Joe Tsai, Alibaba's chairman, just spent HK$82 million buying 720,000 shares on August 25. The week before, he had done the exact same thing. His CEO, Eddie Wu, matched the gesture with 350,000 shares at an average price of HK$111.6, totaling roughly HK$40 million. Combined, the two executives injected about HK$120 million of their own capital into the company. This is not a rounding error. This is a signal.
But here is the part that should make every protocol engineer sit up straighter: the HK$80 billion rights issue that preceded these purchases was oversubscribed nearly three times. Global sovereign wealth funds and long-term investors fought for the right to fund Alibaba's "full-stack AI capability and AI infrastructure." The market is not just optimistic. It is euphoric. And that is precisely when we need to apply our code audit eyes.
I have spent the last seven years watching capital flow into decentralized systems, and I have learned one uncomfortable truth: the loudest signals of confidence are often the ones that deserve the most scrutiny. When an executive buys shares, we read conviction. When a sovereign wealth fund oversubscribes, we read validation. But what we rarely read is the underlying architecture of the promise. What exactly is Alibaba buying with HK$80 billion? And more importantly, who gets to audit the governance of that spend?
Let me be clear about my position from the start: I am not here to argue that Alibaba's AI strategy is wrong. The company is one of the few entities on the planet with the cloud infrastructure, the data assets, and the distribution channels to make full-stack AI work. Alibaba Cloud already serves millions of enterprises. Tongyi Qianwen, their large language model, is embedded across their ecosystem. The pieces are real. The question is whether the architecture of this investment respects the principles we have spent a decade building in decentralized systems: transparency, community governance, and verifiable accountability.
The first thing to understand is what "full-stack AI" actually means in capital terms.
When Alibaba says it is investing in full-stack AI infrastructure, they are not talking about buying a few thousand GPUs and calling it a day. Full-stack means chips. It means server racks. It means data center cooling systems. It means model training clusters. It means the middleware that connects all of it. It means the application layer that makes it usable. And it means the talent to operate all of it. This is a bet on owning the entire vertical stack, from silicon to software. In blockchain terms, it is like trying to build the entire L1, L2, and application ecosystem with a single treasury. No protocol has ever succeeded at that. The ones that tried collapsed under the weight of their own complexity.
I remember attending the Prague Consensus Workshop back in 2017, where we spent weeks teaching developers that trustless systems are not built by throwing money at a monolithic vision. They are built by creating incentives for independent actors to contribute to a shared infrastructure. The moment one entity controls the entire stack, you have reintroduced the very centralization that blockchain was designed to eliminate. Alibaba is not a blockchain company, of course. But the governance lessons still apply.
The second point is about the nature of the capital itself.
An HK$80 billion rights issue is not venture capital. It is not a Series B. It is a public market instrument that dilutes existing shareholders and commits the company to a specific capital allocation plan. The fact that it was oversubscribed nearly three times tells us that institutional investors believe Alibaba's AI infrastructure will generate returns. But here is the uncomfortable question: who is accountable if it does not? In a decentralized protocol, capital allocation is governed by on-chain voting, and every transaction is auditable. In a public company, capital allocation is governed by a board of directors, and the audit trail is... well, let us just say it is less transparent.
I am not arguing that Alibaba should become a DAO. That would be absurd. But I am arguing that the principles of decentralized governance—transparency, verifiability, and stakeholder participation—can and should inform how we evaluate corporate AI infrastructure investments. The executives are putting their own money in. That is a strong signal of personal conviction. But personal conviction is not the same as accountable governance. When I audited Aave's interest rate model back in 2020, I found that the parameters were completely arbitrary, disconnected from real market supply and demand. The team was genuinely convinced their model was sound. Conviction without accountability is just enthusiasm with a balance sheet.
Let us now look at the technical debt this investment is likely to create.
Full-stack AI infrastructure is not a one-time capital expenditure. It is a continuous operational commitment. Chips need to be refreshed every 18 to 24 months. Models need to be retrained. Data centers need to be maintained. The HK$80 billion will get Alibaba to the starting line, but it will not keep them competitive. The real question is whether Alibaba's operational revenue from AI services can sustain the ongoing capital expenditure required to maintain this infrastructure. Based on my audit experience with cloud providers, the answer is usually no. The initial capex is the appetizer. The opex is the meal.
This creates a familiar pattern. In the DeFi world, we saw protocols raise massive treasuries during bull markets, only to find themselves unable to sustain their operational costs when the market turned. The ones that survived were those with clear revenue models and disciplined treasury management. The ones that failed were those that mistook fundraising for product-market fit. Alibaba has real revenue. That is an advantage. But the scale of this investment suggests they are betting that AI infrastructure will become their dominant revenue stream. That is a bold bet, and it is not guaranteed to pay off.
Here is where I will offer a contrarian angle, because I think it is important to test the prevailing narrative.
The prevailing narrative is that executive stock purchases and oversubscribed rights issues are unequivocal signals of confidence. I want to challenge that. Executive stock purchases can also be a form of communication—a way to signal to the market that the company is stable, that the leadership is committed, that the ship is not sinking. This is especially true in a bull market for AI stocks, where any negative signal could trigger a sell-off. In other words, the purchases might be as much about market management as they are about genuine conviction.
I have seen this pattern in the crypto world. When a founder buys their own token, it is often interpreted as a bullish signal. But sometimes it is simply a liquidity event dressed up as confidence. The token purchase provides the founder with a public narrative while simultaneously supporting the price. The market reads the narrative, and the founder reads the exit. I am not saying Joe Tsai and Eddie Wu are exiting. They are clearly not. But I am saying that we should be careful about conflating public market signals with fundamental value. The signal tells us something. It does not tell us everything.
The second contrarian point is about the opportunity cost. HK$80 billion is an enormous amount of capital. It could have been used to acquire AI startups, to fund open-source research, to build developer ecosystems, to support community-led innovation. Instead, it is being poured into proprietary infrastructure. In the blockchain world, we have learned that the most resilient systems are those that embrace open standards and community contribution. Alibaba's full-stack approach is the opposite. It is a walled garden approach. It might work. But it will only work if Alibaba can build and retain a developer ecosystem that believes in their stack. That is a hard sell when the stack is proprietary.
I am reminded of the "Art & Algorithm" gallery I curated in Prague during the NFT frenzy. We showcased artists who used blockchain for provenance rather than speculation. The artists who thrived were not those who built the most complex proprietary systems. They were those who embraced open standards, collaborated with communities, and built for the long term. The proprietary systems faded. The open ones persisted. I see a parallel here. Alibaba's AI infrastructure might be technically superior. But if it is not built on open principles, it will struggle to attract the ecosystem it needs to sustain it.
Let us talk about the regulatory angle, because it is unavoidable.
Alibaba is no stranger to regulatory scrutiny. The company was fined for anti-competitive practices, and it has had to navigate a complex relationship with Chinese regulators. AI infrastructure investments will only intensify this scrutiny. Data security, algorithm transparency, content moderation, cross-border data transfers—these are not optional considerations. They are core requirements. In the blockchain world, we have spent years advocating for regulatory frameworks that enhance community autonomy rather than hinder it. The same principle should apply to AI infrastructure. Alibaba has the resources to lead in this area. The question is whether they will choose to do so.
I had the privilege of advising the EU regulatory task force in 2025 on decentralized governance guidelines. The biggest lesson from that experience was that regulators are not the enemy of innovation. They are the gatekeepers of trust. If Alibaba can build AI infrastructure that is transparent, auditable, and compliant, they will have a competitive advantage that no amount of capex can buy. If they treat regulation as an obstacle to be overcome, they will spend the next decade fighting fires instead of building value.
Now, I want to bring this back to the core principle that guides my analysis.
Build for humans, not just nodes. This is not a slogan. It is a design principle. When we build systems, we are not just building technology. We are building relationships, communities, and trust. Alibaba's AI infrastructure investment is a massive bet on technology. But the real value will come from how that technology serves people. Will it empower small businesses to compete with giants? Will it enable researchers to accelerate discovery? Will it give ordinary users control over their data? Or will it simply concentrate more power in the hands of a few corporations?
I do not have the answers to these questions. But I do know that the answers will determine whether this HK$80 billion investment creates lasting value or becomes another monument to technological hubris. The executives are betting their own money. The sovereign wealth funds are betting their mandates. The market is betting on future returns. But the real stakeholders—the users, the developers, the communities—are not at the table. That is the decentralization dilemma, and it is not going away.
Education is the ultimate yield. I have said this for years, and I believe it more strongly now than ever. The best investment Alibaba can make with its HK$80 billion is not in chips or data centers. It is in educating the next generation of developers, users, and regulators about what AI infrastructure can and should be. That is the investment that compounds. That is the investment that builds trust. That is the investment that ensures the technology serves humanity rather than the other way around.
So here is my forward-looking judgment: Alibaba's AI pivot will be remembered either as a masterstroke or as a cautionary tale. The distinction will not be determined by the size of the investment or the conviction of the executives. It will be determined by the governance structures they build, the openness of their ecosystem, and their willingness to engage with the communities they claim to serve. The market is giving them the benefit of the doubt. The rest of us should be paying attention.
The question I leave you with is this: when the next bull market comes, and the next massive infrastructure investment is announced, will we be better equipped to ask the right questions? Will we demand transparency? Will we insist on accountability? Or will we be seduced by the numbers, the names, and the promises? The choice is ours. And it always has been.