Contents
1. Investment trends in AI startups,
2009-2019 .............................................................................. 8
2. Exit trends for AI startups ..............................14
3. Business recommendations ..............................16
4. Policy recommendations ..................................17
C
ov
er
il
lu
sr
at
io
n:
A
le
x
/ A
do
be
S
to
ck
4 The road to AI – Investment dynamics in the European ecosystem
Methodology
The 2019 AI Global Index supplements the 2018 joint study entitled "Joining the
dots – A map of Europe's AI ecosystem", which presented a detailed picture of the
startups, research labs and communities in Europe, and divided European countries
into three categories based on the density and rigorousness of their AI ecosystems.
Among these countries, the United Kingdom (UK), France and Germany emerged as
the three leaders of the European AI ecosystem.
Roland Berger and France Digitale conducted desk research and built an original database
that covers 28 European Union members plus Norway, Switzerland and Israel.
France, Germany, the UK and Israel account for 80% of investments in AI startups
from 2009 to 2019. As such, the following analysis will focus on these four
countries, as the investments from elsewhere are too scant to draw relevant analysis
from them.
1. Investments1 trends in AI
startups, 2009-2019
A flourishing European AI ecosystem
The number of startups funded is one of the indicators
illustrating the dynamism of an AI ecosystem that sup- ports
entrepreneurship, attracts investments and realiz- es startup-
friendly public policies.
In line with the 2018 study, the leaders of the Euro- pean
AI ecosystem boast the highest number of start- ups funded
per 10,000 inhabitants in 2019. The UK is the top winner
with 590 AI startups funded, a ratio of funded startups to
population of %. France ranks second with 235 AI
startups funded and a ratio of %, while Germany ranks
third with 218 funded startups and a ratio of %.
In comparison with the European leaders, Israel
stands out from the crowd with 189 new AI startups
funded and a striking ratio of %. The country's strong
performance in funded AI startups to population calls for a
closer look into the dynamics of the ecosystem shaping its
performance. The dynamism of the Israeli AI ecosystem stems
from a local culture that strongly advo- cates an ethic of
entrepreneurship and internal business combined with strong
ties with the American market.
Ever-increasing
investments in European
AI startups
The four leaders in terms of number of AI startups fund- ed
(the UK, France, Israel and Germany) attract 80% of the
total amount of funds raised in this sector over the 2009-
2019 period, representing USD bn out of a total of
approximately USD bn in funds raised by AI start- ups.
The UK and France account for the biggest invest- ment
flows, with USD bn invested in the UK between 2009 and
2019 and USD bn in France.
1 The term "investment" encompasses BA, VC, CVC, private equity funds, crowd
equity and IPOs
Following its regular average
growth of +58% per year,
France is expected to score
the highest absolute growth,
with USD bn by the end of
2019 (projected results for the
second half of the year).
In the period from 2009 to 2019, Israel leads the way in
terms of funds raised per startup, with an average of USD
m in investments. France comes second with an average of
USD m. Germany and the UK follow be- hind,
respectively raising USD m and USD m per startup.
However, when considering investments in AI startups only,
France takes the lead over its European counterparts.
Globally, investments in AI startups have been at-
tracting more and more funds since 2014, with a com-
pound annual growth rate of 55% for each country.
For the first time, our data show a significant leader- ship
change. Following its regular average growth of
+58% per year, France is expected to score the highest
absolute growth, with USD bn by the end of 2019
(projected results for the second half of the year) due to
booming investments in Series C during the first half of 2019
with ContentSquare, Wynd, Shift Technology and record
fundraising of USD 230 m for Meero.
France topples Israel and comes out on top in terms of
investment attractiveness. There has been a spurt of funding
in France, while Israel has seen constant
1,268
1,241
Global
average
growth
+55%
1,025
902
802
760
610
634
621
567
516
465 447
401
510
452
401
309 316
258 239
273
228
255
191 188
213
128
109
144
100 118
48
83
52
AI-dedicated fundraising, or: the French breakaway
Funds raised 2014-2019 [USD million]
2014 2015 2016 2017 2018 20191
Total
USD 528 m USD 986 m USD 1,356 m USD 1,925 m USD 2,992 m USD 4,724 m
France United Kingdom Israel Germany Others
1 Projections for the entire year, figures available only for H1 2019
Source: Roland Berger
growth. Even though the figures for investments in Isra- el in
2014 were low (basis of USD 48 m in 2014), the country's
compound annual growth rate is extremely high at +80%
each year, a harbinger of Israel's astound- ing dynamism. In
Germany, the number of VC-funded startups has increased by
77% with the average amount invested remaining stable
between 2018 and 2019. In Israel, within the same period,
the number of VC-fund- ed startups has increased by 8%,
while the average amount invested has grown by 50%.
Overall, these fig- ures paint a very positive picture of
European dyna- mism, especially in France, which has
witnessed the number of VC-funded startups rise by more
than 100%. Mostly driven by seed rounds, this astounding
dyna- mism is fostered by both a promising pool of talents
and policies favoring entrepreneurship.
Driving forces of the European
ecosystem: The US and the UK are
taking charge
There is a visible polarization of funds emission, with the US
and the UK identified as the two main investment-is- suing
pools both for their dynamism and the significance of their
investments, the two countries having invested a total of USD
bn in Europe over the last 5 years.
However, these figures should not elude the fact that, by
comparison, the United States remains the indisput- able
leader of AI startup dynamism. In 2018, the United States
counted 70 exits for an overall investment of USD
bn and 510 transactions with average fundraising of
around USD 10 m. Europe counted more transactions
overall (983) with 62 exits for USD 3 bn invested, but the
2 Institut Montaigne. (2019). Brexit : qu’en pensent les entreprises ?
[online] Available at: quen-
pensent-les-entreprises
3 (2018). Intégrer « l’effet Brexit »: cerner les enjeux, évaluer les
risques, relever les défis ! [online] Available at: -paris-
There is a visible polarization
of funds emission, with the
US and the UK identified as
the two main investment-
issuing pools both for their
dynamism and the
significance of their
investments, the two countries
having invested a total of
USD bn in Europe over
the last 5 years.
average round reached only USD 3 m. In 2019, the aver- age
fundraising amount has rocketed in the United States,
shooting up from USD 10 m to USD 24 m with 55 exits, 378
transactions and USD bn of investment. Though
average funds raised also increased in Europe, the leap
remains significantly less impressive with 53 exits, USD
bn of investments, 516 transactions and average
fundraising of USD m.
In the perspective of a highly potential Brexit vote, UK
companies' willingness to move their business activities
beyond their borders2, the pressure of the weaker pound on
earnings, as well as weaker GDP growth in the UK (% in
the second quarter of 2019) creates new risks for the UK
economy. Accompanied by a prolonged delay in Brexit, such
volatility and economic pressures may weigh on future
European investments originating in the UK3.
Interdependency of
the European AI ecosystem:
The case of the UK, France and Germany
To varying extents, startup ecosystems in France, the UK and
Germany rely heavily on their domestic investors4. In
France, 73% of investors are French. In the UK, 65% of
investors are English. In Germany, 64% are are Ger- man.
In parallel, European investors represent % of foreign
investors in France (of which % are from Ger- many and
the UK), 11% in the UK (% are from France and Germany)
and 17% in Germany (% are from France and the UK).
In comparison with France and Germany, the UK AI
ecosystem's lower reliance on its domestic investors is a
reflection of the country's ability to attract a diverse ar- ray
of foreign investors. The top 5 investors in the UK startup
ecosystem are headquartered in 46 different countries,
whereas the same figure is 25 for Germany and 20 for
France.
Among the top 5 foreign investors within the AI start- up
ecosystems in the UK, Germany and France, US in- vestors
are highest in number, representing % of the foreign
investors in the UK, 14% in Germany and % in France.
The fact that US investors outnumber any other European
investors shows the underperform- ing state of cross-country
investments across the three European leaders, against the
backdrop of a prominent transatlantic link.
In line with the low number of active Chinese inves- tors
in Europe, China makes up just a small proportion of the
three countries' startup ecosystems. In France, Huami and
Hax, and in the UK, Lanting Capital, Alibaba
4 The number of investors is based on the top 5 investors in startups
available on Crunchbas
5 Forbes (2019). [online] Available at:
intelai/2019/02/11/ai-and-healthcare-a-giant-opportunity/#3d37d11a4c68
6 ai-
Group, Tencent Holdings, Arm Accelerator, Cherubic
Ventures and ZhenFund represent % of foreign inves- tors,
respectively. In Germany, on the other hand, Aliba- ba Group,
China Accelerator, Hax, Hillhouse Capital Group Linear
Venture, Sequoia Capital China and Tinavi Medical
Technology represent % of foreign investors. Overall, the
UK can boast of being more attractive to investors in and
outside of Europe than France and Ger- many, an
attractiveness that has enabled the diversifica- tion of the
UK's AI ecosystem. Furthermore, the three AI ecosystems'
heavy reliance on their domestic investors proves that Europe
is still far from being a "Digital Sin- gle Market", indicating
a need to further bring down barriers and foster cross-
country European investment.
Startup maturity:
The golden age of Series B and C
Healthcare and biotech (representing 13% of AI start- ups),
entertainment/media/culture (9%), financial ser- vices (8%)
and defense/security (4%) are driving the growth of AI
startups.
2019 trends in European AI startups by industry fol- low
the 2018 trends in which 70% of the startups fo- cused
on B2B services with 35% using AI for general/ cross-
sectoral applications.
Among the sector-specific applications of AI, health- care
and biotech witnessed a surge in European AI start- ups. The
four percentage point increase from 2018 re- flects the
swelling global use of AI to improve drug discovery,
diagnostics and patient monitoring and care, with the total
public and private sector investment in healthcare AI
expected to reach USD bn by 20215. Further, healthcare
AI startups received USD 864 m with 75 VC-backed deals and
financing rounds in the second quarter of 2019, up from USD
764 m in the second quar- ter of 2018. These figures are a
strong sign of the contin- ued enthusiasm in harnessing AI for
healthcare6.
AI-dedicated
investment-issuing pools
The case of France, Germany and the UK (2019 up-to-date figures).
INVESTORS' COUNTRY OF ORIGIN (%)
Others
USA
Chin
a
Europe1
(excluding France,
the UK and
Germany)
Franc
e
German
y
United
Kingdom
FRANCE GERMANY UNITED KINGDOM
1 Europe: European Union + Switzerland +
Norway Source: Crunchbase
In terms of startups' maturity, a structural shift in in-
vestment strategy has been observed internationally since
2015. While most investments were occurring at seed level
in 2015 (USD 171 m out of USD 685 m, repre- senting 27%
of total investments in AI startups), this figure has now
shrunk to 8% of overall investments. Investments in
Series A startups peaked in 2017 when they represented
44% of all investments, but they have decreased over the past
three years to a more moderate 25%. Today, 68% of
investments in AI startups occur at either Series B (31% of
all investments) or Series C lev- el (37% of all investments,
approximately USD m). Scaleup investments (Series D+)
have been replaced by ambitious Series C. Indeed, we can
observe growth in the average amount of investments in
Series C startups, which rose from USD 24 m in 2015 to USD
71 m in 2019, a factor of three. All in all, investors seem to
be more confident when it comes to their future exit from
com- panies, encouraging them to be bolder. They are
in- deed investing more money in Series C startups, as
evidenced by an investment of USD 170 m for Innoviz, and
Meero's USD 230 m fundraising. Startups do not need to
wait until the fourth round to finance interna- tional growth
anymore, which results in average Series C investments
being USD 80 m and the average for Se- ries D being
approximately USD 55 m. The average Se- ries B investment
is growing as well with a record fund- raising of USD 125 m
for the Berlin-based insurtech wefox Group. Series A
remains stable.
Now past the first phase of development during
which investments were targeted towards seed-level
startups, the ecosystem is today well established.
Hence, the current second phase sees investments tar- geted
to accompany growth (Series B and C) while con- tinuing to
fuel seed projects. These joint flows keep the ecosystem
afloat, preventing a loss of impetus.
Now past the first phase of
development during which
investments were targeted
towards seed-level startups,
the ecosystem is today well
established. Hence, the current
second phase sees investments
targeted to accompany growth
(Series B and C) while
continuing to fuel seed
projects.
10
3
Spotlight on Series B and C startups
Funding maturity evolution 2015-2019 [USD million]
In comparison with the US market, Europe is
not growing as fast as it could, underlining a
need for syndication
2018-2019 Evolution
311 685 929 1,484 1,795 3,909
100%
983
24
2014 2015 2016 2017 2018 2019 Transaction
s
Total amount
[USD bn]
Average deal
[USD m]
Seed Series
A
Series
B
Series
C
Series D Europe 2018 Europe 2019 USA
2018
USA 2019
Source: Roland Berger Source: Crunchbase - Full year 2018 - H1 2019
27% 25% 22% 16% 18% 8%
25%
44%
32%
40%
18%
23%
31%
16%
13%
32%
39% 11% 14%
26%
37%
29%
18%
7%
4%
13%
5%
7%
516 510
378
3
86
72
27
28
70
12 The road to AI – Investment dynamics in the European ecosystem
Spotlight on R&D
and patent registration
Strong R&D is one of the ways through which an AI ecosystem cements its position
internationally.
An analysis of the International Journal of Computer Vision – the
most cited European AI journal – between 2015 and 2019 highlights both
the strength of Europe's AI ecosystem(s) in terms of R&D, as well as
the extent of collaboration among European countries.
The US and China based institutions are at the forefront of AI
research, representing half of the institutions featured in the journal. At
the European level, the UK, France and Germa-
The international race in academia, or: most
prolific countries publishing research papers in
the International Journal of Computer Vision
(2015-2019)
362
ny-based institutions represent two thirds of the institutions featured
in the journal, a reflection of their established and mature AI
ecosystems.
The extent of collaboration within and outside of Europe provides an
insight into the greater trends within the Europe- an AI ecosystem. In
academia, there is strong collaboration among European academic
institutions, with 65% of Europe- an papers co-authored with a
European institution.
Parallel to the flow of funds from the US to Europe, both pay
attention to fueling noteworthy partnerships in academia. Among the co-
authored papers in the International Journal of Computer Vision, half are
co-authored with US academic in- stitutions. China, on the other hand,
represents a small yet significant proportion of co-authored papers, at
13%.
In addition to academic research, patents shed light on an AI
ecosystem's contribution to research and innovation.
The UK, Germany and France leverage their vibrant AI
ecosystems and are the source of 52% of AI patents granted. When co-
applying for a patent, European institutions pre-
fer institutions based in North America and China over their European
counterparts. Indeed, with 64% of co-applicants based in the US, 8%
in Canada and 8% in China, these three countries appear to be the most
attractive partners for Eu- rope, making up 80% of European
institutions' co-applicants.
Overall, in line with the trends in investment flows, R&D is
monopolized at the global level under the tripolar structure composed of
the US, China and the UK, fueling the growing
Others 47
Belgium 15
Italy 17
Spain
Switzerland
Germany
France
United
Kingdo
m
292
191
16
gap with the remaining European countries. Europe USA China Israel
Source: International Journal of Computer Vision
623
530
263
228
38
39
48
49
101
111
135
141
185
199
The road to AI – Investment dynamics in the European ecosystem 13
Patent competition: distribution of
patent registrations among European
countries, the US and China (2015-2019)
16,497
International Journal of Computer Vision:
focus on European contributors (2015-
2019)
2,595
Norway
32 Austria
Luxembour
g
Denmark
Belgium
2,517
Greece
%
Slovenia
%
Czech Rep.
%
Denmark
%
Austria
%
Finland
Ireland
%
Hungary
%
Romania
%
United Kingdom
%
Spain
Ital
y
Finlan
d
Sweden
Switzerland
Netherlands
Ireland
%
Sweden
%
Nether-
lands
%
Belgium
%
Italy
%
At the European
level, The UK,
France and
Germany-based
institutions represent two
thirds of the institutions
featured in the journal, a
reflection of their
established and mature
AI ecosystems.
France
Germany
Spain
%
United
Kingdo
m
Switzerland
%
Germany
%
France
%
Europe USA China
Source: International Journal of Computer Vision Source : Crunchbase
2. Exit perspectives for AI startups
Over the past five years, the number of exits has in-
creased by 64% (CAGR) overall. The UK, Israel, France and
Germany are the leaders on exits, with 66% of total exits
occurring in those countries from 2014 to 2019.
Among the big four, the UK outperforms the others with
28% of all exits, representing 66 exits out of the 236
counted in Europe and Israel over the 2014 to 2019 period.
Exceeding the number of exits in France (28) and
Germany (24) over the period, Israel ranks second with 38
exits, which is 16% of the total, the same per- centage as
Italy, Norway, Finland, Denmark, Russia, Ireland, Poland,
Hungary, Portugal, Ukraine, Luxem- bourg, Cyprus,
Belarus, Bulgaria, Turkey, Austria, Lith- uania and Greece
put together. In 2019, Spain caught up on its European
neighbors, ousting Germany from the winning quartet with
6% of overall exits in Europe, and Israel taking its place.
Israel ranks first in terms of startup exits with a 10% ratio
of exits per number of funded startups (38/374) for a total
of USD 446 m, followed by France with a 7% ratio (28/430,
USD 25 m), Germany with a 6% ratio (24/383 USD 133
m) and the UK with a 5% ratio (66/1267, USD 315 m) over the
period from 2014 to 2019.
Corporates are the main acquirers
of AI startups, followed by private
equity firms and investment companies
Between 2014 and 2019, acquirers have been mostly
corporates (92%), of which 70% are tech companies,
followed at some distance by private equity firms (% of
all acquirers) and investment companies (%).
American acquirers are by far the most numerous, with no
less than 42% of all acquirers coming from the US,
followed at far lower levels by the UK (13%), Germany
(%) and France (7%). Although Israel is one of the
main recipients of AI-targeted investments, very few
acquirers (2%) are based in that country. Looking at
Over the past five years, the
number of exits has increased
by 64% (CAGR) overall. The
UK, Israel, France and
Germany are the leaders on
exits, with 66% of total exits
occurring in those countries
from 2014 to 2019.
Among the big four, the UK
outperforms the others with
28% of all exits, representing
66 exits out of the 236 counted
in Europe and Israel over the
2014 to 2019 period.
2019 in isolation, acquirers are mostly tech companies and
corporates (96%) and most of the acquirers come from the
US (32%), the UK, France and Germany, with Israel coming
in equal fifth with China, Singapore and Italy, having two
investors each (less than 4% each).
Taking some critical distance in considering the
identity of buyers, three methodological issues per-
taining to tech M&A deals focusing on AI-dedicated
companies can be identified as being behind the high
number of unprofitable deals7. In this context, it ap-
pears that buyers tend to focus on what they can get out of
purchasing an AI company rather than on what they need to
do to make their new business successful. Moreover, buyers
are usually in "take mode", which en- ables the seller to
foster competition between poten- tial buyers and
consequently increase their price. In- deed, the seller
intends to extract all of the cumulative future value from
the transaction. Finally, it is quite clear that buyers do
not necessarily understand the market they are trying to
penetrate8.
The final limit to AI startup acquisitions, though far
from the least important, touches upon the increas- ing value
of tech M&A deals worldwide. Undoubtedly, acquiring tech
companies is getting more and more expensive. Though
total deal values are increasing, fewer deals are
concluded today. In 2015, 4,422 tech M&A deals were
recorded for a total deal value of USD 574 bn. Only three
years later in 2018, 3,617 European AI tech M&A deals were
concluded for a total deal value of USD 573 bn.
7, 8 Harvard Business Review (2011). The Big Idea: The New M&A Playbook.
[online] Available at: playbook
Between 2014 and 2019,
acquirers have been mostly
corporates (92%), of which
70% are tech companies,
followed at some distance by
private equity firms (% of
all acquirers) and investment
companies (%).
The treacherous path of M&A deals
TECH M&A WORLDWIDE
Tech M&A deals worldwide 2009-2018
[USD bn; # of tech M&A deals]
Deal volume distribution by acquirer age and type [% of
tech exit deals; 2018]
3,023
144
3,293
180
3,788
225
3,654
192
3,963
3,284 388
247
4,422
574
4,050
456
573
3,617
3,555
341
2007-2017
1997-2007
1987-1997
1977-1987
1967-1977
1957-1967
1947-1957
1937-1947
1927-1937
1917-1927
Before 1917
60% of tech
exit deals are
driven
by young acquirers,
companies
founded within
the past 20 years
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 0 5 10 15 20 25 30 35
Total deal value of tech M&A Total number of tech M&A deals Startup Corporate PE/VC fund
Source: Press research, Statista, Roland Berger
3. Business recommendations
Venture capital activity in Europe is on the rails. France
remains a leader in artificial intelligence in Europe.
Funds keep fueling the development of late-stage start- ups
while steadily supporting early-stage projects. How- ever,
major rounds are continually being entrusted to American
and Chinese investors, each with their own strategy, with
amounts substantially exceeding Europe- an funding flow
capacities.
I. On future investments
Overall, there is a lack of growth funds in Europe. To curb
this shortage, new investments should be made on the basis of
strategic joint approaches by venture capital and private
equity funds. However, venture capital funds are not
taking the plunge due to a lack of resourc- es. Similarly,
private equity funds are standing on the sidelines as they
seemingly have not yet realized the sense of such a strategy.
As for corporates, such actions exceed their financial means.
II. On tech M&A
Globally, it appears that M&A activity is driven by Amer- ican
acquirers and is mostly entrusted to tech compa- nies as
private equity funds are not positioned in this tech
segment. Very few corporations are investing in AI-driven
startups, most probably because accomplish- ing a
successful acquisition remains a difficult task, with a 70%
failure rate in tech M&As.
To mitigate this negative trend, tech M&A requires the
adaptation of acquisition and integration processes. In
dealing with acquisitions, acquirers should adopt the
perspective of the venture capital world by investing in future
revenues. They should also work closely with management
teams to ensure a shared vision. Finally, key assets (among
them tech and people) should be properly secured,
leveraged and integrated. In terms of integration, buyers
should protect the newly acquired activity from being
cannibalized by the core business, while defining agile and
inclusive governance to avoid the departure of the initial
founders. Finally, relevant KPIs should be listed and
efficiently monitored.
To succeed in developing European AI champions,
countries should organize themselves to support the
growth of startups internationally and thought should be
given to developing a dynamic private equity ap- proach
to tech. Companies should also be adequately supported in
leveraging AI assets.
4. Policy recommendations
With AI being a top priority for the new EU Commission and
as the new European Parliament takes up its legis- lative
work, this new term will prove decisive in whether the
continent is able to embrace change and make poli- cies work
to boost artificial intelligence in the EU. This study proves
that Europe has the potential to become the world leader in
innovation, provided the European institutions develop a new
and ambitious strategy and promote AI-friendly regulations.
Europe will only be able to push its new AI standards
globally if its ethical ambitions are coupled with efforts to
boost a top-notch AI industry across the bloc.
In the spirit of the European Commission's Coordi-
nated Plan on AI9, we recommend a three-pillar ap-
proach based on funding, talent and regulation, to sup- port
the growth of investments and promote the
independence of the European AI ecosystem.
I. The European AI
ecosystem should take action
to support the funding of AI
startups
The EU's AI startups suffer from limited and polarized
funding. The average European VC fund is EUR 60 m, half
the size of a typical US fund, and 90% of the avail- able VC
funds are found in just eight EU Member States (Denmark,
Finland, France, Germany, the Netherlands, Spain, Sweden
and the United Kingdom)10.
9 European Commission (2018). Coordinated Plan on Artificial
Intelligence. [online] Available at:
digital-single-market/en/news/coordinated-plan-artificial-intelligence
10 European Startup Network (2017). A Manifesto for Change and
Empowerment in the Digital Age. [online] Available at: https://
europeanstartupnetwork.
eu/wp-content/uploads/2017/10/
1. The EU's financial institutions should take appropri- ate
measures to develop pan-European funds to foster cross-
border investments, especially in a post-Brexit context.
2. The European Investment Fund (EIF) should play a
major role in supporting the emergence of late-stage
funds. For example, within its fund-of-fund activities, the
EIF has EUR bn in European VC funds. But its growth
is limited, and new investment vehicles need to be created in
order to unlock the necessary capital to empower European
scaleups to challenge Chinese and North American
decacorns.
3. The tax framework for venture capital across the EU
should be standardized to mobilize capital inside and
outside of the Union to avoid double taxation. Today, for
example, a Danish pension fund does not recog- nize the
French FPCI (Professional Fund of Invest- ment Capital)
and hence will not invest in it. A pan-Eu- ropean VC
ecosystem will thrive only if savings in Member States
can be mobilized equally in every other Member State.
4. Corporate venture capital has become a major inves- tor
in tech and AI. 681 unique corporate investors ac-
counted for % of total investments in tech (EUR bn)
in 2018. In order to support the flow of corporate funds
across Europe, tax depreciation schemes for cor- porates are
encouraged within EU Member States.
5. In the case of business angels, whose investments
continue to be small and concentrated in only a few
Member States, the Member States themselves may
grant tax breaks. For instance, Belgium granted a tax
break of 45% for investing in new shares issued by a
startup and 30% for investing in new shares in an SME or
startup fund.
6. The EU institutions should facilitate access to
cross-border crowdfunding, which has come to play an
important role in equity investment. In 2015, European
crowdfunding platforms raised EUR 422 m, represent- ing
more than 10% of all venture capital raised that year (EUR
bn). Despite the potential, the legal uncertain- ties
surrounding crowdfunding investments have limit- ed its
growth beyond national borders. To counter this, the EU
should foster the transparency of cross-border crowdfunding
by setting up a simple and transparent cross-border
framework favoring mutual recognition of nationally
regulated crowdfunding platforms11.
7. Finally, the control of foreign investments in compa- nies
deemed sensitive for their use of critical technolo- gies such
as AI, robotics, cybersecurity and quantum technologies is
strengthening both at EU and national levels. Given the high
number of foreign investments in European AI startups,
additional controls over the in- vestments might place a
strain on European startups. In order to mitigate a downward
pressure on the invest- ment in European AI startups in the
coming years, Eu- rope should support the growth of
European funds that can be directed towards acquisition of
and investment in European startups.
11 European Startup Network (2017). A Manifesto for Change and
Empowerment in the Digital Age. [online] Available at: https://
europeanstartupnetwork.
eu/wp-content/uploads/2017/10/
12 New York Times (2017). Building AI That Can Build AI. [online] Available at:
II. Capital follows talent
Global AI talent is scarce, with just 10,000 people hav- ing
"the education, experience and talent needed" to develop
AI technologies12.
In this context, Europe needs to follow a holistic
strategy by maintaining and supporting the level of ac-
ademic excellence in its academic institutions, retain- ing
European talent in the face of "brain drain", and attracting
talent outside Europe.
In educating future academics in the field of AI, Eu-
ropean higher education institutions should integrate a
cross-disciplinary curriculum, with a particular focus on
applied ethics and humanities. Further, available funding
and attractive . positions need to be put in place, to
support both vibrant and competitive aca- demic research
and the level of attractiveness of aca- demic institutions
for researchers.
Moreover, to attract and retain talent, Europe must
position itself as an attractive destination for entre-
preneurs.
First, the complex recruitment processes and limit- ed
work permits, which vary between countries, should be
simplified with the creation of a European Startup Visa.
In light of the initiatives taken by several EU countries
to attract entrepreneurs, such as Denmark, France, Ireland,
Italy and the Netherlands, the visa, de- livered on a
multiannual basis, would simplify the ad- ministrative
procedure of recruitment.
Second, a unified share option scheme should be
established for startups and scaleups to support their global
reach and facilitate competition with larger cor- porate
players. This scheme would give startups and scaleups the
opportunity to issue standardized share options across the
28 countries of the European Union.
III. The EU regulatory
frame- work needs to make
room for AI
The EU must seek greater
harmonization and strategic alignment
The gap in the maturity of Europe's AI ecosystems may be
further exacerbated by the diverse array of national AI
strategies. Since Member States were encouraged to develop
and implement national AI strategies (. by the Villani
report in France) before a European approach was adopted, the
continent must pay attention to the harmonization of
national strategies and European efforts. European
countries must work in synergy to mutually compensate for
different strengths and weak- nesses in patents,
infrastructure, investment capacity and human resources.
Beyond GDPR:
The EU must ensure the free flow of data
A thriving data-driven economy is essential for a func-
tioning Digital Single Market. Yet barriers between Eu-
ropean countries can make it difficult for entrepreneurs to
fully exploit the potential of AI technology.
Even though data is the key ingredient for AI appli-
cations, Europe has imposed the strictest rules in the world
on the use of personal data, reflecting widespread concerns
over privacy. The General Data Protection Reg- ulation
(GDPR) has had a crucial positive impact world- wide in
terms of creating a more regulated data market. More
importantly, the regulation on the free flow of non-
personal data, applicable as of May 28, 2019, is a major
follow-up to the GDPR and is a pillar in facilitat- ing cross-
border business in the EU and enabling the scaleup of
innovative data services.
Now efforts must be intensified. Although the EU intro-
duced the GDPR to harmonize data protection rules
across countries, there remains a patchwork of different
interpretations in the EU on the extent to which compa- nies
can process private and public data. Plus, the regu- lation on
the free flow of non-personal data initially did not adequately
address how it would interact with the GDPR. Also, the
regulation on non-personal data did not account for the reality
that many large datasets inevita- bly contain a combination
of both non-personal and personal data. The recent
publication of practical guid- ance for businesses on how to
process mixed datasets contributes to addressing how
organizations should ap- proach such challenges.
The EU must remove
barriers to data flows worldwide
In Europe, data sharing is still the exception rather than the
rule, putting European entrepreneurs at a disadvan- tage to
others. Additionally, a no-deal Brexit may further hinder
access to data, affecting AI innovation and dyna- mism in
Europe.
Regulations limiting the ability of startups or any oth- er AI
company to transfer, download or upload data across the
world sends the wrong signal. In order to en- sure data access
and enable European startups to exploit the flow of data,
especially between Europe and the UK, as well as with the
rest of the world, European institu- tions must be guaranteed
in any (future) international trade agreements and their
counterparts, and adden- dums to current agreements should
be made. The agree- ments must ensure a level playing field
for all parties in- volved by subjecting them to the same
limitations and liabilities. The EU-Japan agreement is a model
in this re- gard, in that it allows personal data to flow freely and
safe- ly between the two partners. Both parties agreed to recog-
nize each other's data protection systems as "equivalent", thus
creating the world's largest area of safe data flows.
Conclusion
"The road to AI – Investment dynamics in the European
ecosystem" reiterates the trend highlighted in our 2018 study. The European
AI ecosystem, under the leadership of France, the UK and Germany, has
experienced growth in the last five years, with the annual growth rate of funds
raised by startups reaching 55% and France leading in investment attractiveness.
However, the European AI ecosystem is still fragmented and suffering from
a lack of integration, which is further endangered by the scheduled Brexit on
January 31, 2020.
That is why 2019 represents a milestone for the future of the unified European
AI ecosystem. The newly appointed EU Commission, with its ambitious
political roadmap, should seize the opportunity to design a favorable regulatory
framework for the next decade, in sync with the European AI ecosystem.
