Korva is a scroll of the internet’s best ideas.
Bet on human desire for more
Early predictions
In the early days of computing, people said there was a market for five computers in the world, and that no one needed more than a given amount of RAM.
AI's open demand
AI follows the same shape. Models were on an exponential of improvement, and demand for AI at a high enough level and a low enough price was basically uncaptured, a rare new kind of commodity for the world.
Efficiency and demand
Efficiency does not change that. Algorithms get more efficient and models get better, but the work comes down to turning electricity into useful intelligence, and with demand like this, there will always be a want for more of it, however good the other layer gets.
Why does cheaper power increase total use?
In 1865 economist William Stanley Jevons noted that improvements in steam engine efficiency in Britain were accompanied by rising coal consumption. Cheaper power made steam worthwhile for more uses, which more than offset the savings per engine. Economists now call such rebounds the Jevons paradox.
How much energy does computing need?
Physicist Rolf Landauer of IBM argued in 1961 that erasing a single bit of information must release a minimum amount of heat, equal to Boltzmann's constant times the temperature times the natural logarithm of two. Real chips use many orders of magnitude more energy per operation than that limit. All of the electricity a chip consumes ends up as heat, which is why data centers need cooling.
How did computers go from rooms to pockets?
ENIAC, completed in 1945, weighed about 30 tons and filled a large room. The transistor, invented at Bell Labs in 1947, and the integrated circuit, demonstrated in 1958 by Jack Kilby at Texas Instruments, replaced bulky vacuum tubes. Each step cut the cost per calculation, and new uses followed. A modern smartphone holds billions of transistors.
More from Invest Like the Best with Patrick O'Shaughnessy
