AI × Web3: Building Infrastructure for a New Era

When the technological paradigm truly shifts, we often see a wave of excitement first, rather than a system. The current wave of AI is no exception.

As a primary investor, I always believe that betting on the transformative forces deep within the industry is far more valuable than chasing superficial narratives.

In the past year, I have examined 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 route a project takes, it ultimately needs to integrate AI's collaborative logic to enhance competitiveness and efficiency.

For example, in the RWA sector, we need to think about how to use AI for risk control optimization, off-chain data verification, and dynamic pricing in the future. Alternatively, Consumer or DeFi projects that urgently need to improve user experience also require AI for user behavior prediction, strategy generation, incentive distribution, and more. Other fields also have similar demands.

Therefore, whether it is asset digitization or experience optimization, these seemingly independent narratives will ultimately converge into the same technical logic: if the infrastructure does not possess the integration and carrying 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 being "used more and more"; 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 for the first time, turning ideas into products, thus giving rise to an entire open ecosystem.

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

This is a problem that is currently difficult to solve in the Web2 world, and it is also some of the underlying logic I am focused on in the AI + Web3 track: making AI collaborative, transferable, and profit-sharing is the truly worthwhile system to build.

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

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

My investment logic is not simply to combine the public chain narrative with AI and then bet based on the team's background.

Investment is essentially a choice of worldview, and I have always been asking a core question: 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 participate, contribute, and continue to benefit? Only with such a logic in place do I believe that these types of projects could potentially become disruptors, rather than "just another public chain."

To find the answer, I basically examined all the AI projects I could access until I encountered Sahara. The response given by Sahara's co-founder 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 token economics is usually limited to Gas Fees or governance, rarely able to truly support a positive cycle of the ecosystem, and even more difficult to sustain the development of an emerging track.

I am well aware that this road is full of challenges, but precisely because of this, it is an irresistible revolution—this is also the reason why I am firmly investing.

As I emphasized earlier in my discussion on the "evolution from Web2 to Web3": the true paradigm shift lies not in creating a single product, but in building a supporting system.

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

From Investment to 8x Valuation Follow-up Heavy Investment

If I initially invested in Sahara because it is doing exactly what I believe is the true leading mission of AI - building the AI economy and infrastructure system, then what drove me to rush in and invest at an 8x pre-round valuation in just half a year is the rare strength I felt in this team.

Two co-founders, one of whom is the youngest tenured professor at USC, specializing in AI. The value of a tenured professor in their 90s is not only reflected in the academic field but also in the fact that they still have dreams, energy, and the determination to realize those dreams at such an age. Knowing this professor for over a year has shown me what it means to work for more than ten hours a day, with stable emotions and humility.

Another co-founder is a former investment director at a well-known blockchain laboratory, responsible for North American investments and incubators, and his understanding of Web3 goes without saying. His level of self-discipline is astonishing: he only sleeps in multiples of 1.5 hours, insists on working out no matter how busy he is to maintain his condition, and avoids sugar to keep his mind clear, working over 13 hours a day. I once joked that he is a robot, to which he simply replied: "I am lucky to have the busyness of today." His source of dopamine comes from pushing the progress of projects every day; dreaming is his passion and requires no other fuel.

I am very glad to have met them, which has also changed me. I have also started to sleep regularly as much as possible, my emotions are gradually stabilizing, and I have started working out...

So when someone says that Sahara gained the favor of capital due to luck, I will not hesitate to add, "The pursuit of capital is a natural 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 certain well-known investment institutions have invested in Sahara. Sahara has ushered in an investment era for a large technology company entering the Web3 AI field, and its receipt of the company's AI award is a significant reason for the investment. In addition, some funds heavily invested in AI, national banks, and others are also guests of Sahara. What you can see is a group of institutions more focused on traditional technology and industrial resources beginning to quietly bet on AI × Web3 because of Sahara.

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

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 queued), serving clients that include several leading technology companies, and achieving revenue in the tens of millions of dollars.

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

The ultimate challenge of public chains: ensuring continuous benefits for all contributors and driving a positive economic cycle.

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

Model training and data optimization rely heavily on a large amount of annotation and interaction support; 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 participating 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, it not only enhances the quality of the model but also allows more people to understand and participate in this decentralized AI ecosystem, linking data contributions with benefits, thus forming a true virtuous cycle.

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

At the same time, users participating in data annotation 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 protecting 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 supported by long-term value?

Imagine that if you want to build an AI application, you naturally hope that 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 likely to become 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 can achieve a long-term, traceable, and sustainable incentive system.

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

But this is not just a revenue model for data labeling and model training stages. Sahara builds an economic system that covers the entire lifecycle of AI models, with a built-in profit-sharing mechanism at every stage of 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 compounding effect for model combination calls and cross-chain reuse. A trained model, like building blocks, can be repeatedly called and combined by different applications, with each call creating new revenue for the original contributor.

For this reason, I agree with Sahara's underlying belief: a truly healthy AI economic system cannot be just about data plundering and the buying out of models, nor can it be about allowing a few people to reap all the benefits. It must be open, collaborative, and mutually beneficial—where everyone can participate, every valuable contribution can be recorded, and rewards can continue to be received in the future.

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

Although I am optimistic about Sahara, I will not cover up the challenges that the project will face because of my investment stance.

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 beginning to be open, full-chain, and standardized: it supports 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 scarcity of this architectural design, it also carries 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 assess 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 land based on its standards remains uncertain.

Undeniably, Sahara has achieved key validation: it provides relevant data services to several leading technology companies and addresses some of the industry's most challenging data demand issues, becoming an early signal of the feasibility of this system.

However, it is important to note that these collaborations mainly come from the Web2 world. The true factors that will determine the long-term development of Sahara are still the maturity of the entire Web3 AI sector.