AI × Web3: Who will build the infrastructure for this era?

When the paradigm of technology truly changes, we often first see the hype rather than the system. The wave of AI that we are currently experiencing is no different.

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

In the past year, I have seen a large number of projects in RWA, consumer applications, information finance, and many others - they are undoubtedly exploring the intersection between the real world and on-chain systems. However, an increasingly obvious trend is that no matter which path the projects take, they will ultimately have to enter the collaborative logic of AI, using AI to enhance competitiveness and efficiency.

For example, in RWA, thinking about how to use AI for risk control optimization, off-chain data verification, and dynamic pricing is the future direction; or consumer applications or DeFi that urgently need excellent user experience also require AI to accomplish user behavior prediction, strategy generation, incentive distribution, etc. Other tracks in different directions also have similar demands.

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 have 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 and thereby giving birth to an entire open ecosystem.

AI is also on this path: Agents will become everyone's intelligent co-creation body, helping you turn expertise, creativity, and tasks into automated productivity tools, and even achieve monetization.

This is a question that is difficult to answer in today's Web2 world, and it reflects some of the underlying logic I see in the AI + Web3 space: 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 operations of AI from the chain-level structure: Sahara.

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

My investment logic is not simply to add the narrative of public chains with AI and then bet on whichever team's background looks better.

Investment, at its core, is a choice of worldview, and I have always been asking a fundamental question: Can the future of AI be owned by more people together?

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 continuously benefit? It's simple, I only believe that such projects have the potential to be disruptors if this logic emerges, rather than being just "abandoned public chain +1".

To find the answer, I basically scanned all the AI projects I could access until I encountered Sahara. The response from Sahara's co-founder was: to build an open, participatory ecosystem that everyone can own and benefit from.

This statement is simple, but it precisely hits the Achilles' heel of traditional public chains: they often serve developers in a one-sided manner, and the design of token economies 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 firm investment.

As I emphasized in my previous article discussing 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 predictions 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 to be the true leading mission of AI – building an AI economy and infrastructure system. Then let me say that within just six months, I was eager to follow up with an investment at eight times the previous round's valuation because 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 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. Having known this professor for over a year, I have witnessed what it means to work for more than ten hours a day, with stable emotions and humility.

Another co-founder, who previously served as the head of North American investments and incubators at an investment firm, undoubtedly has a deep understanding of web3. He is astonishingly self-disciplined: he sleeps in multiples of 1.5 hours, insists on working out no matter how busy he is to maintain his state, and refrains from any candy to keep his mind clear, working over 13 hours a day. I once joked that he was a robot, and he simply replied: "I am lucky to have the busy life I have today." His source of dopamine comes from advancing project progress every day; dreaming big is his passion, and he doesn't need any other fuel.

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

So when someone says that Sahara gained the favor of capital due to luck, I always unreservedly add, "The capital's pursuit is an inevitable result." I vividly remember how difficult it was for primary market financing in this round, yet Sahara was being chased for investment by the primary market.

What everyone should remember is that certain well-known investment institutions have invested in Sahara. Sahara has opened the investment era for a certain tech giant's entry into the Web3 AI field, and winning the company's AI award is an important reason for the investment. In addition, some heavily invested AI funds, national banks, and others are also guests of Sahara. What you can see is a group of institutions that are more focused on traditional technology and industrial resources, starting to quietly bet on AI × Web3 because of Sahara.

Capital will only pay for a certain direction and execution ability - this is the 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 are in queue). Their clients include several leading technology companies, and they have already achieved revenue at the million-dollar level.

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

The Ultimate Challenge of Public Blockchains: 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 interactive 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 diminishes the value-driven nature of community co-creation.

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 improves the quality of the model but also allows more people to understand and participate in this decentralized AI ecosystem, linking data contribution with benefits to form 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 annotation, 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 annotation also received token rewards distributed 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 precisely 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 would 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 soon 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 the actual 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 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 invocation, combination, and cross-chain reuse after the model goes live, allowing value to be captured over a longer period.

Model developers, optimizers, validators, compute power contribution nodes, etc. 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, like building blocks, can be repeatedly called and combined by different applications, and each call creates new revenue for the original contributor.

For this reason, I agree with Sahara's underlying belief: a truly healthy AI economic system cannot simply be about the plundering of data and the buyout of models; it cannot just benefit a few people. Instead, it must be open, collaborative, and mutually beneficial—where everyone can participate, every valuable contribution can be recorded, and rewards can continue to be gained in the future.

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

While I am optimistic about Sahara, I will not cover up the challenges the project will face due to my own investment position.

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

Its system is designed from the beginning 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 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, 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: serving multiple leading tech companies,