies is currently working on AI products, though she notes improvement: in 2018, 87 percent of established companies said AI was not relevant to them, while now 22 percent are planning an AI project and 6 percent are already using AI.

Four steps toward AI leadership

Cann outlines four priorities: clear and proportional regulation; more education on AI and entrepreneurship, from schools through universities to SMEs and corporates, so that research talent also learns to commercialize; direct state investment in young companies, noting the federal government has earmarked five billion euros for AI that young companies have so far barely accessed; and tax and other incentives that make it more attractive for investors and established companies to work with AI startups.

A blueprint for proportional regulation

Cann details the association's regulatory recommendations: proportionality, since risk depends on the use case rather than the algorithm, so high-risk applications should be regulated while low-risk ones keep a light bureaucratic touch; measuring AI against human performance rather than an unattainable 100 percent accuracy standard; regulating AI through updates to existing sector rules, such as in mobility or health, rather than writing new AI-specific law; treating a single general AI regulation as impractical, since AI is hard to define legally and not all software is labeled as containing it; and ensuring regulation does not add extra bureaucratic burden on SMEs and startups, which she calls the main drivers of innovation.

Q&A highlights

Challenged on why SMEs need extra incentives rather than being drawn in by promising research, Cann says many SMEs still do not understand AI and need education and peer learning from other SMEs, not only from startups. Asked whether regulation will deter founders, she says clarity matters more than the amount of regulation, since current law predates the digital transformation. Asked about liability when AI itself acts, she calls it an unresolved regulatory issue needing clarity. Asked about the main bottleneck for scaling startups, she says it is partly cultural: investors want proven numbers and a finished product rather than backing a vision, unlike in the US, Israel or China, and says the state must also fund young companies directly. Asked whether founders should consider leaving Germany, she says startups already look abroad for VC funding and that Germany must help companies scale and exit domestically rather than losing them. Asked why the well-known funding problem remains unsolved, she says there is a gap between agreeing on the problem and acting on it, and points to the promised five-billion-euro federal investment as a step toward closing it.

Quotes

It's not that the talents are missing, it's about an entrepreneurial mindset that we have to get going.

Vanessa Cann

I think we have to invest more in visions than in numbers.

Vanessa Cann

Questions answered in this talk

How does Germany's AI startup ecosystem compare to Israel's?

According to Vanessa Cann at Rise of AI 2020, Israel, though about the size of the German state of Hesse, has around 1,300 AI startups compared with roughly 300 in Germany, with per-capita AI investment about 30 times higher.

Why do German AI startups struggle to raise venture capital?

According to Vanessa Cann at Rise of AI 2020, 57.5 percent of German AI startups want VC funding but only 21.1 percent get it, a gap of 36.3 percent that is especially pronounced for later-stage capital.

How does the German AI Association want AI to be regulated?

According to Vanessa Cann at Rise of AI 2020, the association wants proportional, risk-based regulation built on existing sector rules rather than a single general AI law, with AI measured against human performance and no extra bureaucratic burden placed on SMEs and startups.

Mentioned

AI ecosystem, venture capital gap, proportional regulation, risk-based regulation, value by design, research spin-offs, sector-specific regulation

Keep exploring

More from Vanessa Cann

More from Rise of AI Summit 2020

More on Enterprise AI & Adoption

Transcript

Auto-generated transcript, not edited. · 4,579 words

Show the full transcript

and we are live again from rise of ai's futures space welcome back everyone i hope you fed your algorithms and optimized that analog metabolism of yours during our lunch break we will dive right in again with our next keynote the topic is challenges of the german ai ecosystem really do we have challenges well i thought everything was going just great but wait and see our speaker has a background in political consulting for fortune 500 companies she is a mentor for startups and an advisor to women in ai but first and foremost she is managing director of the german ai association germany's biggest ai company network please give it up for Vanessa Cann hello and thank you for the introduction it's great to be here today and speaking from the drivery i would like to use the next 20 minutes to speak to you about the challenges of the german ai ecosystem or at least what i perceive as being the challenges so actually we have the ideal preconditions we have a world-class ai research so

actually we rank second when it comes to ai research just behind china we're actually above usa so we actually have ai research as a huge asset to our ecosystem and we're really great at research and also another asset we have a world-class industry so we're fourth here when you look at the statistics of the gdp which is a great indicator of industry performance and and that is also a huge asset towards our ai ecosystem but still we are not world class at building ai products and when you look at this ecosystem map for example um and that's just one of the couple that are around there um it labels all the ai startups the 100 global ai startups that cb insights found promising here there's a couple of ecosystem maps around they all vary in the logos and company names they give but they have one thing in common they only show two or only a few ai companies here in this in this field as conox another so only two ai companies from germany

and that's just on a list of hundred for that that we have these huge assets too few and here's why the one point and argument i want to make is that ai research is still not transferred into products and services so especially our spin-offs so the startups that are being been being funded and founded from our research institution they're facing large obstacles often they have to share their revenue they have to give up a lot of the equity so things that make a bad foundation for later stage for finding later stage capital and bcs to invest in and so it does not really come as an a surprise that ai companies in germany are quite rare you can see this in a comparison between israel and germany in israel and keep in mind that israel only has a landscape about as big as our state of hessel so only a fraction of germany but still they have around 1300 ai startups while germany only has 300.

when you look at this from an ecosystem perspective so this is our landscape how it looks in germany so it's it's um it's around 300 companies in germany well this is the ecosystem map in israel so there's just a load more ai startups there and that is a really really big issue here and the second point and argument i want to make is that our ai companies are still lacking funding so there's a big gap between what is actually the need of the ai startups when it comes to venture capital funding and what the capital market is giving them you can see here that actually when you when you look at all startups in germany so across all branches across all technology fields you can see that ai startups are actually getting more funding than all startups put together but still when you look at the preferred source of funding that you can see here in pink it's 57.5 percent of asis startups are wishing for vc funding but only 21.1 percent are actually getting

it so the venture capital gap the funding gap here is large at 36.3 percent and it's particularly later stage capital that is missing so compared to israel for example we have a per capita investment this is 30 times higher in ai companies than it is to germany and that's just something that's um that's also showing its impact on the ecosystem so when it comes to the results our companies are just staying small the teams have to focus for a long time on actually building their product their service whilst the companies in israel and also in the usa can really grow and scale and internationalize they usually invest a lot in becoming more visible they invest in marketing and therefore become more visible towards potential clients they invest in in also an invisibility towards venture capital firms so then again they get more investment and that's why these companies are going so so high and are so so much better in their performance and the third point i want to mention at this point is that

our ai companies are still seeking clients and that might come as a surprise because i said in the beginning our established industry is one of our biggest assets but even though three of four ai startups are already cooperating with established companies and you can see here that most of them has have really seen established industry also as the asset of the german market so most of them are working together with other companies are working on a b2b basis or working with different companies and not b2c with clients and and customers so it's really down to working with established industries and still it's especially smes so the 90 of our economy those that are really that's the backbone of our society and our economy and that rarely meet startups because startups are based there where the venture capitals are so they rarely meet them usually based in berlin or in munich some of them in in hamburg or even where the big tech universities are so around the kaiti in kaiser or but that's not where

actually the smes are sitting and that's a big issue so so they rarely meet and they really have the opportunity to actually collaborate on that basis so it doesn't really come at a surprise that our industry at the moment is not very much investing in ai especially the smaller companies aren't investing in ai around every seventh company at the moment is is working on ai products so that's still a wide range of opportunities that we have here that we haven't really um yeah and up to so it's a big issue but i am optimistic because the discretion discussion is winning in speed uh also of course because the corona crisis so i think that's one of the uh the big benefits we've seen in this year that in 2018 87 percent of of ai company of established companies have said that ai is not the topic for them whilst today actually 22 are planning to use ai a planning ai project or six percent are already using ai so if you want to learn more

about what artificial intelligence is all about and how our ecosystem is comprised please feel free to check out our study that we brought about in autumn together with the german startups association the telecom hubram and also the institute for applied artificial intelligence based in stuttgart you can find it on our website both in german and english language whichever you prefer but i think the question that is open now is how can we still become world leaders in ai or can we still become leaders in ai and i think my answer to this question is definitely yes we can we can still still become world leaders in ai and here's just four points that i want to outline which i think are the steps we should go towards a brighter AI feature and also a better ai ecosystem so the one point i want to stress is we really need clear and proportional regulation that is something that's very important for our small companies i'll dig into that in a second because i think that's one

of the most substantial points at the moment the second point i want to mention is we really need more education on ai starting in schools going for universities and also ending in like a smaller and medium-sized companies and also corporates they really need to know what asai is about and what benefits ai can have for them and the same with entrepreneurship especially the people the experts working in in research they need to learn more about entrepreneurship because it's not that the talents are missing it's about an entrepreneurial mindset that we have to get get going so that the research results that we already have really hit the road and and are transformed into products and services the third point that i find important is investing in young companies this is not only like up to the business angels and the vcs to invest in companies i think it's also down to the state so now it's five billion that the federal government has deployed for for the use of artificial intelligence young companies have so

far only seen little of this i think we really have to invest in young companies because that is where the invasion is coming from and then the fourth point i want to mention is the incentives we have to give towards investors and also established companies to invest in ai and i think that's all down to also tax incentives but giving them the opportunity to really see the potential of ai and working with ai companies but as i said i want to dive deeper because the time is only short into one topic the clear and proportional regulation you've just seen one um yeah one keynote about this topic and and you've seen that especially on the european level it's a very important topic right now and there's a lot going on so we've discussed this on quite a wide range in our association on how we want to get going when it comes to regulations what our recommendations are and this is what we ended up with so one point is for us that proportionality is

key so in in general it's that algorithms are not good or bad it really boils down to the use case you deploy ai in so it depends on which use cases you use artificial intelligence in and there are use cases that are a threat that are risk towards our democratic values or our risk towards our lives and these are the regulations that we have to have in place so we have to regulate risk-based use use cases but there's also very many use cases that don't actually present a threat and they should have the bureaucratic lean way to just go ahead and bring their product their service to the market because otherwise we won't have the potential to to also strive on a worldwide basis the second point i want to mention is that ai applications should be measured against human performance so eventually ai is not at a hundred percent accuracy and i think it will never actually be there in many use cases so what we actually have to measure it against is that

human performance is not at hundred percent either so we should still have to have an ai application that is better than what the human benchmark is but our orientation should always be the human benchmark so the ai should be better than than the a than the human but still we cannot say that an a application should be at hundred percent before it hits the road and then the third point is that ai use cases should be regulated by by using existing regulation so actually the wheel does not have to be reinvented in that way we've always as a society have regulated different fields different industries more stricter because they were a more a bigger risk towards our our life but vigorous towards our democracy for instance the field of mobility or the field of health so we should really look at these industries that are already actually regulating ai use cases and they should be reviewed and in many cases updated so that it's clearer how we can actually regulate ai so we should really

go ahead and regulate ai along the industries that are already regulating the field so that we don't have a doubling on that end and now many stakeholders are actually pushing for the fourth point so they're pushing for general ai regulation but we actually do not see a general ai regulation as being feasible on the one hand it's very difficult to define what a.i actually clearly is especially on a legal basis and it is also very difficult to see which software actually entails ai because not all software is labeled with the ai inside so it's very difficult to actually see which software do i have to regulate so in the end we'll probably have to check every software every digital process to see if there's an ai inside and if it has to be regulated on that sort and then the fifth point i want to stress and that's something that's very important for us is that regulation must not place an extra burden on smes and startups those are the main innovation drivers at the

point and they will be in the future so we really have to take care that we are not burdening them even further with the bureaucratic and bureaucratic regulations we should really see that we foster smes in startups and using ai and that's one point that's very important for us and then to come to an end who we are we are the german ai association and um the members of our association are both innovative smes they're startups ai experts all with a focus on the development and application of ai technology and with being so we are the german we have germany's largest ai network and it's our mission to really strengthen germany's ai ecosystem we're pushing towards a more sovereign european union we want to strengthen the ai ecosystem by bringing together entrepreneurs across europe so that they can exchange but also so that they can learn from each other and and build their network and then the third thing that's very important for us is also bringing and fostering openness and innovative strength in established

companies so that they can master the digital transformation so i'm looking forward to tackling ai together let's tackle ai together please feel free to reach out for me to me when you are interested in either becoming part of our network or you're an established company and want to see how you can interact with ai companies and you can find my email address here you can also find me on linkedin and now i'm looking forward to your questions thank you yes thank you very much vanessa we don't have too many questions from the audience yet so please dear audience feel free to ask more we're delighted to take them let me maybe first challenge some of the things you said which i widely agree upon but still i think there are some ideas that we could discuss um one thing is you would love to have incentives for smes and established companies right and isn't this part of what is wrong with germany that having great ai technology promising startups great research not being an incentive

enough that we have to give the companies an additional cookie to even try to work with startups isn't that like the the primordial sin maybe of our ecosystem well i think i think i i agree to you in some parts but i think it's also because ai is such a complex topic i think many smes actually do not know what ai is so far about and also how it can benefit them so i think we really have to help smes and learning and understanding more about ai we have to break it down to the use cases so it really boils down to which which fields ai can be used in and can really be beneficial for the companies and i think they also they should not only learn it from the startups they should also learn it from other smes so i think it's very important for smes that are the pioneers in this field that they they market this more openly because i think it's very important for our industry not to be afraid

of sharing data not to be afraid of working with artificial intelligence but really embracing what ai can do for for them and i think um corona showed how much potential there is inside it so i really hope that many are seeing this potential right now right thank you there are two maybe complimenting maybe contradicting questions the first one is will regulation of ai prevent companies from starting at all yeah so i hope not because i think in the end it's it's not so much about more regulation or less regulation i think it's down it all boils down to the clarity of regulation so at the moment our ai companies are working with very outdated um legislation so all the legislation that they are working with is of course or was put into place before the digital transformation actually came into being and especially ai so i think it's good that the ai commission is very much pushing for regulation in this field but i think we really have to yeah we have to find the

middle way in that sense that we really have to look at which fields are risky and these have to be regulated right and you might have answered a little bit of the second maybe complementary question doesn't regulation always have to place a burden on the companies that are regulated to be effective at all i i actually do not think so because what i can see right now is that many ai companies human by design or value by design is very much in their dna so i think a lot of companies are deciding to to launch their company here in the european union because they they think values are so important in artificial intelligence so so many of them are already discussing how data security privacy transparency and so so on can play a part or be incremental to their product so i think we really have to get there and that we give them principles so that it's easier for them to see how they can actually produce a product that's value by design or

human by design in the end right i see let's talk about another idea that i don't want so much to challenge as to maybe dive a little bit deeper into you said there in many fields we should not even try to invent new regulation because we have existing regulation in place for a certain sector for instance um but for a lot of the laws that we have i think we have an actor in mind that we can hold responsible for something for instance like the driver of a car obviously this is a thought that is um we're going into the uncanny valley when we think about an ai for instance or any piece of software as an actor this is something we will surely have to solve what is your view on this do you see this problem how would you solve it yeah so i i think it is an issue i think it's a question especially for the ai startups that are working very much with with companies that are then kind of

building the product for the end consumer because i so i think it is a very big regulatory issue and i think we need clearance on that and that's why it's so important that this is being discussed right now right um you told me or you told us and i've heard it uh very often that we have we especially have problems in the endgame for startups right when it comes to to scaling up um is this the like the main problem of your members as well think in part it is it definitely is a big issue um also in corona times many were also seeking for for example state funding uh i was saying that like kind of the state also has to has to start investing in in young companies and not only into the established companies um but i think i think it's it's very much a culture question like the entire ecosystem in a way is a culture question and i think we can see this in the vcs too so i think

it's very much showing your business plan looking into numbers and and having a product that's hundred percent and you can't really expect that from a company that's only in like in the starting positions and as build is building up that in the company right now so i think we have to invest more in visions than in in numbers and i think that that's something where we have to learn uh compared to to us or israel china yes i absolutely agree of course this probably will not change overnight and now there are a lot of people thinking about founding a company or maybe they just founded the company one or two years ago and think man maybe i didn't made the wrong decision to to start one in germany like is there a possibility if you're like a couple of years in and you you see that you can't make it here because there simply is not enough capital to change gears to maybe move to the silicon valley or would you would you in maybe

some cases advise for people to leave germany before starting a startup well i think what we're seeing is already that startups are actually not only seeking vc money in in europe but they are already looking abroad i think that's something especially from the german ai association's perspective that we want to and have to change so i think it it has to be has to be our um obligation as as germany to not only see how our companies can launch here with for example money that we're already investing with scholarships like exists and those are programs that really help companies to launch in the beginning but we really have to um yeah seek how we can play out in the end game too and how we can help our companies also exit here and not um kind of giving the the fruits to other countries what we have kind of at the beginning planted i guess right thank you there's a question from marcos i'm not sure if i understand it completely are your members willing

to exchange openly on big data and ai um i maybe he means exchanging big data right or maybe exchanging code or ways to work with with one another because i mean the exchange about it like just talking about it they're all willing to but uh yeah yeah what do you know the position of your members uh definitely so i think it very much goes into the direction that ai startups especially have to find solutions about the awareness of data and they're often working with established companies who are giving them data and smes especially aren't too much open at the moment to sharing their data they're very much still hesitant and and worried about what might happen i think that this give the gdpr didn't really help in that terms and that we still don't have enough clarity on on what is allowed and what not but i think especially startups are rooting out uh in our network but also across the european union and trying to work together and collaborate more more easily because i

think that's the only way how they eventually will grow right how good is the image of the german startup scene when it comes to ai how open are the doors of vcs overseas for instance um so more open as i showed and showed in earlier so they are they are better off than the startups in general but they are more off often looking also for vc funding so um more of the ai startups are very very much dependent on bc capital to to actually uh go through with that company um but these vcs are not waiting for german ai companies to come over um not so much but but they're definitely interested and and they have a global perspective so they don't actually invest in a national market they invest in the companies that they find more promising and if it's a german company then they will invest in it right um let me ask you one question because you've obviously having a lot of talks also with people from vcs people from the realm

of politics and so on and i think the problem is widely known right the problem is widely known that we have a problem with investors we have a problem in the in the endgame of startups so to speak um many people agree but i think there is not much underway to solve the problem right do you do you agree and do you have an idea why yeah i think there's still a step between agreeing and really um understanding for yourself okay i think that that might often be the gap actually uh so so at the moment i am seeing that there are plans as i said the federal government is planning to invest five billion so so i think there's really maybe two more as we heard from dorado yeah yeah they're thinking about it yeah so i think i think there's a lot of um interest in investing in this area but the question is how to invest and and how can we really bridge the gap between uh research and and actually then

what we get onto the streets i think that's the biggest challenge that i'm seeing right now and i think it's it's really up to all players from vcs to to sme so to the potential clients to to really the research institutions to just become more open and really uh foster young companies to be established right and that is why your work is so important and why we are so happy that you're contributing to ai thank you very much thank you thank you thank you

Rise of AI 2027 · Berlin · May 2027

Be in the room for the next one.

Ten years of talks started as conversations in one curated room in Berlin. The next edition brings 300 selected leaders together in May 2027.