Yahoo refused to buy Google because it didn't believe search could be big
Yahoo and P&G made the same mistake: they didn't believe search would move to the front end, and they didn't believe Americans would sit on the sidewalk drinking Starbucks. When founders step back to the second line too early, the company loses the ability to take a bet.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
McKinsey sells half a century of database
Wang Huanan says McKinsey is essentially a "briefcase company" with no assets, yet consultants in their twenties can sketch out a client's hundred-year history in a month or two, because half a century of business accumulation stands behind them. In 1997 and 1998 he worked on a case for Ping An Insurance in Shenzhen, charging 1 million US dollars for three months — nearly 10 million RMB at the time — with two or three people doing it. Ma Mingzhe's reasoning was, "We want to aim at world-class companies, so we have to use a world-class consulting firm." This was the first time he realized that China's best private enterprises had started using world-class consulting services very early.
— Wang HuananWhen you know nothing, never read the literature
While working on Pringles at P&G, Wang Huanan had no understanding of how Americans ate snacks. His methodology was: "When you know nothing, never read the literature — go find users and listen." Four days out of five each week he was running focus groups across the US, three a day, seven or eight people each, reaching over two hundred users a month. After three months he could put himself in the shoes of a twenty-year-old American kid. The product judgments that came out of it were very concrete: chips can't shatter all over the back seat the moment you open the lid, and you can't end up with oily hands after eating them — hence the tube, the lower oil, and the flavor powder dusted on the surface.
— Wang HuananYahoo was a Yahoo with no soul
Wang Huanan attributes Yahoo's decline to two things: the founders stepped back to the second line too early — and "genuinely believed they had no management ability, so they stepped back"; and the technological soul was replaced by a commercial soul. His exact words: "A technology company with no founder, taken over by professional managers too early, is especially prone to decay," because professional managers run things well but can't ride the technology wave. By contrast, when Google brought in professional managers it made clear: "You teach me, you guide me, and at a certain point when my wings are full I'll come back" — so the founder did come back.
— Wang HuananYahoo missed Google, P&G missed Starbucks
Yahoo and Google talked about an acquisition; the conclusion was not to acquire, on the grounds that "search will never come to the front end — search is just a steam engine." At the time Yahoo outsourced search to Google, which fed Google for a long time. Google's two founders couldn't find a business model and said they might as well sell to Yahoo, but Yahoo didn't believe search could get big, and didn't believe users would care about a search brand. Wang Huanan puts this alongside P&G missing Starbucks: P&G assembled a big team to study it, and concluded it didn't believe Americans would sit on the sidewalk drinking coffee the way Europeans do. His summary: "America loses, China loses — no entrepreneurial spirit, founders left their core positions too early."
— Wang HuananGoogle dares to bet, and doesn't touch you after buying
On acquisitions, Wang Huanan sees the difference in two things: Google dares to gamble where it matters — when YouTube came out, Google bought it in one move, at a price that seemed unbelievably expensive at the time; and after buying, it doesn't remake the company first, letting it run freely for several years. He cites Keyhole: Google barely touched it for ten years, and that company is today's Google Maps. Yahoo, by contrast, "didn't dare gamble where it mattered, but gambled in a lot of places," and after buying wanted to remake things fast. He treats Tencent's acquisition of Zhang Xiaolong's mailbox as a positive case: no tasks assigned, as long as he doesn't leave the system, let him live flexibly, and once he finds his own thread, back him to do interesting things.
— Wang HuananWhat Google lacks isn't algorithms, it's closed-loop data
Wang Huanan thinks Google has a congenital defect: there's no closed-loop data on its plate. Users search and click, but after they click through to a delivery site, Google doesn't know whether they ordered, whether they repurchased, whether they were satisfied. ChatGPT, by contrast, is interactive learning — every conversation the user has adds content for it, so a hundred million users means a hundred million people helping it improve. He judges that Google "did indeed lose the first round, and I estimate it lost this round badly," because this is a game of scale, and once it's opened up it's very hard to catch up. But he also says Google doesn't lack the precursors of algorithms, compute and data, and may win it out one vertical scenario at a time.
— Wang HuananChina and the US will split half the world between them
Wang Huanan is betting that data carries political and boundary coloring — Chinese people's answers, knowledge and ways of thinking are different — so a Chinese ChatGPT will definitely be very different, and therefore "China's and America's chat and AI will share half the world between them." He divides the labor by company size: algorithms are basically a big-company matter, though small companies can take part in some of it; compute is entirely a big-company matter — Amazon has dozens of AI groups, and it's normal for one group to burn 10 to 20 million US dollars a month; data is a combination play for mid-sized and large companies. He cites Babytree: nearly 150 million pieces of maternal and infant knowledge can be made into a vertical-scenario bot, using a closed small loop to fight the big loop.
— Wang HuananThis is already an arms race in high tech
Wang Huanan says the ChatGPT team is under 100 people, 80-some, with many Chinese among them, and many of those Chinese came out of China, so this has become a talent war. He describes how Google plunders talent: no work for you to do, a million US dollars a year, just come — the thinking being "as long as you don't go to a competitor." He judges that China is fine on compute, not that far behind on algorithms, but may have only a hundred top engineers while others have a thousand. His conclusion: "Today this is an arms race in high tech" — the US has one internally, there's one between the US and China, and inside China there are still contradictions and competition.
— Wang HuananIn their own words · checked verbatim
I think we lacked this founder's soul. I think it was a Yahoo with no soul.
我觉得我们缺了创始人的这种灵魂 我觉得他是没有一个灵魂的雅虎
Wang Huanan28:25
Search is just a steam engine. Search is just a machine at the bottom of a bag. Whoever builds a machine and says it becomes a car — no one expected that search was just a back-end tool that would come to the front end.
搜索就是一个蒸汽机 搜索就是一个袋子底下的那么一个机器 谁去做一个机器说 它就成了车了 没想到搜索 它就是一个走到前台来的后台工具
Wang Huanan30:19
I think Google is one, daring to gamble where it matters; and two, after buying something, not changing the other party first — letting them play freely for several years, and only then thinking about how to glue the companies together.
我觉得谷歌是一个敢在重要的地方赌博 第二是买来东西以后先不去改变别人 先让别人自由的发挥若干年 然后再想想哪家公司怎么黏在一起
Wang Huanan33:21
What Google lacks is scenarios and data. It has a lot of data, and it also lacks a lot of loops.
谷歌差的是场景和数据 它既有很多的数据 又缺少很多的回路
Wang Huanan37:23
Today this is an arms race in high tech. In high tech there's an arms race inside the US itself, there's an arms race between the US and China, and inside China there are some contradictions and competition.
今天是高科技领域的军备竞赛 高科技领域在美国自己有军备竞赛 美国和中国之间也有军备竞赛 中国自己内部有一些矛盾和竞争
Wang Huanan1:03:18
Humans may just be a machine — only we're machines made of protein — and then, in God's words, we really are a machine.
人类可能就是一个机器 只是我们是有蛋白质构成的机器 然后在上帝的言论 我们真是一款机器
Wang Huanan1:11:40
Figures
| Share of Asians in P&G US talent at the time | 0.4% | 13:10 |
| US users Wang Huanan reached per month at P&G | Over 200 | 14:13 |
| Amazon AI group burn rate | Calculated in tens of millions of US dollars per week | 42:25 |
| Pieces of maternal and infant knowledge at Babytree | Nearly 150 million | 43:26 |
| Size of the ChatGPT team | Under 100 people, 80-some | 1:02:45 |
| Annual salary Google used to poach talent | 1 million US dollars | 1:03:18 |
Glossary
- money follows eyeball
- The internet's eyeball-economy logic around 2001: wherever user attention goes, money flows.
- Transformer
- The underlying architecture of ChatGPT; Wang Huanan explains it as the operation in linear algebra that turns a matrix and makes it simple.
- Alphabet
- Google's parent company after the reorganisation, overseeing investment and grand strategy, with the Google CEO reporting under it.
How to listen
Executives and founders who have lived through foreign-enterprise or internet-platform battles, and anyone who wants to understand why Google lost the first round and how the China-US AI division of labor will be carved up.
1:04:46 to 1:07:46 covers Lu Qi doing investment and rebuilding Babytree — personal judgment, skippable.