Editor’s observe: I’m within the behavior of bookmarking on LinkedIn and X (and in precise books) issues I feel are insightful and attention-grabbing. What I’m not within the behavior of doing is ever revisiting these insightful, attention-grabbing bits of commentary and doing something with them that may profit anybody apart from myself. This probably recurring column is an effort to appropriate that.
Each single individual on the web, me included, has put out a tackle DeepSeek this week. The final path of journey was from, “This Chinese language startup simply beat the West at AI, and NVIDIA is cooked” to “As all of us have clearly recognized this entire time, there’s been an excessive amount of consideration on brute pressure compute and never sufficient on algorithmic optimizations” then lastly deciding on one thing like, “I’ve all the time recognized this precise factor DeepSeek did can be carried out, I simply didn’t do it myself or inform anybody about it.” However blended into the navel gazing had been just a few good concepts that, for no matter motive, all appeared to narrate to Jevons’ Paradox.
Briefly, Jevons’ Paradox means that as technological developments enhance the effectivity of a useful resource’s use, the overall consumption of that useful resource may very well enhance as an alternative of lower. This happens as a result of elevated effectivity lowers prices, which in flip drives higher demand for the useful resource. William Stanley Jevons put this concept out on the planet in an 1865 ebook that appeared on the relationship between coal consumption and effectivity of steam engine know-how. Extra fashionable examples are vitality effectivity and electrical energy use, gasoline effectivity and driving, and AI.
Which brings us to DeepSeek. I refuse to summarize what they did, how they did it, how a lot they stated it value, what it did to tech shares, and what each tech CEO needed to say about it. To the bookmarks!
Mustafa Suleyman, co-founder of DeepMind (later acquired by Google) and now CEO of Microsoft AI, wrote on X on Jan. 27: “We’re studying the identical lesson that the historical past of know-how has taught us time and again. The whole lot of worth will get cheaper and simpler to make use of, so it spreads far and broad. It’s one factor to say this, and one other to see it unfold at warp velocity and epic scale, week after week.”
Microsoft reported Q2 fiscal 12 months 2025 financials this week. Throughout the Q&A portion of the decision, CEO Satya Nadella additionally touched on this. “When token costs fall, inference computing costs fall, which means individuals can eat extra.” Extra on Microsoft’s Q2 right here.
Now to AI luminary Andrew Ng, contemporary off an attention-grabbing AGI panel on the World Financial Discussion board’s annual assembly in Davos. Posting Jan. 30, on LinkedIn, he posited that the DeepSeek of all of it “crystallized, for many individuals, just a few vital traits which have been taking place in plain sight: (i) China is catching as much as the U.S. in generative AI, with implications for the AI provide chain. (ii) Open weight fashions are commoditizing the foundation-model layer, which creates alternatives for utility builders. (iii) Scaling up isn’t the one path to AI progress. Regardless of the huge concentrate on and hype round processing energy, algorithmic improvements are quickly pushing down coaching prices.”
Extra on open weight, versus closed or proprietary, fashions. Ng commented that, “Quite a few US firms have pushed for regulation to stifle open supply by hyping up hypothetical AI risks resembling human extinction.” This very, very a lot got here up on that WEF panel. “It’s now clear that open supply/open weight fashions are a key a part of the AI provide chain: many firms will use them.” If the US doesn’t come round, “China will come to dominate this a part of the availability chain and lots of companies will find yourself utilizing fashions that mirror China’s values far more than America’s.”
Ng wrote that OpenAI’s o1 prices $60 per million output tokens, whereas DeepSeek’s R1 prices $2.19 per million output tokens. “Open weight fashions are commiditizng the foundation-model layer…LLM token costs have been falling quickly, and open weights have contributed to this development and given builders extra alternative.”
Now to Pat Gelsinger, the previous CEO of Intel and VMware, posting to LinkedIn on Jan. 27. The DeepSeek dialogue “misses three vital classes that we realized within the final 5 a long time of computing,” he wrote.
- “Computing obeys the gasoline legislation…It fills the out there house as outlined by out there assets (capital, energy, thermal budgets, [etc…]…Making compute out there at radically lower cost factors will drive an explosive enlargement, not contraction, of the market.”
- “Engineering is about constraints.”
- “Open wins…we actually need, nay want, AI analysis to extend its openness…AI is far too vital for our future to permit a closed ecosystem to ever emerge because the one and solely on this house.”
And our closing bookmark is from a ebook, Insull, a biography of Thomas Edison’s right-hand man, Samuel Insull, by historian Forrest McDonald. Summarizing Insull’s strategy to creating electrical energy a mass-market product, McDonald described the largely forgotten titan’s philosophy: “Promote merchandise as cheaply as attainable—not as a result of worth competitors dictated it; removed from it. Slightly, it stemmed from Insull’s radical perception, which Edison often shared, that decrease costs would convey higher quantity, which might decrease unit prices of manufacturing and thus yield higher income.”
That’s it for now. Comfortable studying.
