There have been several media articles the past few days scoffing at the $500M annual loss that Deepmind is making for its owner Google. Is that a lot of money? Yes, it is. Is it actually expensive? Well, that depends on what you compare it to.
So for those of you who don’t know, Deepmind are effectively the Yankees of A.I. research. The best of the best. Most trophies, all that. So yes, the total salary of almost $500M divided by just 700 employees seems excessive. But these aren’t developers. These are the best researchers in the world, in the hottest field right now. So what does an A-team look like?
So yes, expensive when compared to tech companies like Apple. Same ballpark as top banks like Goldman Sachs, but remember they have 36,000 employees! Deepmind pays a high salary for a low number of specialists, most of who have a Ph.D. If you took top executives at Apple and Goldman Sachs, we’re talking seven figures easy. And what about them Yankees? Yeah, forget about it. It’s a lot of money, but not for an A-team.
Another way to look at the value Deepmind brings is their brand. They are one of a handful of labs in the world with a singular focus on developing General Artificial Intelligence. They believe if they can solve intelligence, then they can solve any problem. Cancer. Quantum Gravity. Any problem. Big statement.
As part of that journey, they have tackled some of the hardest problems humans have developed, like playing Chess and Go. Deepmind beat the humans, all of them, and is now developing even harder problems like multiplayer online games. Deepmind’s AlphaGo made the headlines globally when they beat the world’s top-ranked player Lee Sedol back in 2016.
So what other reference points do we have for such high-brow, high-budget marquee projects?
These are literally the biggest undertakings of mankind to date. So it’s interesting to see that $500M a year definitely registers on the scale. It’s big-spending for sure. The LHC project is considered one of the most expensive science projects ever and has uncovered the Higgs Boson, making it one expensive particle. But after adjusting for inflation, the 20th-century big-ticket items like building the nuclear bomb and going to the moon take the cake. Plus, it’s worth noting that most of these projects spent this budget over a decade. Deepmind is just a few years in, although also growing it’s spending steadily… but what is it worth to own the leading A.I. research lab in the world right now? A lot. Probably more than $500M if you asked for bids.
So if we look at other similar companies that have come up in recent years, is this really a crazy amount in the world of unicorns? In the race to escape velocity and market monopoly, we’ve seen startups raising bigger and bigger rounds to outgrow and outspend each other.
Yeah, not really. Deepmind isn’t even in the race. But there’s a significant difference to point out. These other guys are spending like it’s someone else’s money, which it is actually, because they’re trying to build a moat through money. Becoming so big it’s impossible to compete. To even think of competing would require a massive war chest.
Deepmind is real deep tech. They’re building all kinds of crazy I.P. that while most of it gets shared through public research, Google has first dibs to commercialize. Many are rightly concerned about the long-term sustainability of the cash-burning unicorns, but I wouldn’t put Deepmind in that group. The spending is at least producing something.
So finally, if we back up to how much Google actually paid for Deepmind. It matters in the sense of how deep the hole is, to begin with. If you overpay massively, you better see ROI big time. Makes sense?
Dirt cheap, is the answer. Google paid more than double that for Youtube. Now, what these other companies did have was millions of users. But that’s a different play. Youtube has worked out pretty well for Google. Deepmind is an investment in the future, not the next quarter, and Google can afford it. They know someday the ad teet will run out of its golden milk, so they better diversify while they can. Hence, Alphabet.
Google’s competitors would pay multiples of that half-bil to take Deepmind off their chest if Google wanted to ship it. Not gonna happen. Recently, Microsoft paid a cool billion for a minority stake in the other big A.I. lab, OpenAI. If anything, Google’s investment has already had a killer return. Having the best A.I. research team in the world, and by far the strongest brand in A.I. across the industry, is simply priceless.
So in conclusion, it’s not expensive at all.