AI × Web3: Builders of Infrastructure in the New Era

When the technological paradigm truly shifts, we often see a frenzy before the system. The wave of AI we are currently experiencing is no different.

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

In the past year, I have come across a large number of projects exploring the intersection between the real world and on-chain systems. However, an increasingly evident trend is that no matter which route the project takes, it ultimately needs to enter the collaborative logic of AI, utilizing 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; whereas Consumer or DeFi projects require AI to accomplish user behavior prediction, strategy generation, incentive distribution, and more.

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

In my opinion, the future of AI is not just about becoming "stronger and more widely used"; the real paradigm shift lies in the reconstruction of collaboration 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 and turn 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 monetize them.

This is a question that is difficult to answer in today's Web2 world, and it reflects some underlying logic I see in the AI + Web3 track: making AI collaborative, transferable, and profit-sharing is the system that is truly worth building.

What I want to talk about today is 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 choice.

My investment logic is not just the narrative of public chains combined with AI, and then seeing which team seems to have a better background and placing a bet.

Investment, in essence, is 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? It's simple, I only believe that projects like this can become disruptors with the emergence of this logic, rather than just "abandoned public chain +1".

In order to find the answer, I basically scanned through all the AI projects I could access until I came across Sahara. The co-founder of Sahara told me: to build an ecosystem that is open, participatory, and where everyone can own and benefit from it.

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, let alone sustain the development of an emerging track.

I fully understand that this path is full of challenges, but that is precisely why it is an irresistible revolution – and the reason why I am firmly investing.

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

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

From Investment to 8x Valuation Follow-up Heavy Investment

If I initially invested in Sahara, it was because what it does aligns perfectly with my vision of the true leading mission of AI—building an AI economy and infrastructure system. The reason I rushed to invest again at an eightfold pre-round valuation within just six months is that I felt a rare strength in this team.

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

Another co-founder, a former investment director at a certain investment institution, responsible for North American investments and incubators, has an undeniable understanding of web3. He is astonishingly disciplined: he sleeps in whole multiples of 1.5 hours, insists on working out no matter how busy he is to maintain his condition, and doesn't touch a single piece of candy to keep his mind clear, working over 13 hours a day. I once joked that he is a robot, to which he simply responded: "I am lucky to have such a busy day today." His source of dopamine is making progress on projects every day; dreaming is his passion, and he needs no other fuel.

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

So when someone says that Sahara gained the favor of capital because of luck, I always unreservedly add, "The pursuit of capital is the inevitable result." I vividly remember how difficult it was to secure primary financing in this round, yet Sahara was being pursued by investors in the primary market.

What everyone remembers is that certain well-known investment institutions have invested in Sahara. Sahara has opened up an investment era for a large technology company entering the Web3 AI field, and winning the company's AI award is an important reason for the investment. In addition, some funds heavily invested in AI, national banks, and so on are also honored guests of Sahara. You can see that a group of institutions that are more focused on traditional technology and industrial resources are quietly betting on AI × Web3 because of Sahara.

Capital will only pay for a certain direction and execution capability - this is a 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:

Over 3.2 million accounts have been activated on the test network, with more than 200,000 data platform annotators (millions in the queue). The clients they serve include several leading enterprises, and they have already achieved 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 already gone deeper and steadier than 99% of "AI Narrative projects".

The Ultimate Challenge of Public Chains: Ensuring Continuous Benefits for All Contributors and Driving Positive Economic Cycles

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 a large amount of labeled data 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 labeling, which not only increases costs but also weakens the value-driven aspect 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, not only is the model quality improved, but it also allows more people to understand and participate in this decentralized AI ecosystem, linking data contribution with benefits, thereby creating a true virtuous cycle.

A typical example is a project on a public blockchain that quickly built a high-quality dataset covering multiple languages and accents by leveraging Sahara's decentralized data collection and human-machine collaborative labeling, significantly improving the training efficiency of its TTS and voice cloning models. This also propelled its open-source project to gain thousands of GitHub stars and over 2 million 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 protecting the rights of all participants—this is the underlying logic driving the explosive growth of the entire ecosystem.

Why is this considered a scenario with long-term value support?

Imagine 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 to 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 ahead of the competition.

More importantly, this is not a one-time transaction. Through Sahara, you are connecting to a potential early user community; and these contributors are 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 ongoing profit sharing, with earnings dynamically linked to usage behavior.

But this is not just a revenue model for data labeling and model training stages. Sahara constructs an economic system that covers the entire lifecycle of AI models, with built-in profit-sharing mechanisms 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 a one-time transaction or buyout.

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

For this reason, I agree with Sahara's underlying belief: a truly healthy AI economic system cannot be just data plundering and model buyouts, nor should it be just a few people reaping all the benefits. It must be open, collaborative, and mutually beneficial—where everyone can participate, and every valuable contribution can be recorded and continuously rewarded 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 overlook 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 very beginning to be open, full-chain, and standardized: it supports deployment on any EVM-compatible chain, and also provides standard API interfaces, allowing 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 lies not 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, Sahara's key 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 land based on its standards remains uncertain.

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

However, it should be noted that these collaborations mainly come from the Web2 world. The long-term development of Sahara is still determined by the maturity and penetration of the entire Web3 AI sector. Sahara benefits from the big trend of Web3 AI, but to truly unleash the value of its infrastructure,