Five games, 9–15 March 2016, won 4–1 by the program. Lee Sedol took game four, which remains the only game a human has won from AlphaGo in a formal match.
These two lists are the reason the page exists. A match is a result about a particular series of games under particular conditions, and almost everything usually drawn from one does not follow from it.
What stood behind the Seoul match was 48 first-generation tensor processing units spread across several machines. That is DeepMind's own statement, but it appeared only eighteen months after the match, in the AlphaGo Zero paper: 'AlphaGo Lee is the program that defeated Lee Sedol 4–1 in March 2016… This player was also distributed over many machines using 48 TPUs, rather than GPUs.' The much-quoted '1,920 CPUs and 280 GPUs' has nothing to do with this match: it is the largest row of the scalability table in the 2016 paper (Elo 3168), while the version that played Fan Hui in October 2015 was distributed over 1,202 CPUs and 176 GPUs. The confusion traces to DeepMind's pre-match line that AlphaGo would use 'about the same computing power' as against Fan Hui. How many machines and general-purpose processors served those 48 TPUs was never published. The hardware itself ran on Google Cloud in the United States; what sat in the Seoul room was Aja Huang, placing the stones.
Sources: D. Silver et al., 2017; D. Silver et al., 2016; Wikipedia, 2026Every line here states the boundary it was measured inside, and the boundary is printed above the number rather than under it: eight hundred watts of silicon and eight hundred watts at the wall are not one figure at two precisions, they are two different figures. Training and playing are kept apart for the same reason and are never added: one was spent once, before the match; the other over the hours people sat at the table.
Reconstructed from the configuration; nobody metered the match. 48 first-generation TPUs × 40 W. The forty watts is the measured busy power of a TPU die from Table 2 of Jouppi et al. (ISCA, 2017); the same table gives a 75 W nameplate TDP and 28 W at idle. That paper is directly on point: one of the convolutional networks in its benchmark set is AlphaGo's. Using TDP instead, the same multiplication gives 3,600 W. Accelerators only: the machines driving them are not counted, because their number was never made public.
Reconstructed from the configuration; nobody metered the match, and this is a floor. Jouppi et al. give a four-TPU server a measured busy draw of 384 W and a nameplate TDP of 861 W. Forty-eight TPUs make twelve such servers: from 12 × 384 ≈ 4.6 kW to 12 × 861 ≈ 10.3 kW. A range rather than a number, because the low end is a measurement under one particular load in Google's datacentre and the high end is a nameplate ceiling. Both are floors for the whole picture: DeepMind never published how many general-purpose machines served AlphaGo Lee's search, so none of their watts are in here.
There is no figure. The famous contrast — 'AlphaGo a megawatt, Lee Sedol twenty watts' — goes back to a blog post Jacques Mattheij wrote during the match: he took the 1,920 CPUs and 280 GPUs from the scalability table of the 2016 paper, allowed 200 W apiece, arrived at roughly a megawatt and concluded 'fifty thousand times as much'. He gives no source for the 200 W, and more to the point that configuration never played Lee Sedol: Seoul was 48 TPUs. The megawatt is an estimate of the wrong machine. DeepMind has never published a measurement of its own.
A separate quantity, and not a substitute for the match one. Training AlphaGo Lee took months on an incomparably larger fleet — the 2016 paper mentions 50 GPUs for network training — and the energy involved is orders of magnitude beyond five games. Neither DeepMind nor Google published a consumption figure for training, so this stays empty. Merging training with the match is the standard methodological error in this subject: different scope, different order of magnitude.
Prices are held in the money of their own year and inside a stated boundary: the match hardware, the whole project and rented computation are three different sums, and which one is on offer is written above each.
There is no invoice. The hardware was Google's and ran in Google's own cloud; no rental price and no internal rate for 48 first-generation TPUs was ever published — the first-generation TPU was never sold or rented to anyone outside.
Sources: N. P. Jouppi et al., 2017; D. Silver et al., 2017What AlphaGo cost to build was never disclosed. Google's 2014 purchase price for DeepMind does not belong here: that is the price of a company, two years before the match, and it says nothing about this match's hardware or labour. The firm money is the $1 million purse, which DeepMind gave to charity after winning, and Lee Sedol's $170,000 — $150,000 for playing and $20,000 for the game he won.
Sources: Wikipedia, 2026two hours of main time each, then three byo-yomi periods of 60 seconds
The time control is the match regulation. No average time per move for either the program or Lee Sedol appears in the sources checked, so both lines are empty. The search rate is empty on purpose too: the 2016 paper offers only a comparison drawn for the Fan Hui match — 'AlphaGo evaluated thousands of times fewer positions than Deep Blue did in its chess match against Kasparov' — and no positions-per-second figure was published for the 48-TPU Seoul configuration. Comparing AlphaGo with Deep Blue by positions per second is misleading in any case: one counts alpha-beta nodes, the other Monte Carlo simulations with two neural networks inside each.
Sources: Wikipedia, 2026; D. Silver et al., 2016On any chart the human side needs at least two lines: the brain (20 W) and the person at the board (105–126 W). A single line against the machine's watts is not a comparison, it is a choice of answer. And neither line is truly commensurable with the machine's: the machine's number excludes its training, and the human's excludes the twenty years it took to learn the game.
The adult brain at rest. A physiological floor, not the price of a single move.
The twenty watts is a conversion, not a meter reading, and two independent routes arrive at it. One: Raichle and Gusnard (PNAS, 2002) give the brain as roughly 2% of body weight and roughly 20% of the body's oxygen and calorie consumption; 20% of an adult's basal rate (see body_resting, 60–90 W) is 12–18 W, and 20% of a seated person's rate (100–125 W) is 20–25 W. The other route is more direct: the classic Kety–Schmidt measurements put the brain's oxygen consumption at 3.3–4.0 ml per 100 g per minute (Sokoloff et al., 1955). For a 1,400 g brain that is 46–56 ml of oxygen a minute, and at an energy equivalent of about 20.9 J per ml it comes to 16–20 W. Both routes land on 16–21 W, and 20 is the rounding up. This does not earn an M: oxygen was measured, watts were not.
The whole body at rest — basal metabolic rate. An adult lying down, fasted, thermally comfortable.
Converted from kilocalories per day. One kcal/day = 4,184 J / 86,400 s = 0.04843 W. The Mifflin–St Jeor equation (1990), now the working standard in place of Harris–Benedict: a man of 70 kg, 175 cm, 40 years gives 1,599 kcal/day, which is 77 W; a man of 80 kg, 180 cm, 30 years gives 1,780 kcal/day, 86 W; a woman of 60 kg, 162 cm, 40 years gives 1,252 kcal/day, 61 W. Hence 60–90 W, covering ordinary adults of either sex. This is lying down and fasted; a person sitting at a board spends more — see the next row.
A person sitting at a board. If watts are going to be set against watts, this is the honest figure to use.
From the metabolic heat production tables underlying the thermal comfort standards ISO 8996 and ASHRAE 55. The 'met' is defined as 58.2 W per square metre of body surface and describes a person seated at rest in thermal neutrality. The standard adult is 70 kg, 175 cm, with a DuBois surface area of 1.8 m². That gives: seated quiet, 1.0 met — 105 W; seated at office work, 1.2 met — 126 W. This is the body's total heat output, not the brain's share. A caveat: the ISO and ASHRAE tables descend from Passmore and Durnin's 1967 review and have not been revisited since. The spread among advisers — '70–100' against '100–120' — is not a disagreement but two different definitions: basal rate in the first case, a seated person in the second. '150–300 W' for someone sitting is simply wrong: on the same tables 150–210 W is standing and light physical work, and 300 W is walking.
The increment from hard mental work. There is no reliable figure, and that is itself the finding: the increment is small enough to be hard to measure at all.
The most direct test was done seventy years ago and has not been overturned. Sokoloff, Mangold, Wechsler, Kennedy and Kety (Journal of Clinical Investigation, 1955) measured cerebral blood flow and oxygen consumption in 13 people at rest and while doing mental arithmetic. The EEG changed unmistakably during the arithmetic — a shift to low-voltage, high-frequency activity — and pulse and blood pressure rose, while the brain's oxygen consumption did not move: in their words, the control and arithmetic values were identical. The authors state the conclusion worth remembering the paper for: neither the degree of mental activity nor the EEG pattern correlates with total cerebral oxygen consumption. Modern imaging did not overturn this so much as sharpen it. Locally, blood flow in a working region rises a great deal; across the whole brain, the task-evoked increase is usually under 5% of the ongoing baseline (Raichle and Mintun, Annual Review of Neuroscience, 2006). Five per cent of 20 W is under a watt, and that is a ceiling rather than a reading. What is not here and will not be: the claim that a grandmaster burns 6,000 kilocalories a day. It comes from an ESPN piece of 2019 quoting Robert Sapolsky, and traces back to a tripled breathing rate multiplied by 2,000 kcal. No calorimeter appears anywhere in that chain. What does rise during a hard game is sympathetic arousal — heart rate, blood pressure, breathing, muscle tone. That raises the whole body's expenditure, not the brain's, and how much it raises it over a chess match has never been published.
These four lines are the same on all nine pages, because the human side of the table was the same in all nine matches. Twenty watts is the brain and not the person: what sat at the board was a whole body, and setting an organ against a machine's wall socket is a way of choosing the answer in advance.
The match was played in Seoul, at the Four Seasons, from 9 to 15 March 2016: five games, a million-dollar purse. AlphaGo was two neural networks — one proposing moves, one judging positions — wrapped around a Monte Carlo tree search. The Nature paper notes that against Fan Hui it evaluated thousands of times fewer positions than Deep Blue had against Kasparov, winning on selection rather than volume. The distributed version ran on roughly 1,920 CPUs and 280 GPUs, and Google said tensor processing units were used in Seoul as well. Lee Sedol, holder of eighteen world titles, publicly expected 5-0 or 4-1 his way.
D. Silver et al., Mastering the game of Go with deep neural networks and tree search, Nature 529(7587), 2016; distributed configuration and purse from the match record; the TPU statement is Google's own.Go was supposed to be the reserve, good for another decade. Some ten to the power of a hundred and seventy positions, brute force useless, and no one knew how to write an evaluation function for it: a strong player judges a position by a method he cannot put into words. In October 2015 AlphaGo had beaten Fan Hui, the European champion, 5-0 — but the gap between a European champion and Lee Sedol was wider than the gap between Fan Hui and an amateur. Seoul was the first real test, and few people were backing the program.
The position count is the usual published estimate (see the Solved Games page on this site); the Fan Hui match is reported in Nature 529(7587), 2016.Two of the games mirror each other. In game two AlphaGo played its thirty-seventh move as a shoulder hit on the fifth line, a move humans do not play; its own policy network put the chance of a human choosing it at about one in ten thousand. Lee Sedol left the room. In game four he found move 78, a wedge that Korea named the divine touch; AlphaGo had given it the same probability, one in ten thousand, and after it the program's evaluation of the position collapsed. It played several bad moves in a row and resigned. The machine played what people do not. The person played what the machine had not allowed for. Final score 4-1.
The one-in-ten-thousand figures for both moves were given by the DeepMind team and appear in the documentary AlphaGo (dir. Greg Kohs, 2017); score and move numbers from the game records.Game four is still the last human win over a top Go program on even terms. In November 2019 Lee Sedol retired from professional Go, telling the Yonhap news agency that even becoming number one would not make him the best, because "there is an entity that cannot be defeated". Move 37 is now teaching material, and the fifth-line shoulder hit has entered ordinary play. Professional Go did not vanish; it rearranged itself around the programs as a source of positions people would never have reached on their own.
Lee Sedol's retirement statement: interview with the Yonhap news agency, November 2019; game four's standing as the last even-terms human win is noted on this site's timeline page.