From its inception, venture capital has been regarded as “The Magic Potion Department” of the investment community. Indeed, who other than magicians could grow unicorns at an industrial scale? Handpicking shells that bear potential pearls from the ocean of ideas, geeks and visionaries has never been easy. It requires special skills, mindset and strong guts to tolerate high risks and see future success where others cannot. Well, times are changing. Now that artificial intelligence (AI) seems to be reshaping each and every industry, VC is no exception. Let us have a peek into them and see if AI can become the new magic wand – or a curse – for venture capitalists.

*From scratch to an industry
*
What do VC funds generally do in order to function properly? First, they do the fundraising. Second, they screen and evaluate new investment opportunities. Third, they negotiate and close deals. Finally, they seek the right time and way for a profitable exit. Also, in the meantime, they have to maintain their back office to collect regular reporting from portfolio companies, communicate with investors and issue regular reporting of their own. Historically, VC funds mostly relied on human talent and labour to perform the majority of all its functions. Investment decisions were often based on gut feelings of key decision-makers or of the investment committee.

However, with time, data came to the aid of VC executives. As VC was increasingly becoming an industry rather than a set of boutiques, it amassed experience that could be turned into statistics. Various databases were developed, including Pitchbook, Crunchbase, Preqin, Dealroom, CB Insights and many others. They provided VC funds with data on the startup world, including acquisitions, funding rounds and much more. Data availability paved the way for its aggregation and analysis, including AI-powered techniques.

*From gut feeling to data-driven decisions
*
The most straightforward and most obviously beneficial way to employ this data is the back office operations. For instance, tracking portfolio companies along with competition, including online review monitoring, sentiment analysis derived from social media, and analysing trends in their online advertising spending. AI can also help track employee satisfaction in portfolio companies and their peers, which is an important indicator of a company's future growth potential and management effectiveness. Aggregating this data can assist decisions on later follow-on rounds and in VC funds’ own reporting.

One step further, AI can be used to identify recent deal activities and investment interest of potential acquirers of portfolio companies, thus saving time on seeking exit opportunities for VC funds. Similar AI can be used to track investment allocations and potential investment interest of institutional and private investors into VC itself. Next step: the wide variety of data available on financial transactions and start-up businesses across the world enable employing algorithms (developed in-house or adopted from external providers) to detect quantitative patterns in historical data from previous startups and extrapolate them to predict a new startup’s outcome.

*From data-assisted to AI-driven investment
*
Such tools have been in development for some time now. Not surprisingly, almost all the largest data platforms offer various AI tools that work on their data. For instance, there is a Pitchbook tool that predicts successful exits. CB Insights identifies emerging trends, investment opportunities, and potential risks. Crunchbase identifies relevant investors that meet their investment criteria. Dealroom helps to find relevant investment opportunities, connect with potential investors, and make informed investment decisions.

Therefore some VCs turn to algorithms to help with screening of their next investment targets. AI helps them discover companies seeking funding; identify early growth start-ups; determine appropriate investment timing. The list of VC funds using AI in different ways increased from just 10-15 in 2018 to 25 in 2019. However, the landscape is continuing to change drastically. Gartner, the global research firm, expects AI to be involved in 75% of venture capital investment decisions by 2025, up from less than 5% in 2021. Venture firms that employ an AI-driven approach include big names like EQT Ventures (AUM more than $227 bln), GV (founded as Google Ventures) managing over $8 bln and a plethora of smaller ones.

*From groundbreaking to mainstream?
*
So far so good it may sound, right? Not entirely so. In a recent study, it was discovered that the use of AI in screening does indeed generally improve the performance of funds. For instance, by helping to select companies with founders already having some success in similar industries. But it was also found that funds relying on AI in their decisions tend to invest in patterns, selecting startups that look somewhat similar to those that were successful before. This is absolutely consistent with AI relying on already published data – or documented past performance. AI may be perfect at identifying success patterns and compiling them, but it is unable (at least at the moment) to create or predict something completely new.

Thus, the study concludes, employing AI decreases the likelihood for VC to fund breakthrough companies, i.e. startups that achieve an IPO or obtain highly cited patents in the future. However, AI adoption by investors might hinder the allocation of capital to breakthrough innovations. Therefore, adoption of AI actually changes investment patterns of funds and how capital is allocated among innovative start-ups. Bringing this logic to the extreme would mean that, if overly employed, AI might actually dry out money flow to groundbreaking technology and actually undermine human progress in general.

*The markets will keep the balance
*
Indeed, the world would have been very different, if investment were driven in such a fashion by AI decades ago. AI would probably invest in WhatsApp, because it initially looked similar to Skype that pioneered the era of messaging software. This would have been a good investment. But it would also invest in, say, a new line of button cell phones in early 2006, just months before the launch of the iPhone. This would have been a disaster. Then, everyone knows that having a college degree increases your chances of success. So does AI when screening the founders’ background. Based on this pattern, it would probably discard pitches produced by Bill Gates, Steve Jobs, Mark Zuckerberg and Sergei Brin - all well known to be college drop-outs. So it would’ve been a world without Microsoft, Apple, Facebook and Google - very major contributors to tech innovation and gatherers of massive user and corporate data arrays. Here our imaginary experiment collapses, because in such a world the AI itself would probably have not appeared for several additional decades if it appeared at all.

Deliberate “pattern investment” is also a danger to investors themselves, not just the general progress. The most successful sectors can quickly become overcrowded and overpriced exposing investors to a bubble bust risk. This phenomena is perfectly reflected by the passive investment in index-bound ETF, a hype of recent years of cheap money and increasingly accessible markets. However, the good thing about markets is that no matter how overheated or brutal they may be in the short term, they are able to keep their balance in the longer run. “When too many investors rely on passive strategies, market efficiency suffers and opportunities for active managers emerge”, says David Trainer in his article published by Forbes. Trainer is the Founder & CEO of New Constructs, an investment research firm. Active investors will be increasingly finding ways to overplay the crowd on traditional markets. And this will probably be true for the VC landscape as well.

After all, combining two wizardries like AI and VC does not necessarily mean that new fancy ways actually are going to replace good ol’ ways. Like many other industries affected by AI, venture capital will most probably choose the middle way. Becoming AI-powered, VC might become more similar to conventional investment funds. It basically means maintaining a range of funds with different risk levels. For instance, a bunch of index-bound ETFs for the masses; some industry or theme portfolios for those with more knowledge and some exotic or individual strategies for the most rich and daring. With VC funds the lineup could look like this: AI-powered funds for most clients (maybe including a broader retail customer base) looking for stability and old-school handcrafted portfolios for those willing to take the risk of going out to the woods to meet some unicorns.