The news arrived with the familiar rhythm of bullish consensus: Citizens Financial Group raised its price target on Alphabet to $515, citing AI infrastructure growth. The stock ticked up. Twitter cheered. The narrative was clean—big tech investing billions into compute is good for innovation, good for earnings, good for the market. But I spent years auditing smart contracts in 2017, watching ICOs promise decentralised futures while building centralised backdoors. I learned that the cleanest narratives often hide the deepest rot. And this Alphabet upgrade, wrapped in the glossy language of progress, is actually a warning siren for anyone who believes blockchain’s promise of sovereignty.
Let me be clear: I am not anti-AI. I am not anti-Google. I am anti-false choice. The current frenzy frames AI infrastructure as a binary bet—either you back the hyperscalers or you fall behind. But the real story is subtler. Alphabet’s TPU clusters, its Google Cloud revenue growing 35% year-over-year, its $50+ billion annual capex—all of it is being used to justify a valuation that assumes centralisation is the only path forward. For those of us who have watched the ICO bubble, the DeFi summer, and the Terra collapse, this feels hauntingly familiar: a narrative so dominant that it crowds out the very alternatives that could save us.
Citizens’ target price is not wrong on the numbers. Alphabet’s AI business is real. Google Cloud posted $11.4 billion in revenue last quarter, and AI-related products like Vertex AI are accelerating. But the analysis misses the one variable that matters most: the fragility of centralised trust. When a single entity controls the compute, the data, and the model weights, it creates a single point of failure that no audit committee can fix. I learned this in 2020 when I built OpenLedger Lab, mentoring 50 developers from underrepresented backgrounds. We launched tokens, formed DAOs, celebrated democratised finance. Then the Terra collapse shattered my idealisation of algorithmic stability. I retreated to a cabin in Virginia for six weeks, disconnected from all screens, and wrote the manuscript for The Soul of Sovereignty. The lesson was visceral: any system that concentrates power, even with the best intentions, eventually betrays the trust placed in it.
Alphabet’s AI infrastructure is no different. Its TPU v5e and v5p chips are impressive—I’ve spent hours studying their architecture, the systolic array designs, the bandwidth optimisation. But the software stack that makes them sing is entirely proprietary. Developers who build on Vertex AI are not building on open protocols; they are renting a walled garden. The same goes for Gemini models: tight integration with Google’s search, Android, and Chrome creates convenience, but it also creates dependency. In crypto, we call this “vendor lock-in.” In traditional tech, they call it “ecosystem growth.” The difference is semantic, not structural.
The contrarian angle here is not that Alphabet will fail—it likely won’t, at least not in the short term. The contrarian truth is that the AI infrastructure hype is a perfect storm for centralising power in ways that will make today’s internet monopolies look tame. When I audited the Tezos mainnet in 2017, I found 14 critical vulnerabilities in the consensus mechanism. I wrote a white paper titled “Code is Law, But Only If It Compiles.” The core insight was that decentralisation is not a feature—it is a constraint. It forces transparency. It forces redundancy. It forces consent. Centralised AI infrastructure, by contrast, optimises for speed and efficiency at the expense of these values. The $515 target price is built on assumptions that treat efficiency as the only metric worth tracking.
Let me drill into the numbers. Alphabet’s capex is projected to exceed $50 billion in 2024, the bulk of it going to data centres and TPU clusters. The analysis report I read notes that Alphabet is NVIDIA’s third-largest customer, after Microsoft and Meta. That means the same GPU shortage that plagued the crypto mining industry in 2021 is now being replicated for AI—except the buyers are trillion-dollar corporations. The report also highlights that Alphabet’s Google Cloud remains marginally profitable, with operating income of less than $2 billion on $11.4 billion in revenue. That’s a sub-20% margin, far below its advertising business (which prints 35%+ margins). What the report does not ask is: what happens when AI infrastructure investment peaks and the returns disappoint? Will Alphabet cut its capex, leaving developers who built on Vertex AI stranded? Or will it double down, using its advertising cash cow to subsidise cloud prices and drive competitors out of business? Both scenarios lead to the same outcome: centralisation.
I am not arguing that decentralised compute networks are ready to replace Alphabet. The projects I follow—Render Network, Akash Network, io.net—are still early. Their total compute capacity is a fraction of a single Google data centre. But the architecture matters more than the scale. Decentralised compute networks distribute trust across thousands of independent node operators, each running standard hardware. They use on-chain proofs to verify work, they pay out in tokens, and they allow anyone with a GPU to participate. The latency is higher, the coordination is messier, and the developer experience is rougher. Yet they offer something Alphabet cannot: sovereignty. When a startup trains its model on Akash, it owns the model, the data, and the deployment pipeline. There is no risk of a terms-of-service change, no fear of a product sunset, no central entity that can decide its fate.
This is the battle that the $515 target price ignores. It is not about computing power; it is about who holds the keys. Alphabet’s infrastructure is a cathedral: beautiful, efficient, but built on the assumption that a few architects know best. Decentralised compute is a bazaar: chaotic, noisy, but resilient because no single merchant controls the market. I have seen this pattern before. In 2018, I turned down advisory roles for ICOs that were building “decentralised” products on centralised infrastructure. They paid well—six-figure tokens—but I refused because the ethics did not align. I spent six months auditing Tezos instead, because its on-chain governance model, flawed as it was, at least attempted to distribute power. That decision cost me financially, but it gave me a integrity I could not have bought.
Now, in 2025, I run a crypto education platform. I watch students flock to courses on “AI and Blockchain” without understanding the underlying tension. They think combining two buzzwords will create value. They do not see that Alphabet’s AI infrastructure is designed to extract value, not distribute it. The $515 target price is not just a number; it is a bet that centralisation will win. Every dollar invested in Google Cloud is a dollar not invested in decentralised alternatives. Every developer who builds on Vertex AI is a developer who learns the habits of centralisation. And every bullish analyst report reinforces the idea that the future belongs to the hyperscalers.
But here is the rub: the hyperscalers themselves are terrified of decentralisation. That is why Alphabet has invested in blockchain companies, why it launched its own blockchain node service, why it sits on the board of the Linux Foundation’s AI project. They know that the ultimate threat to their business model is not a competitor with a bigger data centre—it is a protocol that renders data centres obsolete. They are trying to co-opt the movement, the same way banks co-opted the fintech revolution by partnering with startups instead of fighting them. The risk is that the crypto community, desperate for mainstream adoption, will embrace these partnerships as validation rather than recognising them as co-option.
I saw this happen during the 2024 Bitcoin ETF approval. I published an op-ed titled “Institutionalization vs. Ideology” in a major tech journal, arguing that the ETF structure centralised custody in the hands of a few Wall Street firms. I pointed out that 95% of ETF Bitcoin was held by three custodians. The response was overwhelming—thousands of emails from people who felt their silent doubts were finally voiced. But the market continued to celebrate the ETF. Price action overwhelmed principles. The same dynamic is playing out with AI infrastructure: investors are cheering Alphabet’s growth without asking who controls the infrastructure and what that means for decentralisation.
The analysis report I studied admits that Alphabet’s AI infrastructure is “increasingly being used to justify a valuation that assumes centralisation is the only path forward.” It also notes that the true risk is not technology but “organisational risk” and “anti-trust litigation.” These are real. Alphabet’s search monopoly case is winding through the courts, and a breakup could cripple its ability to subsidise AI cloud. But the report never asks the more fundamental question: what kind of AI infrastructure do we want to build? Do we want a world where three companies control the compute, the models, and the data? Or do we want a world where anyone can contribute a GPU, train a model, and deploy it without permission?
I am an idealist, I know. My INFJ personality makes me prone to seeing the meaning beneath the surface. But idealism without rigour is just naivety. I spent three months in 2022 recovering from burnout by writing The Soul of Sovereignty, and during that solitude I developed a framework for evaluating any technology: does it increase or decrease the number of people who have meaningful agency? Alphabet’s AI infrastructure decreases agency. It concentrates power. It makes developers dependent. It creates a digital feudalism where the landlord collects rent on every interaction. The $515 target price is the rent cheque.
Crypto’s role, if it is to be true to its founding principles, is to build the alternative. Not to compete head-on with Alphabet—that is a losing battle. But to carve out a parallel infrastructure that serves those who value sovereignty over convenience. The early signs are promising. Decentralised compute networks are growing. ZK-proofs are making it possible to verify AI inferences without revealing data. Token incentives are attracting GPU providers from around the world. But these projects need capital, talent, and attention—all of which are flowing toward Alphabet’s cathedral.
So here is my takeaway: the next time you see a target price upgrade on a big tech stock citing AI infrastructure, ask yourself what is being left out. The numbers may be right, but the framing is a choice. The choice to ignore the ethical cost of centralisation. The choice to assume that efficiency is the only virtue. The choice to bet on the bazaar instead of the cathedral. I know which side I am on. Truth is immutable, unlike the price action. The $515 target will come and go, but the architecture of power will shape our digital future for decades. Let us build the one we can trust.