DeepSeek Begins Wealth Creation as It Transforms into a Commercial AI Company

Author | Painting

For a long time, DeepSeek didn’t seem like a company.

Liang Wenfeng does not seek funding, does not give interviews, and does not make any external statements.

The outside world doesn’t know how many people DeepSeek has, how much money it has made, or where it’s ultimately headed.

Until recently, it suddenly began hiring on a large scale.

I believe this matter is even more important than the 51 billion yuan funding.

This means that, for the first time, DeepSeek has transformed from a mysterious laboratory into a truly commercial company.

During DeepSeek’s most mysterious period, there were no product manager positions, no large-scale R&D collaboration processes, and no demand for Agent commercialization roles.

Liang Wenfeng funds the entire team with money earned from his quantitative fund. The company exists to publish papers and release models that reshape industry expectations, waking up the global AI community on Lunar New Year’s Day.

Today is different.

Hiring on a large scale, accepting external funding, forming the Harness team, and even team leader Cui Tianyi stepping forward to explain hiring criteria to the media—taken together, these actions only indicate one thing: the era of DeepSeek as a myth is over; an AI company that must confront market competition has now emerged.

At the most grassroots stage of WeChat, Taobao, and Douyin, ordinary people had no access. The real large-scale benefits were only unleashed after companies began establishing organizations, building platforms, and introducing equity incentives.

Today, DeepSeek has also entered this stage.

As a company, it began to need to scale talent acquisition in bulk, not just support a few geniuses.

But DeepSeek is not the whole story.

First, let's look at a few sets of numbers.

DeepSeek, Series A funding of 51 billion yuan, post-investment valuation of 400 billion yuan.

Zhipu, listed on the Hong Kong Stock Exchange in January 2026, saw its market capitalization exceed one trillion Hong Kong dollars six months later, with a gain of over 1,900% in six months.

Moonshot, Baichuan, Jieyue, and ZeroOne Wanwu are almost all in the Pre-IPO stage, with valuations rising and capital continuing to flow in.

The message conveyed by this list is that China's AI industry has, for the first time, entered a phase where employees can jointly create wealth with their companies.

If you extend the timeframe, this will become more apparent.

Over the past decade, China has also produced many outstanding companies.

But a new industry segment that truly, like mobile internet, can continuously create wealth, jobs, and industries has not yet emerged.

After the internet dividend ended, big tech companies remain strong, but young people joining Tencent, Alibaba, or ByteDance are more like joining a mature enterprise rather than embarking on a shared growth journey as a startup.

AI is the first new industry in the past decade to reopen this door.

Why can the internet create wealth?

Many people answer listing and cashing out. But listing is just the outcome. The real reason is that the internet created an entire new industry.

Portals, search, e-commerce, payments, advertising, and cloud computing—each layer gave rise to new companies, new jobs, and new wealth. It wasn’t just ByteDance, Tencent, or Alibaba that won; it was the entire industries they spurred into existence.

AI is also replaying this logic.

The model is just the first layer. Beyond that, there are agents, chips, robots, AI office tools, AI education, AI healthcare... Over the next decade, an increasing number of new companies will emerge from these layers.

DeepSeek alone cannot create a wealth-generating cycle. What truly creates this cycle is the fact that the entire AI industry is coming together.

DeepSeek went from founding to a $40 billion valuation in less than three years. Zhipu went public less than six months ago, and its stock has already surged nearly 20-fold. The internet boom once saw similar speeds, but that window closed nearly a decade ago.

Today, it reopened.

Regarding the timing of joining DeepSeek, I’ve seen many comments saying: “It’s already too late.” Early employees made the most, and now that you’re joining, the initial红利 have already been fully consumed.

Not necessarily.

One detail of DeepSeek’s this funding round is rarely mentioned. As I previously noted in my articles, one of the purposes of this financing is to address employee stock option pricing, liquidity, and cash-out issues, in order to retain talent. (Further reading: Why is Liang Wenhong raising funds if he’s not short on money?)

Before this, the options held by DeepSeek employees were like blank pieces of paper with no clear price, as the company had not raised funds, gone public, or had any external valuation references.

After a $5.1 billion funding round, a $400 billion valuation has been established. For the first time, the options held by employees have a price.

From 40 billion to a future listing, there is still a journey ahead. Those who join today stand at the entrance to this journey.

At the end of 2017, ByteDance's funding valuation was approximately $22–25 billion, and most investors at the time considered the valuation high.

In 2021, three years later, the internal valuation corresponding to employee equity incentives and private secondary share transactions once exceeded $300 billion. While valuations in the primary market are difficult to predict precisely, the long-term growth thesis rooted in the globalization of short-form video remains valid.

Options have been covered. Now let’s talk about what DeepSeek is really betting on with this hiring.

First, a judgment: today’s competition in AI is transitioning from the first stage to the second stage.

In the first phase, everyone competes on models.

Who has the stronger large model, higher parameter efficiency, and lower inference cost? At this stage, DeepSeek has won, creating a model that outperforms GPT with fewer resources, prompting the global AI community to reassess its technical pathways.

In the second phase, everyone is competing in Agent development.

OpenAI's new Agent product, Codex, saw its weekly active users surge from 600,000 to 5 million in the first five months of 2026, an increase of approximately 730%, evolving from a programming tool into an entry point for delegating entire tasks to autonomous Agents.

Anthropic’s Claude Code, not to mention, is now the global leader among AI agents, scoring higher than Copilot and Cursor in developer surveys and becoming the most widely used AI programming agent on GitHub. Google’s Project Mariner enables AI to directly take over your browser and navigate web pages for you.

DeepSeek established the Harness team based on this very assessment. Cui Tianyi stated, "There is no rule against hiring foreigners," implying: We are fighting in this global battle.

Harness is targeting Codex and Claude Code, aiming to enable Chinese AI products to truly enter the global Agent competition.

Many people think DeepSeek is hiring for R&D.

I actually think it's redefining the next generation of engineers.

In the past, a great engineer needed to know how to code, understand systems, and comprehend architecture.

In the Agent era, an excellent engineer needs to understand reasoning, understand Agents, and understand how models truly accomplish tasks.

DeepSeek is hiring more than just employees.

It is actually setting the standard for the most expensive talent in the AI industry over the next five years: those who can drive products from 0 to 1 in the Agent era.

Each technological revolution redefines a generation of engineers.

In 2003, those who joined Alibaba later learned internet commerce.

In 2011, those who joined Tencent WeChat later learned about social products.

In 2014, those who joined ByteDance later learned algorithmic recommendation.

Are you going to DeepSeek today? Is the option worth it? What’s the salary? Many people are doing the math.

After calculating, there’s still one more important thing that hasn’t been accounted for.

That’s why those who entered the core of the AI industry acquired, ahead of time, a scarce skill that the industry will need for the next decade.

For example, back then, Alibaba sparked a wave of internet commerce trends—how to build an e-commerce platform from zero users, assemble a team, and turn an idea into a product with millions of daily active users. This methodology is something no resume could ever claim after the internet bubble.

The social products developed by Tencent have learned how to drive user growth, how to understand connections, and how to transform a tool into a platform.

ByteDance drives algorithmic recommendations—how to make data-driven decisions and how to cold-start a new product in an extremely short time. This set of capabilities was rarely understood before TikTok's global surge.

Today, what DeepSeek teaches may be Agent.

In five years, will every company need an Agent engineer? No one knows.

But it’s certain that very few people today truly understand Agents.

An engineer who joined DeepSeek today is working at the forefront of inference optimization, agent architecture, and multimodal collaboration. Globally, the number of people who truly understand these areas combined is fewer than a few thousand today.

In five years, when AI agents become standard for every company, where will these thousands of people be?

DeepSeek is currently opening only 36 positions for recruitment.

But this wave of wealth creation involves far more than just DeepSeek.

ZhiPu, MiniMax, Moonshot, Baichuan, and Jieyue together offer thousands to tens of thousands of positions across every stage, including model development, agent applications, data labeling, product operations, and commercialization.

Greater opportunities lie beyond these companies.

Especially for ordinary people, the real benefit of AI has never been joining AI companies, but rather turning every industry back into a node where ordinary people can start businesses.

For example, designers are redefining creation, sales teams are redefining customer relationships, teachers are redefining instruction, and doctors are redefining diagnosis.

Every time a productivity tool is rebuilt, the ones who ultimately gain the greatest allocation power are not the developers of the tool, but those who first use the new tool to rebuild their business.

After the explosion and widespread adoption of mobile internet, the beneficiaries weren’t just employees of Tencent, Alibaba, and ByteDance, but more importantly, bloggers who learned to create content online, merchants selling products through e-commerce, and every ordinary person using traffic to do business.

AI is the same this time. The arrival of the Agent era is creating an opportunity for ordinary people in every industry. Find the intersection between your core skills and AI, and use AI to amplify what you do best—you may see results sooner than by joining DeepSeek.

Years later, the gap between them and their peers is no smaller than that of programmers who joined ByteDance in 2014.

Outside the layout:

DeepSeek is launching a large-scale hiring drive, seemingly offering positions across only 36 categories.

But what it truly tells everyone is that Chinese AI has entered a new phase where it can consistently create wealth, companies, and careers.

The new wealth creation cycle does not belong to DeepSeek—it belongs to this generation that learns to use AI.

This article is from the WeChat official account "Beyond the Layout," authored by Hua Hua.