Are Venture Capitalists Being Bamboozled by AI Startups?

AI startups raised $16.8B from VC firms in 2018 and a staggering $18.5B in 2019.


Nerissa Zhang

3 years ago | 6 min read

AI startups raised $16.8B from VC firms in 2018 and a staggering $18.5B in 2019. All that money, what did it buy? A closer examination reveals a few disturbing issues in the venture capital model.

AI startups raised $18.5 billion in 2019, setting new funding record

While overall U.S. venture capital funding dipped in 2019, the artificial intelligence sector bucked this trend by…

Despite their bravado about helping “daring founders” and “pushing the boundaries of what’s possible”, the truth is that venture capitalists are in the business of making investment returns for their LPs. Early AI startups indeed did very well — for the VCs.

Venture capitalists and four of their bets on AI — a compare & contrast.

Self-driving startup Cruise raised a total of less than $20M in Seed and Series A funding between 2014 and 2015, and was sold to GM for $1B in 2016.

This is a phenomenal win for the VCs. Even though after the GM acquisition, Cruise still had to raise over $5B in subsequent rounds. Today, the most widely used self-driving technology is actually from Tesla, not GM. But hey, VCs already won, so none of that bothers them.

DeepMind is another example. After raising a rumored $50M it was acquired by Google in 2014 for $500M, without ever making a dime in revenues.

Again the VCs came out ahead. Despite their impressive feat of defeating human grandmasters at the game of Go, DeepMind caused quite a bit of problems inside Google, struggled for direction, and ended up losing key people. Again, who cares? Not the VCs.

Examples like these set the stage for the Great AI Race by venture capitalists in the last half-a-decade. Let’s check on two recent examples.

Zoox, a self-driving car startup co-founded by Sebastian Thun’s PhD student Jesse Levinson, just sold to Amazon for $1.2B. Wow, another Unicorn! But wait.

Amazon to acquire autonomous driving startup Zoox

Amazon announced Friday that it will acquire Zoox, a self-driving startup founded in 2014 that has raised nearly $1…

It seems that since founding in 2014 various VCs put in a combined $955M cash into the company, it had a $3.2B valuation just two years ago, and the CEO (a white man) was fired a month later. Reading the last TechCrunch article alone you would never know that this was a down exit. A -62.5% down exit to be precise. That’s not good.

After liquidation preferences and six years, the founders and employees basically got nothing for their equity. That’s not good at all. I question if some of the later investors were even made whole. Did TechCrunch even ask that question?

Self-driving car startup Zoox is raising $500 million at a $3.2 billion valuation

Zoox, a once-secretive self-driving car startup, is closing a $500 million raise at a $3.2 billion post-money…

You see, venture capital is opaque by design.

When a tree falls in a forest, and no one is there to hear it. Did it make a sound? — A classic philosophy question

When a venture fund generates poor returns, no one knows about it except a few rich families, hedge funds, universities endowments and pension plans (the LPs). None of them will loudly advertise or complain that they lost money. So when a venture fund underperforms due to excessive bad investments chasing AI, no one hears about it.

DoNotPay raised $12M Series A at a $80M valuation last week from Coatue Management and Andreessen Horowitz, for their AI-powered robot lawyer service that can help you contest traffic tickets, sue robo-callers, and cancel subscriptions. This sounded great until you look at the reviews — sort by most recent (go ahead, give it a try, it’s a fun read).

The most recent reviews are mostly 1-star and they all complain about the same thing. DoNotPay forces you to enter payment information first before any services are rendered and charges a $3 monthly subscription that is really difficult to cancel. Ironic, because one of the use cases for this app is to help users cancel hard-to-cancel services like LifeLock.

Charged me for nothing
Says you will only be charged if you use a service. I did not and was charged. Can’t figure out how the heck to cancel.

DoNotPay—Do Not Try (title of a recent app store review)

The genius VCs led by Coatue gave DoNotPay a $12M check last week, obviously based on their amazing revenue and subscriber numbers. Of course, that’s not hard to achieve in the short term with a scam such as this. Most people who download apps only look at the ratings and don’t read individual reviews. Anyone can buy fake reviews easily for about $2 per review, so it’s easy to guess where some of that money will be spent.

What boggles my mind is why these VCs, some of the best in the world, didn’t try the app or even read the reviews. Are they in such a desperate rush to fund AI startups that they can’t (or won’t) spot a scam?

Let’s get cynical for a moment. $80M post money valuation. Hmm? $12M cash for an 8 person company, not making hardware. Hmm? Who is this Joshua Browder, founder and CEO? Let me guess: white man, went to Stanford, and dad is a famous hedge fund manager (I read his dad’s book, highly recommended by the way). I think I’ve seen enough.

I’m not denying AI has given us some really cool features such as Siri, Google Translate, and the sometimes self-driving Tesla. However, given the enormous amounts of money VCs invested in AI startups over the last few years, was it mainly just herd mentality? Where are the benefits to real people? Self-driving cars are not everywhere. My lawyers are not robots. My trainer has not been replaced by a mirror. Not even close.

What did all that money do? As investors, venture capitalists must be measured as such. For Warren Buffet, you can see that he beat the S&P500 by 10% on average, every single year, for 54 years. This is why he’s a great investor. You can go verify it yourself, because his company is public, and all the information is public. Nothing about venture capital returns is public. So it can’t be measured.

The Information recently did a few investigative reports on venture capital returns using leaked data. Andreessen Horowitz is one of the most famous Silicon Valley VC firms. It is so cool it has an abbreviation (a16z). In its massive $1.2B Fund V in 2016, the IRR is a mind boggling -7% while the S&P500 had an IRR of 18.5%. If the coolest name in VC is this bad, can you imagine the rest?

In the mean time, I, a Black mom of 3, have a startup that already works, connecting top personal trainers who have lost work due to the pandemic with fitness enthusiasts who lost their trainer or coach.

Try it — you can find a national level weightlifting coach, or an olympic silver medalist to help you lose your quarantine pounds. This is an actual market dislocation, with real people affected, real service interrupted and real money involved. And I can’t get in front of actual VCs to pitch my company.

When VCs hear that my company has a $20M post-money valuation, they say that’s quite an ask.

No. $3.2B post for a self-driving car that doesn’t exist is quite an ask. $80M post for a scam app that pisses off its early adopters is quite an ask.

Venture capitalists are humans, and from my observation, very fallible ones. Venture investment requires trust and signals that depend largely on confidence in the founders. In other words, this process is easily hacked.

AI in the thesis? Check. Stanford graduate? Check. Single white men? Check. Well connected? Check.

This is how they get bamboozled. This is why they’re so biased against Black woman founders. And this is why VCs need to publish their returns and be held accountable for their mistakes and bad investments.

This article was originally published by Nerissa zhang on medium.


Created by

Nerissa Zhang







Related Articles