Will I ever own a zettaflop?
geohot.github.io
geohot.github.io
> Some customers already have datasets on the order of a petabyte, or 2^50 bytes. Thus the 64-bit capacity limit of 2^64 bytes is only 14 doublings away. Moore's Law for storage predicts that capacity will continue to double every 9-12 months, which means we'll start to hit the 64-bit limit in about a decade. Storage systems tend to live for several decades, so it would be foolish to create a new one without anticipating the needs that will surely arise within its projected lifetime.
* https://web.archive.org/web/20061112032835/http://blogs.sun....
And some math on what that means 'physically':
> Although we'd all like Moore's Law to continue forever, quantum mechanics imposes some fundamental limits on the computation rate and information capacity of any physical device. In particular, it has been shown that 1 kilogram of matter confined to 1 liter of space can perform at most 10^51 operations per second on at most 10^31 bits of information [see Seth Lloyd, "Ultimate physical limits to computation." Nature 406, 1047-1054 (2000)]. A fully-populated 128-bit storage pool would contain 2^128 blocks = 2^137 bytes = 2^140 bits; therefore the minimum mass required to hold the bits would be (2^140 bits) / (10^31 bits/kg) = 136 billion kg.
> To operate at the 10^31 bits/kg limit, however, the entire mass of the computer must be in the form of pure energy. By E=mc^2, the rest energy of 136 billion kg is 1.2x10^28 J. The mass of the oceans is about 1.4x10^21 kg. It takes about 4,000 J to raise the temperature of 1 kg of water by 1 degree Celcius, and thus about 400,000 J to heat 1 kg of water from freezing to boiling. The latent heat of vaporization adds another 2 million J/kg. Thus the energy required to boil the oceans is about 2.4x10^6 J/kg 1.4x10^21 kg = 3.4x10^27 J. Thus, fully populating a 128-bit storage pool would, literally, require more energy than boiling the oceans.*
* Ibid.
If we instead consider a million 18TB hard drives, and estimate they each need 8 watts for 20 hours to fill up, 2^64 bytes take 160MWh to write on modern hardware. And they'll weigh 700 tons.
Edit: The quote is inconsistent about whether it wants to talk about bytes or blocks, so add or subtract a factor of about a thousand depending on what you want.
Or a third of a billion 24 TB drives, which is one of the larger sizes currently available.
Some random search results say the global hard drive market is around an eighth of a billion units, but of course much of that will be smaller sizes.
So that should be physically realizable today (well, with today's commercial technology), with only a few years of global production.
To be clear, the first quote was talking about 2^64 bytes, and you're talking about 2^64 blocks.
Edit: Though confusingly the second part talked about 2^128 blocks.
Also these days I'd assume 4KB blocks instead of 512 bytes.
That's 16 exabytes. Wikipedia cites a re:invent video to say that Amazon S3 has "100s of exabytes" in it.
So it not only could theoretically be done, but has been done.
For the record, 44TB drives have been announced in March 2026:
* https://www.seagate.com/ca/en/stories/articles/seagate-deliv...
I have a bunch of NVMe enclosures that generate 750W of heat when going full blast, and 600W when completely idle. How heat does the equivalent number of HDDs generate when working/idle?
Many petabytes fit in a single rack and many data sources generate several petabytes per day. I'm aware of sources that in aggregate store exabytes per day. Most of which gets promptly deleted because platforms that can efficiently analyze data at that scale are severely lacking.
I've never heard of anyone actually storing zettabytes but it isn't beyond the realm of possibility in the not too distant future.
With larger virtual memory addresses there is still the issue that the ratio between storage and physical memory in large systems would be so high that cache replacement algorithms don't work for most applications. You can switch to cache admission for locality at scale (strictly better at the limit albeit much more difficult to implement) but that is effectively segmenting the data model into chunks that won't get close to overflowing 64-bit addressing. 128-bit addresses would be convenient but a lot of space is saved by keeping it 64-bit.
Space considerations aside, 128-bit addresses would open up a lot of pointer tagging possibilities e.g. the security features you allude to.
No, I mean k8s style architecture, where you take physical boxes and slice them into smaller partitions, hence the dataset on each partition is smaller than the raw hardware capability. That reduces the pressure towards the limit.
Many CPUs, not just x86, have a "physical bits" length that is less than the address size in the ISA. This saves transistor and power budget, since address buses can be smaller. Of course, it means there is a lower maximum RAM config for that design.
The software would still shuffle around the full ISA word/double word/quad word or whatever. In a typical OS, the MMU and page mapping logic would potentially interpret all the bits to map to the more limited physical address range. It didn't mean storing smaller pointers in your software data structures etc.
I'm not an expert, but I think it varies by ISA whether it is defined how the higher address order bits, above the physical range, are handled. Some may allow applications to set them for address tagging, while the CPU ignores those bits. But, others may require them to all be zeroed to get predictable behavior.
I believe the Bekenstein bound for holographic information on a 1 liter sphere, using space at the Planck scale for encoding, instead of matter, is about 6.7×10^67.
I confess I got that number by taking round trips through multiple models to ensure there was a clear consensus, as my form of "homework", as this is not my area of expertise.
As far as figuring out energy or speed limits for operations over post-Einsteinian twisted space, that will require new physics, so I am just going to wait until I have a 1 liter Planckspace Neo and just measure the draw while it counts to a very big number for a second. (Parallel incrementing with aggregation obviously allowed.)
Point being, there is still a lot of room at the bottom.
Interesting thought. Space can expand faster than the speed of light over significant distances, without breaking the speed limit locally.
But what happens if complex living space begins absorbing all the essentially flat local space around it? Is there a speed limit to space absorption? If space itself is shrinking, due to post-Einsteinian structures/packing, then effective speed limits go away. As traversal distances, and perhaps even the meaning of distance, disappear. So, perhaps not. I call this the "AI Crunch" end-of-the-universe scenario.
That is the computer I want. And I believe that sets a new upper bound for AI maximalism.
They talk quite a bit about this sort of thing at the end...
Secondly, I recently tried to work out what year on the Top500 list[1] I could reasonably be for around US$5000. It's surprisingly difficult to work out mostly because they use 64 bit flops and few other systems quote that number.
It came out three years after Snow Crash, which already ironically referenced "The sky above the port was the color of television, tuned to a dead channel".
I agree that Neuromancer wasn't a great novel, though it obviously had vibes that resonated with many people. The novel being otherwise a bit of a dud actually speaks to how strong the vibes were to overcome that.
As with much from this thread of cyberpunk writing, the cities and world are the most important characters, and the storyline is just an excuse to wander through their streets.
Though about the world building: he threw out a lot of neologisms on the page, and later other writers gave them meaning.
https://news.ycombinator.com/item?id=45303483
Jeff Geerling made a $3000 raspberry pi cluster and shared the linpack scores, so I looked at when it’d hit different spots in the top500 list. He’d have won from ‘93 to June ‘96, and then been knocked out of the top 10 in November ‘97.
That’s with a pretty substantial constraint, making it out of raspberry pi’s, and a lower budget. With $5000, and your pick of chips… I bet you could hit the turn of the century…
Edit: There is a problem with that - 10MW * 8 billion is something like half the solar power from the sun which implies wasted heat will heat earth considerably.
* https://en.wikipedia.org/wiki/Metric_prefix
In 2025 we got robi— and quebi—:
What will it take to get this before you die? What are physical limitations to shrink things more and more and to speed things up more and more? He talks about solar, but what are the physical limits and how can we get there?
I think there's interesting physics here, but this sounds like just a rich guy craving more power.
Comma does some cool stuff, if relatively entry-level, and this post is good napkin-maths and was a fun read, but there is so much more depth and a hundred ways in which this post is wrong or over-simplified to the point of near irrelevance.
That someone isn't as intelligent as they think they are doesn't place an upper bound on their intelligence.
The GTA fix shows humility, literally the first sentence of the "recon" section is "First I wanted to check if someone had already solved this problem". Geohot wasn't interested in if anyone had tried and failed to "solve search", or why it might be a difficult problem. He assumed that Twitter were a bunch of idiots.
The whole approach of the GTA fix author is curious and humble. Very low ego.
I bring this up to present an alternate view of the future that a lot of thought has gone into: the Matrioshka Brain. This is basically a Dyson Swarm but the entire thing operates as one giant computer. Some of the heat from inner layers is captured by outer layers for greater efficiency. That's the Matrioshka part.
How much computing power would this be?
It's hard to say but estimates range from 10^40 to 10^50 FLOPS (eg [1]). At 10^45 FLOPS that would give each person on Earth access to roughly 100 trillion zettaflops.
[1]: https://www.reddit.com/r/IsaacArthur/comments/1nzbhxj/matrio...
You’d need self-replicating machines to build it, naturally. You’d need some ability for them to mine from asteroids and process the materials right there on the spot. And they’d need to be able to build both the processor “swarmlets” (probably some stamped-out solar/engine/CPU package) and more builders, so that the growth can be exponential. Oh, and the ability to turn solar energy into thrust somehow using only fuel you can get from the mined asteroids. Maybe a prerequisite is finding a solar system that has a huge and extremely uranium-rich asteroid belt.
You would need a CPU design that can be built using the kind of fidelity that a self-replicating machine in space under constant solar radiation can achieve. But if you can get the scale high enough, maybe you can just brute force it and make machines on the computational scale of a Pentium 3, but there’s 10^40 of them so who cares. Maybe there’s a novel way of designing a durable computing machine out of hydrocarbons we have yet to discover.
The machines would have to self replicate, and you’d need to store the instructions somewhere hardened. And that can be built out of materials commonly found in asteroids. Maybe hydrocarbons. Hell, may as well use RNA. These things need to be as good as humans at building stuff, so really this is just creating artificial “life” that self has DNA and is made of cells that build proteins needed to create the machine. Maybe they reproduce by spreading as little DNA seeds that can attach to an asteroid with the right chemistry, and we just spew them into the cosmos at a candidate star and hope the process gets kickstarted. Hell, we could make it spew its own DNA at the next stars over as soon as it’s done. We’d have a whole galaxy computing for us, all we’d need is the right DNA instructions, the right capsule for them, and a way to launch them.
Maybe another civilization has already done this…
You can't have "one giant computer" when the speed of light is a 16-minute ping time from one side of the swarm to the other. Also cooling. Space is a vacuum, you can't just use convection. The inner layers would melt before they could radiate it away.
Even maintenance and power distribution, you're talking about trillions of nodes that need active course correction to avoid a chain-reaction of collisions.
There's so many reasons this is not feasible and more of a whimsical thought experiment. I've barely even touched on most of the issues.
> One million Claudes. To be able to search every book in history, solve math problems, write novels, read every comment, watch every reel, iterate over and over on a piece of code until it’s perfect – spend a human year in 10 minutes. 50,000 people working for you, all aligned with you, all answering as one.
We are already near the limits of what we can do if we throw compute at Claude without improving the underlying models, and it is not clear how we can get big improvements on the underlying models at this point. Surely geohot knows this, so I am surprised he thinks that "one million Claudes" would be able to e.g. write a better novel than one hundred Claudes, or even one Claude.
Hard disagree. If I had a million Claudes worth of compute I'd be livestreaming my entire reality feed to a local server 24/7 and having it organize my observations and thoughts, synthesize new ideas, implement prototypes and discard infeasible ones while I sleep. If you're in the business of knowledge creation, a million Claudes isn't enough. Text is an easy modality, I want foundation models that operate on text, images, audio, video, streaming point clouds, ...
The results will be lackluster. Additional agents will not improve the result.
> "Heaven is exactly like where you are right now, but much, much better."
If a million Claudes of compute were accessible, people would not be doing the same things they are now, but more so. They'd be doing very different things we likely can't imagine - in the same way that Alan Turing imagined machines learning from experience, but didn't imagine downstream products like Sora, ad tech, or social media, or their cultural and economic effects.
Maybe a million is a stretch but thousands is completely doable right now. That's thousands of Claudes. Knock yourself out.
If it ever becomes possible to say "build me a unicorn" you're going to get millions of people trying to do the same thing, and you no longer have the same economy.
Because the features that generate unicorns stop being unusual.
Startup slop instead of art slop.
Which is the real problem with AI. Work gets cheap, original value stays expensive no matter how much compute you throw at it.
Because if it gets cheap too, it gets commoditised and stops being valuable.
And globally that applies to everything.
AI will either have to be tightly rationed, or it will murder the competitive economy.
You may as well make the same argument about commanding a billion Nintendo 64s. We cannot simply scale up Claude instances like you say, its ability to produce interesting output is bounded by the underlying model.
(Maybe that's why they call it Gas Town?)
This is why the focus on raw compute is a poor use of engineering time. We have plenty of it, we are just headed in the wrong direction.
I’d say there odd a bit of a flaw in the read 50,000 books part though. The LLM reading that much doesn’t really get you 50k books of value as a person. You’re the bottleneck not the flops
This is all just such crazy coincidence.
Everything is coming together so quickly.
There is an issue of the "non-uniformity of the spread of the future" though with fast development, and the faster the development the stronger the non-uniformity and the tensions it creates. Strong non-uniformity and resulting tensions have tendency to resolve catastrophically on their own at some point if not solved/smoothed by the other ways before.
The "simulation argument" to me is ridiculous.
Life started on earth shortly after the solar system formed. It's a quarter the age of the universe that it took for intelligence and civilization to arise. A long, long, looooong time.
From a numbers perspective, our minds are all shockingly rare. The universe probably doesn't produce many of us through stellar and then biological evolution.
Taking that into consideration, contrast that with the flip side.
If high technological simulation exists that becomes indistinguishable from reality, it could simulate quintillions of minds. It seems like we're on the path to that technology, with a great degree of probability.
Given that, and the fact that historical simulation is likely easy for the future, it seems more probable to me that we're one of the quintillions of simulations rather than the origin timeline.
I'm half joking, but I'm half not. It's a fun thought experiment with absolutely no basis in science.
Do you even really know if you were alive a second ago? Perhaps you were just instanced into existence with a set of memories - memories you are not randomly accessing. A very deliberate version of a Boltzmann Brain.
Again, pseudoscientific tomfoolery, but fun to ponder.
Most of the universe, by a large percentage (or so I'm told), is further away than 200 light years.
What signs of intelligence would we see or detect in our own solar system from a distance of 200 light years should we scan it with all our latest and best tech?
It's strange that life has existed on earth for much of the age of the solar system itself, yet intelligence only arises at the end of the habitable period of earth. There's not much time left.
* Did David give any answer to the question of what could be observed about intelligent life on Earth from a distance of 200 light years?
That was the question asked.
* Who out there is giving lectures in which they propose other intelligent life exists "out there" ?
> yet intelligence only arises at the end of the habitable period of earth.
Human habitable .. and some human descendants may very well survive AGW, if not there's likely time for non human intelligence to come along on the planet.