On December 6, 2013, several important things happened on the same day.
The most important thing is that Google sent them their draft sharing contract. Regarding how Meng Fanqi’s AI technology should share the profits, there are more ways than ever.
Meng Fanqi had previously signed a service contract with Baidu for a single technology. It was just a real-time detection algorithm for images. There were already quite a lot of things that needed to be agreed upon.
Now, Meng Fanqi needs to agree with Google on how various types of technologies should be divided.
As we all know, there are some technologies that make money and some that don’t. There are those who seem to be losing money but are actually making money, and there are also those who are looking like they are making money but are actually losing money.
If epoch-making products such as self-driving cars and humanoid robots can be marketed, they will undoubtedly bring huge profits, but this is still far away and is in the pure money-spending stage.
For technologies that can be monetized immediately, technologies that are not so easy to monetize but are very important, and technologies that are purely invested in research and development, although they will not be monetized, Google very much hopes to make breakthroughs.
Different types of technologies and directions are divided into complex ways.
Google's contract is very thick in this regard. It stands to reason that the technologies Meng Fanqi has released so far are purely image-related technologies, but this contract still stipulates many other types of technologies.
Very strict and strict.
For example, for Google, what technology is monetizing the fastest and generating the highest revenue?
There is no doubt that it must be advertising revenue. Google's advertising revenue in a quarter is usually about 20 billion US dollars, accounting for more than 80% of its total revenue.
When the harvest is good, the accumulated revenue in a year is close to 100 billion U.S. dollars, which is the foundation to support all Google projects.
Currently, a valuable way for Google to make money is by predicting users' click choices with increasing accuracy.
According to its internal statistics, even a 0.1 percent accuracy improvement will generate hundreds of millions of dollars in additional revenue.
If Meng Fanqi makes a breakthrough in the recommendation algorithm, it will naturally be tested directly. If the effect is good, it will be deployed directly online.
The sharing in this case is very simple. According to the extra income brought to the company, the commission will be drawn in proportion to it, which is easy to understand.
It may be difficult for laymen to understand how to calculate the extra income after technological improvements.
In fact, regarding the recommendation algorithm, how much users’ willingness to click has increased, and how much their willingness to consume has increased, these have been Google’s biggest gains for many years.
Therefore, the internal algorithms and statistics in this area are extremely advanced, and each quarter's forecast of advertising revenue for the next quarter is quite accurate.
Meng Fanqi's experience is mainly in image algorithms. He has learned a lot about natural language and large models, but not so much about recommendation algorithms.
Fortunately, it is still early. Although Meng Fanqi rarely pays attention to recommendation and advertising algorithms after graduation, he still has some impressions of the classic algorithms as of 2018-2019.
Especially some papers produced by big companies such as Google, Facebook and Ali Penguin, I have read them carefully several times.
"If I hadn't read the contract, I would have forgotten that Google is actually a well-hidden advertising company..." Meng Fanqi murmured to himself. Everyone in the world thinks that Google is a technology giant, but its main income depends on advertising, and the rest accounts for less than 20%.
"If that's the case, my top priority in the past few months should actually be recommendation and advertising algorithms." Meng Fanqi thought about it for a moment. Google's share in this area is very high, reaching 20-30% within one or two years, but in the future It will be gradually diluted, with only the first two years providing much more.
If Meng Fanqi now uses Google's more classic recommendation algorithm from 2015 to 2016, Meng Fanqi personally estimates that the increase in advertising revenue for Google will be between 5 and 10%, close to 10 billion US dollars.
The 2-3 achievement is more than 2 billion US dollars, which is close to 5 billion US dollars in two years. This will be the most terrifying income he can get at this stage.
Google has actually adopted some methods to limit Meng Fanqi’s share of the revenue, mainly through the collaborator model. If there are Google personnel or other personnel involved in the technical achievements, Meng Fanqi’s contribution will be diluted.
For example, if the recommendation algorithm was completed by him and four Google employees, then even if he is in a dominant position, only 30 to 35% of the contribution may be counted in the end.
Others did not have a share contract and had no right to share, so the share of 2 billion directly turned into about 5 to 6 billion.
To launch this kind of technology, it is often possible for a team to exceed a dozen or twenty people. If it is allocated to Meng Fanqi, there may even be only tens of millions of dollars left, which is a good calculation by the capitalists.
It's a pity that Google never thought that Meng Fanqi did not need the help of collaborators most of the time.
For cost-saving technological breakthroughs, the sharing method is similar, with 30% for one year.
Mature technologies of technology companies iterate very quickly, and the technical effects will be difficult to calculate over time. Therefore, the sharing periods generally given are relatively short.
But fortunately, the ratio was not low and it came quickly, so Meng Fanqi was quite satisfied.
As for technologies that cannot be monetized at present, they are in the form of long-term sharing. The share ratio may be lower, but it also depends on the proportion of the technology in profits.
Generally speaking, Meng Fanqi is actually quite satisfied, because he is not worried at all about some of the operations Google has done against him.
For example, using collaborators to dilute the contribution of technology.
Another example is long-term sharing of technology that cannot be monetized. If the technology you develop does not make money in the end, you will not get a penny.
But Meng Fanqi happens to know which technologies have the best response and generate the most money.
It can be said that it is completely grasped.
That afternoon, Meng Fanqi finally read the current contract twice with the help of Han Ci.
Based on this share ratio, Meng Fanqi is confident that he can directly steal at least US$2 billion from Google in 2014.
This is much higher than the two-year signing fee of US$20 million originally offered by Google.
This is the charm of gambling-style sharing contracts, and even technology giants like Google cannot beat reborn people in gambling.
Unfortunately, Meng Fanqi was still not satisfied. If he had seen this contract when they met for the first time the day before yesterday, he would have signed it without saying anything.
But today, Meng Fanqi learned that Baidu and Yanjing Changping District have reached a cooperative relationship, which means that Baidu's technology conference later today will raise his worth and status to another level.
"Someone from Google has just come to urge me. I wonder what you think about it?"
After getting the contract, Meng Fanqi needed a quiet room to read the terms carefully. The whole process took a long time.
Someone from Google came outside to inquire about the situation. Tang Juan, Liu Xu and others sent the message to Han Ci who was assisting Meng Fanqi in reading.
Han Ci did not mention this matter until Meng Fanqi read the entire contract and put it aside.
"Let them wait a little longer, there will be a show tonight."
