tokens are going to be as cheap as electricity within the decade
227 comments
The author observes that a call to GPT-5.6 Luna is only 4-5 orders of magnitude more expensive than grep, and then predicts that at current rates of progress, calling an LLM will soon be cheaper than a grep. I think this is a good time to invoke Stein's Law: "If something cannot go on forever, it will stop." These efficiency improvements won't continue forever. It's more likely that the per-call cost of high-quality, compiled software like grep will be a lower-bound that LLMs asymptotically approach, rather than a line that they blow past with perpetual exponential progress. (Barring a true breakthrough in something like quantum computing or room-temperature superconductors.)
non-backreferencing grep is not very difficult to implement in an ASIC either. But it's probably not worth it because of how relatively rarely you use it and of the data transfer costs.
LLMs are great candidates for ASIC-burning because they're slow compared even to network speeds and run all the time. The issue is that you don't want to burn a specific model or architecture that then becomes obsolete.
So you've got two possible futures, and both guarantee large price drops: (a) LLMs keep getting better and better and better, so ability/$ keeps rising; or (b) LLMs plateau in ability, in which they will start getting ASIC'd.
you could make hardware implementation of grep and store the file itself next to it in some ROM but that's not a very useful grep ... while hardware LLM is exactly as useful as software LLM only orders of magnitude faster
When the author wrote Llm can be as cheap as a tool, I read it as not equivalent. They even said the Llm can be embedded into a tool.
Their point was, the higher level use case — like classification — could become as cheap as grep. Which is quite well possible.
I also don't understand where the idea that frontier models are getting better efficiency comes from. The results are certainly improving, but that comes from feedback and multiplexing requests, which cost more.
Something seems off about this.
At some future point where LLM hardware is cheaper than simply running grep, then grep equivalent would benefit from those selfsame hardware improvements and be cheaper to run as well, probably still by the same ratio.
The only aspect that is poorly analyzed by the OP is business model viability. All players are investing insane amounts of money in infrastructure with the expectation that their future profits will justify all that investment. The winner or winners in the AGI race, they believe, will find the proverbial "pot of gold at the end of the rainbow."
The OP glosses over questions of business model viability with a brief qualitative discussion and very little hard data. For example, to earn an annual return > 10% on every trillion dollars of capital sunk into infrastructure, the owners of that infrastructure must earn free cash flow (operating profit less investment) in excess of $100 billion per year in perpetuity. Is that feasible? Why? How?
The OP does not really consider such questions.
You may be right. I'm not so sure. Inference looks like a viable business model for those operators that have SOTA infrastructure in place, but the investment required to have it is enormous, and appears to be never-ending, because if an operator stops investing aggressively, its infrastructure quickly becomes non-competitive, and customers will quickly leave for alternatives. SOTA infrastructure is a moving target.
Only if you also have the model thats better than anyone else's.
As soon as models are free, or there are no newer models (assuming thats going to happen, and thats not a given) then the only thing you can compete on is price.
This means that the only thing you have to differentiate is either price, speed or ease of use. (or regulatory capture...)
We are at pets.com level of spend currently. Unless model development becomes cheaper, then we are going to run out of novel debt but not really debt mechanisms.
One is how much does it cost OpenAI to train the model.
The other is, if I stole OpenAI's model how much would it cost for me to run it?
R&D costs versus operational costs. Operational costs are very likely profitable. R&D is catastrophically expensive currently.
The hubris of this is really astounding too. There is no technology out there that some company develops and has not been reverse engineered and copied and manufactured at scale by competitors before long. You can't stop this from happening. People will leave the company or be poached and proliferate what they have built in the past. Every country that wanted a nuke has a nuke, after all.
The only way to keep the secret fire from leaking out would be to have AGI's first move be to lock the doors and prevent anyone from ever leaving company property again.
Also another take on AI costs falling https://x.com/EpochAIResearch/status/2102510281176023529 Methodology is how much it costs to do the same task a while later. Gives fallen ~47% a quarter.
"It is not too much to expect that our children will enjoy in their homes electrical energy too cheap to meter,..." Lewis Strauss
https://en.wikipedia.org/wiki/Too_cheap_to_meter#Origins
Oddly enough my power bill was metered and big.
Of course, perilous territory for future irony depending on how your prediction plays out.
Later, you were billed for time connected to the "internet" (compuserve or aol or whatever)
Around when the iphone came out, software went from tens or hundreds of dollars to pennies, then free.
On the other hand legal advice has always been expensive, because a good answer is worth it.
Medical advice is worth it. Investing advice is worth it.
(That said, I wonder if with home solar and batteries if electricity will ever "generally" go down in price to normal people)
[0] https://en.wikipedia.org/wiki/Zwentendorf_Nuclear_Power_Plan...
We had batteries full and were spending electricity on all kinds of luxury things like whole-day internet and desalination for several weeks, and now that it has rained for over a week we're starting to turn non-critical systems off to keep the lights on.
"Between digital economics and the wholesale embrace of King's Gillette's experiment in price shifting, we are entering an era when free will be seen as the norm, not an anomaly. How big a deal is that? Well, consider this analogy: In 1954, at the dawn of nuclear power, Lewis Strauss, head of the Atomic Energy Commission, promised that we were entering an age when electricity would be "too cheap to meter." Needless to say, that didn't happen, mostly because the risks of nuclear energy hugely increased its costs. But what if he'd been right? What if electricity had in fact become virtually free? The answer is that everything electricity touched—which is to say just about everything—would have been transformed. Rather than balance electricity against other energy sources, we'd use electricity for as many things as we could—we'd waste it, in fact, because it would be too cheap to worry about."
... What Mead understood is that a psychological switch should flip as things head toward zero. Even though they may never become entirely free, as the price drops there is great advantage to be had in treating them as if they were free. Not too cheap to meter, as Atomic Energy Commission chief Lewis Strauss said in a different context, but too cheap to matter. Indeed, the history of technological innovation has been marked by people spotting such price and performance trends and getting ahead of them."
My issue is that datacenter energy costs are being prioritized for commerce over residential use, so the average consumer is paying more for electricity, because a datacenter needs more electricity and they're getting tax breaks. Assuming this all improves efficiency for new products like automated robotics, there is a debatable benefit. Jevon's Paradox has no ceiling, except the environment, and people's 401ks.
The "most attractive quadrant" is completely meaningless. The whole point of a Pareto curve is that each point on the curve is better than everything else on at least one dimension, and that you can make these comparisons without placing a value judgement on the relative importance of the different metrics. If you make a composite score of the two metrics (any monotonically non-decreasing function, e.g. a weighted sum with non-negative weights), that score will always be maximized by one of the points on the Pareto frontier.
So going by the numbers in the 2nd chart (1st AA chart) from TFA alone:
- there's no reason one would choose Deepseek V4 Pro 0813 (max) even though it's in the "most attractive quadrant", because GLM-5.3-Flash is both cheaper and scores better.
- Claude Fable 5.1 (max with fallback) on the top right* could be your most attractive option if you need the best scoring model and don't care about cost, even though it isn't in the "most attractive quadrant"
- The un-shown model off the left side of the chart could be your most attractive option if you just need lots of cheap tokens and don't care about quality.
* I also dislike the way they place the labels, and that grey line connecting the label to the point is way too subtle.
"Had the atomic bomb turned out to be something as cheap and easily manufactured as a bicycle or an alarm clock, it might well have plunged us back into barbarism, but it might, on the other hand, have meant the end of national sovereignty and of the highly-centralised police State. If, as seems to be the case, it is a rare and costly object as difficult to produce as a battleship, it is likelier to put an end to large-scale wars at the cost of prolonging indefinitely a “peace that is no peace”."
It seems, especially with open weights, that the AI is much more like the alarm clock and not the battleship. $20/mo would have been about $1 in 1944
https://www.orwellfoundation.com/the-orwell-foundation/orwel...
Read the full thread on Hacker News →
Related stories
- Tokens Too Cheap to Meterjyn.devHacker News · 2 points · 8 days ago
- Tokens Too Cheap to Meterjyn.devHacker News · 1 points · 9 days ago
- tokens too cheap to meterjyn.devLobsters · 32 points · 9 days ago
- Tokens Written, Tokens Readhci.profHacker News · 2 points · about 1 hour ago
- Ars Technica · 0 points · 7 days ago
- Human TPS: How fast can your fingers generate tokens?homoagens.github.ioHacker News · 1 points · 3 days ago