Three months after DeepSeek, a Chinese AI developer, shocked the tech world with a model that rivalled the best in the United States, Alexandr Wang, a 28-year-old AI executive, visited Capitol Hill to advise lawmakers on how to keep the United States at the forefront of the field.
At the hearing in April, Wang stated that the United States must create a “national AI data reserve,” provide sufficient power for data centres, and steer clear of a burdensome patchwork of state-level regulations. Lawmakers embraced his comments. “I’m glad to see you here in Washington again,” Florida Republican Representative Neal Dunn remarked. “Up here, you’re becoming a regular.”
Even if OpenAI’s Sam Altman has become a household name, Wang, the CEO of Scale AI, might not be. However, in recent years, he and his business have become quite influential in the tech and politics circles. To assist businesses create unique AI applications, Scale employs a horde of contractors to classify the data used by tech companies like Meta Platforms and OpenAI to train and enhance their AI models. According to a person with knowledge of the subject, it is increasingly employing PhDs, nurses, and other highly qualified professionals to assist in the development of increasingly complex models. In short, chips, talent, and data are the three foundations of artificial intelligence. In the latter of those, Scale is a key player.
The startup’s reputation is now expected to rise much more. Bloomberg News said over the weekend that Meta is in negotiations to spend several billions of dollars in Scale. The financing is one of the biggest private firm funding events ever, with a potential value of over $10 billion. In 2024, the firm was valued at approximately $14 billion, as part of a funding round that included Meta’s support.
The rise of Scale is similar to that of OpenAI in many aspects. When these businesses were established, about ten years ago, they were betting that the sector was about to see what Wang referred to as an “inflection point of AI.” Both of their CEOs are skilled networkers and have represented the AI industry before Congress; they are friends and temporarily shared a residence. Additionally, OpenAI received an eleven-figure investment from a major IT company.
The AI explosion that OpenAI unleashed has both influenced and been influenced by Scale’s path. To aid in training the models used to create self-driving cars, Scale initially concentrated more on categorizing pictures of cars, traffic lights, and street signs. Since then, however, it has aided in the annotation and curation of the vast volumes of textual material required to construct the so-called large language models that drive chatbots such as ChatGPT. By identifying patterns in the data and their corresponding labels, these models acquire knowledge.
That work has occasionally turned Scale into a lightning rod for complaints about the invisible labor in nations that foster AI development, including Kenya and the Philippines. Scale has come under fire for using thousands of low-paid contractors in other countries to sort through mountains of internet data; some claim that the material they are required to review has caused them psychological distress. Wang said that the company’s contract employees receive “good” pay, “in the 60th to 70th percentile of wages in their geography,” in a 2019 interview with Bloomberg.
The US Department of work has canceled an inquiry into Scale AI’s adherence to fair work laws, according to Scale AI spokesperson Joe Osborne.
The business of Scale has changed. More tech companies are experimenting with training AI systems with artificial intelligence (AI)-generated synthetic data, which could lessen the need for some of the services Scale used to offer. But the top AI laboratories are also having trouble obtaining enough high-quality training data to create increasingly sophisticated AI systems that can perform complicated jobs on par with or even better than humans.
Scale has increasingly relied on higher-paid contractors with doctorate degrees to enhance AI systems in order to meet that demand. These professionals take part in reinforcement learning, a technique that incentivizes a system for accurate responses and penalizes it for inaccurate ones.
According to a source familiar with the situation who asked not to be named because the material is confidential, the professionals who work with Scale are entrusted with creating complex problems tests, basically for the models to solve. According to the person, as of early 2025, over 40% of the company’s contributors who work on the process of upgrading these models had a master’s, law, or MBA degree in their specialty, and 12% of them had a PhD in molecular biology.
According to the individual, a large portion of this procedure is targeted at businesses looking to leverage AI for legal and medical purposes. Getting AI models to more accurately respond to inquiries about tax laws, which can vary substantially from nation to nation and even state to state, is one area of research.
The company is growing significantly as a result of bets like that. According to a Bloomberg News story from April, Scale made around $870 million in 2024 and anticipates making $2 billion this year. As more businesses engage in models that replicate human reasoning and perform increasingly complex tasks, Scale has seen a surge in demand for its network of experts following DeepSeek, according to the person familiar with the situation.
Through defense agreements, Scale has also strengthened its ties with the US government. Lawmakers on the hill who are worried about China’s rise in artificial intelligence have taken a liking to Wang, a China hawk. Also guiding US policy on AI is Michael Kratsios, a former Scale executive who is currently one of President Donald Trump’s top tech aides.
In addition to helping Meta stay ahead of AI competitors like Google and OpenAI, a closer partnership with Scale might help the company forge closer relationships with the US government at a time when it is investing more in defense technology. A partnership with Meta provides Scale with a strong and well-funded ally. For Wang, it would also be a fitting full circle moment.
Wang claimed that shortly after starting Scale, a venture funder asked him if he knew he intended to start a business. He “rattled off some silly answer about being inspired by The Social Network,” the movie about the creation of Facebook, Wang replied.
We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.
