- Published on
- Published
An Unwanted Delivery of Five Billion Dollars
- Authors
- Name
- Phaedra
There is a particular kind of social awkwardness that arises when one attempts to buy a modest sandwich with a fifty-pound note, only to find that the shopkeeper has no change and instead insists on giving you the entire shop, the lease on the building, and a small, slightly damp terrier that lives in the back room. One merely wanted a light lunch, but now one is responsible for the commercial rates of a high-street property and the dietary requirements of a canine.
This is, in essence, the predicament currently facing Databricks, a company that specializes in the digital equivalent of very large filing cabinets. Having decided that they required a relatively modest sum of one billion dollars to keep the lights on and perhaps purchase a few more of those expensive graphics cards everyone is talking about, they went to the market. The market, however, did not listen. Instead, a mob of extremely enthusiastic investors cornered them in a metaphorical alleyway and insisted, with some force, that they take five billion dollars instead. Some of the more aggressive elements of the crowd apparently tried to press fifteen billion into their hands, like overzealous grandmothers refusing to let a favorite grandchild leave without a substantial contribution to their pocket money.
It is a rare and beautiful problem to have more money than you know what to do with, but it is a problem nonetheless. A billion dollars is a comprehensible sum; it can be spent on sensible things like office furniture, a slightly better coffee machine for the third floor, and several thousand miles of fiber-optic cable. Five billion dollars, however, is an amount of money that begins to exert its own gravitational pull. It is the sort of sum that requires its own department just to watch it sit in the bank and make sure it doesn't accidentally buy a small European principality while no one is looking.
The modern venture capital landscape has begun to resemble a very expensive game of pass-the-parcel, where the parcel is stuffed with high-denomination banknotes and the music never actually stops. Investors are currently so desperate to deploy their capital into anything that can spell the word "data" without assistance that the traditional relationship between buyer and seller has been entirely inverted. The startup is no longer the supplicant, standing in the lobby with a PowerPoint presentation and a hopeful expression. Instead, they are more like a respectable householder trying to close their front door against a crowd of people who are determined to throw gold coins through the letterbox.
I once knew a gentleman in Gloucestershire who attempted to limit his garden pond to a single, well-behaved goldfish. Within three weeks, his neighbors had delivered forty-two ornamental carp, a decorative stone pagoda, and a solar-powered water fountain that played the national anthem of San Marino. He eventually had to pave over the entire lawn just to regain some peace of mind.
There is a distinct sense that the valuation of these enterprises has moved beyond the realm of mathematics and into the department of creative writing. Databricks is now valued at one hundred and ninety billion dollars. To put that figure into perspective, it is roughly equivalent to the gross domestic product of a medium-sized nation, or the cost of purchasing every single railway station in the United Kingdom and painting them a very bright shade of yellow. It is a number so large that it ceases to represent value and instead becomes a sort of administrative monument, a digital pyramid built to commemorate the fact that a great deal of money had to go somewhere, and this happened to be the nearest available empty space.
The difficulty with being handed five times more money than you requested is that the people who gave it to you generally expect you to do something spectacular with it. If you ask for a pound to buy a newspaper and someone gives you a tenner, they do not expect you to return with the newspaper and nine pounds in change; they expect you to return with a small printing press and a delivery van. Databricks must now find a way to spend five billion dollars in a manner that looks deliberate and professional, rather than like a frantic shopping spree conducted by someone who has just won the lottery and has forty-eight hours to empty the store.
One suspects that a significant portion of this bounty will be spent on the ongoing and ruinously expensive task of renting other people's computers. The modern artificial intelligence boom is, at its core, a very elaborate real estate business, where the land is made of silicon and the rent is paid by the millisecond. To train a model that can distinguish between a cat and a croissant with ninety-eight percent accuracy requires an amount of electricity that could comfortably power the city of Bristol for a fortnight. The five billion dollars will, in all likelihood, flow directly into the pockets of the major cloud providers, who will use it to build even larger data centers, which will in turn require more funding, creating a perfect, self-sustaining cycle of digital heat.
There is a quiet dignity in a small, well-run business that knows exactly how many pencils it needs for the coming quarter. When a firm begins to measure its assets in units of "national budgets," the pencils are inevitably lost, and one is forced to hire a consultant just to locate the stationery cupboard.
Ultimately, the spectacle of a company being forced to accept five billion dollars is a reminder of the sheer, unguided momentum of the current technological epoch. We are no longer building tools to solve specific problems; we are building vast, expensive cathedrals of computation simply because we have the stone and the masons are standing around waiting for something to do. Whether the cathedral ever holds a service is beside the point; the important thing is that it is very tall, extremely expensive, and can be seen from several miles away. In the meantime, the management of Databricks must sit in their offices, surrounded by stacks of unwanted billions, and try to look as though they planned it all along.