AI × Web3: Who is building the underlying systems for this era?

When there is a real shift in the technological paradigm, we often see the hype first, rather than the system. The wave of AI that we are experiencing is no different.

As a primary investor, I always believe that focusing on the deepest transformative forces in the industry is more valuable than chasing superficial narratives.

In the past year, I have been exposed to a large number of projects such as RWA, Consumer, and infoFi. They are all exploring the intersection of the real world and on-chain systems. However, an increasingly obvious trend is that regardless of the project's direction, it ultimately needs to integrate the collaborative logic of AI to enhance competitiveness and efficiency.

For example, RWA projects need to consider how to optimize risk control using AI, validate off-chain data, and implement dynamic pricing; Consumer or DeFi projects require AI for user behavior prediction, strategy generation, incentive distribution, and more. Other projects in different directions also have similar needs.

Therefore, whether it is asset digitization or experience optimization, these seemingly independent narratives will ultimately converge on the same technological logic: if the infrastructure does not possess the integration and bearing capacity of AI, it will not be able to support the complex collaboration of the next generation of applications.

In my opinion, the future of AI is not just about becoming "stronger and stronger" and "having broader applications"; the real paradigm shift lies in the reconstruction of collaborative logic.

Just like the early transformation of the Internet, it wasn't because we invented DNS or browsers, but because it allowed everyone to participate in content creation, turning ideas into products, thereby giving rise to an entire open ecosystem.

AI is also taking this path: Agents will become intelligent co-creators for everyone, helping you transform expertise, creativity, and tasks into automated productivity tools, even enabling monetization.

This is a problem that is difficult to solve in the current Web2 world, and it also reflects some underlying logic of my focus on the AI + Web3 track: making AI collaborative, transferable, and profit-sharing is the system that is truly worth building.

Today I want to discuss the only project so far that attempts to systematically build the underlying operation of AI from a chain-level structure: Sahara.

The essence of investment is worldview, recognizing the value system of choices.

My investment logic is not simply to combine public chain narratives with AI and then look for well-backed teams to bet on.

Investment is essentially a choice of worldview, and I have always been questioning a core issue: Can the future of AI be jointly owned by more people?

Can it leverage blockchain to reconstruct the value attribution and distribution logic of AI, allowing ordinary users, developers, and other roles to have the opportunity to participate, contribute, and continuously benefit? In short, I only believe that such projects can potentially become disruptors if this logic emerges, rather than being "just another public chain project."

In order to find the answer, I researched almost all the AI projects I could access until I came across Sahara. The response I received from Sahara's co-founder Tyler was: to build an open, participatory ecosystem that everyone can own and benefit from.

This sentence is simple, but it precisely hits the soft spot of traditional public chains: they often serve developers in a one-sided manner, and the design of the token economy is mostly limited to Gas Fees or governance, rarely able to truly support a positive cycle of the ecosystem, making it even harder to sustain the development of an emerging track.

I am well aware that this path is full of challenges, but precisely because of this, it is an irresistible revolution—this is also the reason for my steadfast investment.

As I emphasized in my previous discussion on the "Evolution from Web2 to Web3": the real paradigm shift lies not in creating a single product, but in building a supportive system.

And Sahara was one of the most anticipated cases in my prediction at that time.

From investment to 8x valuation follow-up investment in heavy positions

If I initially invested in Sahara because it is fulfilling the mission of what I believe to be the true leader in AI - building an AI economy and infrastructure system, then the reason I rushed to invest again at an 8 times pre-round valuation in just half a year is that I felt a rare strength in this team.

Among the two co-founders, one is the youngest tenured professor at USC, specialized in AI. The value of a tenured professor in American universities from the 90s generation is reflected not only in the academic field but also in the fact that this age group still has dreams, energy, and the determination to realize those dreams. In the more than a year I have known Professor Ren, I have witnessed what it means to work for more than ten hours a day, remain emotionally stable, and be a humble genius.

Another co-founder, Tyler, who previously served as the investment director at a well-known trading platform's lab, was responsible for North American investments and incubators, and his understanding of Web3 goes without saying. He is astonishingly self-disciplined: he only sleeps in multiples of 1.5 hours, insists on working out regardless of how busy he is to maintain his condition, and avoids even a piece of candy to keep his mind clear, working over 13 hours a day. I once joked that he is like a robot, to which he simply responded: "I am lucky to have this busy life today." His source of dopamine comes from making progress on projects every day; dreaming is his passion, and he doesn’t need any other fuel.

I am very fortunate to have met them, which has also changed me. I have started to maintain a regular schedule as much as possible, my emotions have gradually stabilized, and I stick to fitness...

So when someone says that Sahara has attracted capital's favor due to luck, I always unreservedly add, "The pursuit of capital is an inevitable result." I vividly remember the difficulty of primary financing in this round of the market, but Sahara was being chased by investors in the primary market.

What everyone remembers is that a well-known cryptocurrency investment firm, a large trading platform, and a venture capital company invested in Sahara. Sahara has opened the investment era for a well-known technology company to enter the Web3 AI field, and its receipt of the AI award from that company is a significant reason for the investment. In addition, some AI-heavy funds, national banks, and others are also guests of Sahara. You can see a group of institutions that are more focused on traditional technology and industrial resources beginning to quietly bet on AI × Web3 because of Sahara.

Capital will only pay for a direction and execution capability that is certain - this is positive feedback on the depth of Sahara technology, team background, system design, and execution capability.

This is also why it can produce some real and solid structural indicators:

More than 3.2 million accounts have been activated on the test network, with over 200,000 data platform annotators (millions in queue). Their clients include several leading technology companies, and they have already achieved revenue in the tens of millions of dollars.

On this infrastructure chain, at least from "who can do it" to "can it be done", Sahara has gone deeper and steadier than 99% of the "AI Narrative projects".

The ultimate challenge of public chains: ensuring all contributors continue to benefit and driving a positive economic cycle

Returning to our initial judgment logic: In a system where AI and blockchain are combined, is there really a mechanism that allows every contributor to be seen, recorded, and continuously rewarded?

Model training and data optimization rely heavily on extensive support from annotations and interactions; conversely, if there is a lack of user contributions, the project itself has to invest more funds to procure data and outsource annotations, which not only increases costs but also weakens the value-driven nature of community co-construction.

Sahara is one of the few Web3 AI projects that allows ordinary users to "participate in data construction from day one." Its data labeling task system operates daily, with a large number of community users actively involved in labeling and prompt creation. This not only helps improve the system but also invests in the future with data.

Through the mechanism of Sahara, not only is the quality of the model improved, but more people can also understand and participate in this decentralized AI ecosystem, linking data contribution to benefits, thereby forming a true positive cycle.

A typical example is a voice synthesis project on a public blockchain that utilizes Sahara's decentralized data collection and human-machine collaborative annotation to quickly construct a high-quality dataset covering multiple languages and accents, significantly enhancing the training efficiency of its TTS and voice cloning models. This has also driven its open-source project to receive thousands of GitHub stars and over 2 million model downloads.

At the same time, users participating in data labeling also received token rewards issued by the project, forming a two-way incentive loop between developers and data contributors.

Sahara's "permissionless copyright" mechanism ensures the open circulation and reuse of AI assets while safeguarding the rights of all participants—this is the underlying logic driving the explosive growth of the entire ecosystem.

Why is it said that this is a scenario with long-term value support?

Imagine if you want to build an AI application, naturally you hope your model is more accurate and closer to real users than others.

The key advantage of Sahara is that it connects you with a vast and active data network—hundreds of thousands, and in the future millions of annotators. They can continuously provide you with customized, high-quality data services, allowing your model to iterate faster.

More importantly, this is by no means a one-time transaction. Through Sahara, you are connecting to a potential early user community; and these contributors are very likely to become the real users of your product in the future.

This connection is not a one-time buyout; through Sahara's smart contract system and rights confirmation mechanism, it enables a long-term, traceable, and sustainable incentive system.

Regardless of how many times the data is called, contributors will receive continuous profit sharing, with earnings dynamically linked to usage behavior.

But this is not just a revenue model for data labeling and model training phases. Sahara builds an economic system that covers the entire lifecycle of AI models, with a built-in profit-sharing mechanism at every stage, including model deployment, invocation, combination, and cross-chain reuse, allowing value to be captured over a longer period.

Model developers, optimizers, validators, and computing power contribution nodes can now continuously benefit at different stages, rather than just relying on one-time transactions or buyouts.

Such a system brings a compound effect for model combination calls and cross-chain reuse. A trained model can be repeatedly called and combined by different applications like building blocks, with each call generating new revenue for the original contributor.

Because of this, I resonate with Sahara's underlying belief: a truly healthy AI economic system cannot just be about the plundering of data and the buyout of models, nor can it be about allowing a few to reap all the benefits. It must be open, collaborative, and win-win—where everyone can participate, every valuable contribution can be recorded, and future rewards can be continuously obtained.

But the closer we get to the real structure, the more challenges there are.

Although I am optimistic about Sahara, I will not hide the challenges that the project will face because of my investment position.

One of the major advantages of the Sahara architecture is that it is not limited to any specific chain or single ecosystem.

Its system was designed from the start to be open, full-chain, and standardized: supporting deployment on any EVM-compatible chain, while also providing standard API interfaces that allow Web2 systems—whether e-commerce backends, enterprise SaaS, or mobile apps—to directly call Sahara's model services and complete on-chain settlements.

However, despite the extreme rarity of this architectural design, it also has a core risk: the value of the infrastructure does not lie in "what it can do," but in "who is willing to do what based on it."

To become a trusted, adopted, and integrated AI protocol layer, the key for Sahara lies in how ecological participants evaluate its technological maturity, stability, and future predictability. Although the system itself has been built, whether it can truly attract a large number of projects to implement based on its standards remains uncertain.

It is undeniable that Sahara has achieved key validation: providing relevant data services to several leading technology companies and tackling some of the industry's most challenging data demand issues, becoming an early signal of the feasibility of this system.

But what needs to be seen is that these collaborations mainly come from the Web2 world; what truly determines the long-term development of Sahara is still the overall situation.