▲Los Alamos, the place of discovery!

The economic structure of Los Alamos is extremely single. The core is Los Alamos National Laboratory, with 16,487 regular employees and a total of about 17,925 contract workers. It is one of the largest employers in the state.

As you can imagine,Many of the Wang family’s neighbors are physicists, mathematicians, and engineers.

When Wang Tao was a little older, the family would talk about atomic theory, numerical simulations, and national defense projects at the dinner table instead of watching short skits on their mobile phones. There were a lot of parents with Ph.D.s in the school, and Halloween might become a science display.

What kind of child you want depends on what kind of environment and information you want to create.

Children like Wang Tao are naturally very sensitive to data and scientific things. This is the result of information immersion and the result of the intentional or unintentional efforts of parents.

One more thing, when young people are looking for a partner, the knowledge and cognition of the partner are very important. Excellent knowledge and knowledge will benefit several generations and even enhance the status of a family.

Backward knowledge and cognition will destroy several generations and bring a family into a situation of no return.

Wang Tao has been nurtured in such an information environment for a long time. He has his own edge and brilliance, and is confident in using data and scientific methods to solve problems instead of using emotions to solve problems.

02 Working with Silicon Valley engineers in my senior year of high school

In elementary school, Wang Tao showed a sense of number. He actively participates in American primary and secondary school mathematics competitions because if he wins the competition, he can go on free trips, such as going to Disneyland. The fun lies in learning.

His parents did not force him to study for exams like a tiger mother. Seeing that he was interested in computers, they asked someone to teach him programming so that he could understand some algorithms and data structures early on.

Tolerant parents do not worry about their children playing games, but create an environment where they can play and learn, because the conversations at the dinner table and the parents’ words and deeds are enough to stop them from playing games crazy and focus more on scientific research.

In high school, Wang Tao won many good places in mathematics competitions and physics competitions. He learned programming by himself and competed in programming competitions. He could also play the violin and speak Chinese, English, and French. He is a “competitive all-rounder” raised in a scientific research family, but he is not a nerd.

Therefore, he is very resistant to pressure, resilient, and willing to deal with and cooperate with people with an open mind, rather than guarding against this or that.

The young man’s working career began.

At the age of 17, Wang Tao first went to Addepar (a wealth management and financial big data platform) in Silicon Valley to work as a software engineer, not an intern.

Then I entered Quora (a question and answer community) to do back-end, performance, and infrastructure engineering. Here I met my entrepreneurial partner Lucy Guo (a Thiel scholar and later co-founder of ScaleAI).

▲Wang Tao and Lucy Guo (left)

While other kids are still preparing for college entrance exams in their senior year of high school, he is already working with Silicon Valley engineers.

This young man from a small town in New Mexico entered the start-up environment of California’s Silicon Valley where “you are in your 20s and you can handle billions of requests”, which had a great impact on him and made him have a broader perspective.

Of course, you still have to study in college.He applied to MIT and dropped out after one year., are the grades too poor? Not really.

He studied mathematics, computer science, and machine learning. He took graduate-level CS courses in his freshman year, and his GPA was perfect.The reason for dropping out was that the window for starting a business was too obvious and I didn’t want to wait four years.

According to domestic standards, he has failed in life, and there is no chance of him taking the public examination.

Because you don’t have a bachelor’s degree, how can you take the public examination?

At this time, Wang Tao had already integrated into society, Silicon Valley, and technology companies.

Both his parents were scholars. They did not teach him “to be a billionaire in the future”, but gave him three things directly or indirectly:

First, treat science as the way the world works——Technology is not a tool, but the chassis of civilization;

Second, regard “changing the world” as a serious life goal, and he did say so later;

Third, be instinctively sensitive to basic scientific research, computing power, and data.——This later directly influenced him to start a business to create AI training data, and later to join Meta to develop super intelligence.

03 A small thing triggers entrepreneurial inspiration

When he was in college, a small incident triggered his entrepreneurial inspiration.

His milk and sandwiches were often stolen by his roommates, and he had no evidence. He wanted evidence.

So he installed a camera in the dormitory refrigerator and used AI to catch “roommates who stole my milk and sandwiches.” The model was written, but it couldn’t be used – there were too many video frames and no labeled data, so the model couldn’t learn “what is a half-bottle of Coke” at all.

Isn’t this the pain point?

Wang Tao later said this sentence repeatedly:Algorithms are not the bottleneck, clean, trainable data is.

That is to say,The real bottleneck of AI is not insufficient algorithms, but the scarcity of high-quality annotated data and training data infrastructure.

So, he decided to start a data annotation and data infrastructure companyScale AI, got $120,000 in seed funding and started working.

He also told his parents, “Just do it for a summer first.” As a result, he never went back. His parents were deceived and there was nothing they could do.

In 2016-2017, ChatGPT was not there yet, and the hottest thing in Silicon Valley was autonomous driving.

Scale’s early customers are not large model companies, but a bunch of companies related to autonomous driving.

When Wang Tao goes to computer vision conferences like CVPR, he carries a notebook, visits one house after another, opens the demo and says: “You give me a driving video, and I will give you marked data on pedestrians, vehicles, and lane lines.”

He charges per task and delivers much faster than large companies can outsource in-house.

In the early days, there were only a few people in the company, and employees carried backpacks to the customer conference room. The other party thought they were teaching outside the university or couriers and accidentally broke into the conference room.

The company started like this, and the development nodes are very clear:

● 2016–2018: Autonomous driving data annotation

Image, lidar, map, sensor fusion.

● Around 2019: NLP and large model warm-up

Provide text sorting, command data, and preference annotation for cutting-edge laboratories such as OpenAI.

● 2020: Government, Defense

Scale Donovan, SEAL evaluation, US military AI readiness, geospatial, intelligence data.

● 2022–2025: Generative AI infrastructure

Large model post-training, red team, evaluation list, enterprise AI data engine.

In 2019, Scale AI’s valuation exceeded 1 billion, making it a unicorn.

In 2025, Scale AI will be valued at approximately 29 billion.

Customers have changed from car companies to OpenAI, Google, Microsoft, Meta, Amazon, Nvidia, and the U.S. Department of Defense. Revenue in 2025 is estimated to be nearly US$2 billion.

04 14.3 billion in exchange for 234 billion market value growth

At this time, Meta encountered difficulties and wanted to solve the problem. Meta does not lack computing power, but it also lacks high-quality data and data pipelines.Zuckerberg wants a young talent who “understands both AI infrastructure, national defense, and enterprise sales, and can speak to both Washington and Silicon Valley.”, so he really wanted Wang Tao.

Scale AI is a good target. Meta didn’t buy it based on gross profit, but because Zuckerberg wanted to use $14.3 billion for data infrastructure and to replace Wang Tao with leading the super-intelligence team.

Meta is not buying the entire ScaleAI company;Acquire 49% of Scale AI’s non-voting shares for US$14.3 billion in 2025, and at the same time poached Wang Tao to Meta to manage the super-intelligent laboratory, so the essence is to buy data infrastructure and people.

Facts have proved that the 14.3 billion was well spent.Wang Tao led the team to create Muse. Muse does not chat with users, it does work automatically for users.

Users only need to give a final goal, and Muse can complete the entire process independently.: Automatically open browsers, retrieve information, fill in forms, send emails, plan schedules, and organize files. Even if the app is closed, it can still continue to perform tasks in the background.You can think of it as hiring an assistant for $20 a month.

The 29-year-old Wang Tao showed his sword in Meta for the first time, directly surpassing the industry hegemon ChatGPT.

▲Zuckerberg posted a photo with Wang Tao

Within five trading days after Muse went online, Meta’s market value increased by approximately US$234 billion.. US$14.3 billion in exchange for a market value increase of US$234 billion is a good deal.If you choose the right person, making money is never a problem.

Related Reading

• The west “breaks zero”, and the competition for the word “rice” in inland provincial capitals begins2026-09-29
• Original Luo Yonghao makes enemies everywhere, Zheng Gang resorts to “tax evasion” killer weapon2026-09-29
• Li Bin pledged part of Weilai’s most valuable wealth to Geely2026-09-29
• power plant2026-09-29
• ‘Is it really worth it to keep going on?’ Economic pain hits America’s farm belt.2026-09-28