Fifty Grams of Nothing

Here's a detail almost everyone misses. Russell Seitz, a physicist at Harvard, calculated back in two thousand six that the internet weighs about fifty grams. That's two or three strawberries. But what exactly did he weigh? Not the servers, not the cables, not the hard drives. Just the electrons in motion — the flow of charge racing through every server circuit on the planet, every second. Switch the machine off, and those fifty grams vanish. It's a bit like saying the weight of a train is the weight of the air it pushes ahead of it.

Seitz's logic ran like this. A current of one ampere is a flow of roughly ten to the eighteenth electrons per second. Estimate the total power draw of every server on Earth, convert that to current, multiply by the mass of a single electron — and you land on those fifty grams. Seitz himself called it a playful answer to a child's question. The real physics is elsewhere: to keep that flow of electrons moving at all, humanity runs a power infrastructure equivalent to tens of millions of horsepower. The grams are nothing. The energy is everything.

A year later, Discover magazine came at it from the opposite direction — not energy, but information. They put total internet traffic at the time at roughly forty petabytes and asked: how many electrons does it take to store all those bits in memory? In their model, a "one" bit costs about forty thousand electrons, while a "zero" bit weighs almost nothing. The total came out to about five micrograms. Not a strawberry — a single drop of strawberry juice. Both figures are correct arithmetic. They're just measuring completely different things.

Christopher White, president of the NEC research lab and a Bell Labs veteran, put it bluntly: fifty grams is "just the wrong number." His objection wasn't the math — it was the question itself. Today's server farms, built for streaming and giant language models, bear no resemblance to what Seitz was counting in two thousand six. Physicist Daniel Whiteson compared the whole approach to pricing a single doughnut by dividing up the world's GDP: you get a number, but not an answer.

White proposed a third approach. He started from Landauer's principle — a hard physical limit: erasing one bit of information at room temperature always releases a minimum, unavoidable puff of energy. White put the internet's total data at about one point six five times ten to the twenty-fourth bits, multiplied by that energy cost, and ran the result through Einstein's equation to get an equivalent mass. The answer: about five point zero three times ten to the minus fourteenth grams. Fifty-three quadrillionths of a gram. White's own conclusion: under a rigorous physical model, the internet "weighs nothing at all."

But here's the twist. Store that same amount of data in strands of DNA — one gram of DNA can hold up to two hundred fifteen petabytes — and the whole internet would weigh in at about nine hundred sixty kilograms. Nearly a metric ton. Sixty-four thousand strawberries. The very same internet that "weighs nothing" as a stream of excited electrons becomes a solid, physical archive the moment you choose a different medium to hold it.

And that's exactly why there's no single right answer — there can't be. The word "internet" bundles together several entirely different things: protocols, traffic, server buildings, fiber-optic cable, the phone in your pocket. The mass depends entirely on what you decide to weigh. Seitz weighed motion. White weighed a thermodynamic footprint. Biotech researchers weighed a potential archive. All three are right — and none of them actually answers the question.

That's the real lesson here. "How much does the internet weigh" isn't a physics problem to solve. It's an invitation to agree, first, on what we even mean by information — and in what sense it exists as physical matter at all.

Ask me more about Landauer's principle sometime — there's a consequence to it that almost never comes up: a perfectly reversible computing machine would dissipate no heat at all in any calculation, and such a machine is theoretically possible. That single fact changes everything we think we know about the future of energy use in a digital world.

The world is enormous and one question is enough to start. moygid dot online.