Theme 1 – Hardware cost trends and the prospect of local LLMs
Many commenters discuss whether falling RAM, GPU, and ASIC prices will allow frontier‑quality models to run on commodity hardware in the near term.
“We are likely to see LLMs running locally at current frontier‑quality on commodity hardware in the next 3‑6 years” – automatic6131
“RAM prices will crash when demand drops even a little. They’ll probably crash to a lower (inflation‑adjusted) level than before.” – api
“Graph the average compute and RAM in a mid‑high end laptop at an inflation adjusted price point for the past 40 years. It’s very exponential and hasn’t slowed down much.” – api
Theme 2 – Jevon’s paradox and demand elasticity
The discussion frequently turns to whether efficiency gains in LLMs will simply spur more usage (Jevon’s paradox) or hit natural limits.
“Jevon's paradox says that if data centers can serve a lot more tokens per dollar or watt there will be increased demand for data centers.” – bryanlarsen
“Jevon's paradox isn't a physical law, it doesn't magically apply to everything… Thousands of miles of canals were dug in the UK that couldn't be sustained and were abandoned.” – automatic6131
“It’s true that Jevon's paradox doesn't always apply, although this does seem like a classic case.” – bryanlarsen
Theme 3 – Viability of AI business models and profitability
Commenters debate whether inference can be a profitable business given massive capex, subsidies, and uncertain returns.
“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 … in excess of $100 billion per year in perpetuity. Is that feasible?” – cs702
“I think it's clear that inference is a viable business model… But what isn't clear is whether it will be such a profitable business model for any given company that it will justify the investment that company has taken.” – sanderjd
“The labs themselves when they openly say that they’re subsidizing tokens…” – ofjcihen
Theme 4 – Historical analogies and limits of extrapolation
Many invoke past technological promises (“too cheap to meter,” Moore’s law, grep cost comparisons) to caution against overly optimistic extrapolations.
“Too Cheap to Meter reminds me of the promise of Nuclear Power in 1954… ‘It is not too much to expect that our children will enjoy … electrical energy too cheap to meter…’” – abirch
“Stein's Law: 'If something cannot go on forever, it will stop.’ These efficiency improvements won't continue forever.” – jetrink
“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.” – jetrink (and subsequent rebuttals about limits).