The Capital Stack Gets a Second Layer
The AI build-out has a new running mate, and it is not what the consensus expected.
Goldman Sachs projected last week that AI-related capital expenditure will hit $905 billion in 2027 — a figure that rewrites the scale of infrastructure investment globally and makes crypto look, by comparison, like a side project. Around the same time, Arthur Hayes — a figure the crypto market treats with the uncritical reverence it once reserved for equity analysts — predicted that the HYPE token will reach $150. The market, by instinct, filed one story under "the future of money" and the other under "the future of everything else." That classification is increasingly incoherent.
Hong Kong, meanwhile, is building the actual plumbing. The city-state's digital asset expansion — tokenization infrastructure, licensed crypto products, a regulatory on-ramp designed for institutional participation — is not a crypto play in the narrative sense. It is a capital formation play. The intent is to make digital asset issuance as frictionless as bond issuance, to capture the settlement layer that sits between tokenized real-world assets and on-chain liquidity. If it works, Hong Kong becomes the Singapore of the compliant digital asset stack — and the $905 billion flowing into AI servers goes looking for yield on a platform that already speaks the same technical language.
The mainstream framing treats AI and crypto as a competition for institutional mindshare. In that framing, AI wins: the capital numbers are bigger, the enterprise customers are named and credible, the regulatory uncertainty is lower. Crypto, the story goes, is still waiting for its institutional moment, still cycling through scandals, still arguing about the fundamental nature of what it actually is.
That reading is not wrong. It is just incomplete.
The AI stack is being built in parallel, not instead of
Goldman Sachs's $905 billion figure is real. The hyperscaler capex cycle — Microsoft, Google, Amazon, Meta — is not a speculative bet on AI adoption; it is a committed infrastructure build targeting inference and training capacity at a scale that is genuinely without historical parallel outside wartime industrial mobilization. That money is going into data centers, custom silicon, power infrastructure, and the connective tissue of a compute economy. It is not going into crypto mining, and it is not competing with crypto for the same dollar.
What the Goldman forecast misses is the downstream effect on the capital environment. When $905 billion gets deployed into digital infrastructure — power grids, cooling systems, networking fabric, physical real estate — it creates an ecosystem of counterparties, service providers, and derivative financial products that themselves need capital allocation. Some of that allocation will flow into tokenized infrastructure debt, synthetic asset wrappers, and on-chain settlement instruments. Not because the AI builders chose crypto, but because the infrastructure they are building uses the same technical substrate.
Hong Kong's push to become the compliance-first hub for tokenized products is the structural answer to that downstream question. The city has spent the past two years building the regulatory scaffolding — VASP licensing, stablecoin frameworks, tokenized bond pilots — specifically to capture exactly this kind of secondary capital flow. The AI capex is the event; the tokenization infrastructure is the consequence.
Hayes is describing a convergence, not a coin
Arthur Hayes predicted HYPE at $150. The prediction is specific; the mechanism is vague. What would push HYPE from current levels toward three figures is the question that matters more than the target.
Three possible answers: a sustained token burn and fee-revenue model that makes the token genuinely productive; a wave of institutional participation via compliant on-ramps, converting the token into a settlement instrument; or a broader market narrative that treats HYPE as the leading proxy for Ethereum-layer activity and prices it accordingly. All three are plausible. None is guaranteed.
What is guaranteed is the direction of infrastructure. Whether HYPE hits $150 or $15, the tokenization build-out in Hong Kong continues. The compliance framework does not evaporate if a single token underperforms. And that continuity is the more interesting signal — Hayes is making a market call; Hong Kong is making a structural bet. Market calls are fun to watch. Structural bets determine who ends up with the revenue.
The crypto market's tendency to treat price predictions from named personalities as news — to promote them, to argue about them, to treat them as if they contain structural information they often lack — is itself a symptom of a market that has not fully integrated institutional discipline. Goldman Sachs does not announce that AI capex will hit $905 billion and then leave the number without a model. Hayes announced $150 for HYPE and left the mechanism implied. That asymmetry is not a criticism of Hayes specifically; it is an observation about the maturity gap between the two markets.
The convergence is the story
Here is the structural reality that neither the AI bulls nor the crypto bulls want to acknowledge: they are building the same stack.
AI infrastructure requires compute, power, cooling, and networking. Digital asset infrastructure requires consensus mechanisms, settlement layers, compliance verification, and programmable value transfer. Both are digital-first, both require physical data center capacity, both generate revenue from the margin between input cost and output value, and both are creating a new class of financial instruments that did not exist five years ago. AI infrastructure debt will be tokenized. Tokenized real-world assets will use AI for risk assessment. The two markets are not competitors; they are adjacent layers of a capital infrastructure that is still being assembled.
Hong Kong is positioning itself at the intersection. The city's regulatory clarity — and it is genuine regulatory clarity, not just permissive ambiguity — gives it a structural advantage over jurisdictions that are still arguing about whether digital assets are securities or commodities or something else entirely. The SFC's licensing regime, the e-HKD pilot programs, the tokenized green bond issuances: these are not gestures. They are the foundation of a financial architecture that expects to interface with AI-generated capital flows within the next decade.
The competitive question is not whether crypto or AI wins. It is which jurisdiction captures the settlement layer that connects the two.
The stakes
Goldman Sachs's $905 billion AI capex figure is a bet on the United States and its allied supply chain — NVIDIA,台积电, the hyperscalers, the utilities. It is a bet that the compute economy remains Western-led and that the infrastructure build-out is a domestic industrial policy success story. That bet is not wrong.
Hong Kong's tokenization push is a bet on a different axis: that the capital that flows through AI infrastructure needs a settlement and issuance layer, and that the jurisdiction which builds that layer first captures the margin on every digital asset transaction that follows. That is a narrower bet, but a potentially more durable one. Infrastructure outlasts narratives. The e-HKD and the compliant VASP regime will still be there when the AI capex cycle decelerates.
Hayes predicted $150 for HYPE. Whether he is right depends on factors that even Hayes would acknowledge are not fully within his control. What is within the control of financial architecture builders — the SFC, the Monetary Authority of Singapore, the European Central Bank's digital euro team — is the settlement layer that makes those predictions meaningful. Build the layer and the asset follows. Build the narrative without the layer and you get another crypto cycle: loud, profitable for some, structurally fragile.
The $905 billion tells you where the money is going. Hong Kong tells you where it ends up. The two stories are not competing. They are the same story, told from different layers of the same stack.
Monexus covered the Goldman Sachs AI capex projection as a financial markets development; the wire treated it as infrastructure news. The Hong Kong digital asset expansion received modest play in the business sections of most English-language wires — less than the FX or rate coverage that accompanied it. That asymmetry is worth noting: the institutional build-out that will interact most directly with AI capex dollars received a fraction of the coverage of the capex announcement itself. This desk flagged the convergence.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- https://t.me/Cointelegraph/13942
- https://t.me/Cointelegraph/13940
- https://t.me/Cointelegraph/13930