I first heard the figure during a closed-door session with a Sydney-based sovereign wealth fund manager. 'Ten trillion dollars,' he said, his voice flat, as if reading a line from a script. 'That's what Morgan Stanley's CEO is telling clients we'll spend on AI infrastructure over the next decade.' He paused, looking at me. 'You're in crypto. What does that mean for us?'
I didn't answer immediately. In my twenty-nine years watching technology markets, I've learned that the biggest numbers are often the emptiest. They fill a room, but they don't heal it. The $10 trillion prediction from James Gorman, CEO of Morgan Stanley, is less a financial forecast than a strategic narrative—a signal designed to shape capital flows, not reflect them. But as someone who built a career on decoding the ethical architecture of trust, I see something else beneath the surface: a profound silence about the systemic rot this investment will create.
Let's start with context. The prediction, reported by Bloomberg earlier this week, posits that global AI capital expenditure—spanning GPUs, data centers, energy grids, and cooling infrastructure—will reach $10 trillion over the next decade. That's roughly the combined annual GDP of Germany, Japan, and the United Kingdom. It's a number so large it defies intuition. Yet the market has already reacted: NVIDIA shares rose 3% on the news, and cloud provider stocks followed. Morgan Stanley, of course, is both the messenger and the beneficiary. As one of Wall Street's largest underwriters of technology debt, they have every incentive to inflate expectations.
But the real story isn't the dollar amount. It's what this capital will buy—and what it will destroy.
The code compiles, but does it heal?
From my experience auditing tokenized asset frameworks for the Australian Securities Investment Commission, I've learned that infrastructure spending never exists in isolation. Every dollar allocated to centralized compute is a dollar diverted from decentralized alternatives. The $10 trillion figure, if realized, would lock the world into a model where AI intelligence is owned by three or four hyper-scalers—Microsoft, Amazon, Google, and perhaps Meta. This isn't speculation; it's geometry. Only these entities have the balance sheets to absorb such long-term, low-return capital commitments.
The technical implications for blockchain are severe. Layer2 rollups, for instance, depend on sequencer nodes that maintain order of transactions. Today, most sequencers are run by a single entity—the same infrastructural monoculture that $10 trillion would entrench. We've been told for two years that 'decentralized sequencing' is coming. But when capital gravitates toward centralization, the incentive to build alternatives evaporates. The industry will face a choice: either accept that AI infrastructure will be a monopoly, or redirect capital toward verifiable, decentralized compute networks like those emerging from the Ethereum community's work on zk-rollups and shared sequencers.
I've seen this pattern before. In 2021, during the NFT gaming boom, traditional game publishers rushed to adopt blockchain because they believed it would let them mint infinite gear and milk players. They missed the point. The real barrier isn't technology—it's the ethical premise that players should own their assets. Similarly, the $10 trillion AI narrative ignores the ethical premise that compute should be accessible, not extractive.
Feminine wisdom asks not 'how fast can we train?' but 'who gets to set the loss function?'
This is where my contrarian angle emerges. The most revealing aspect of the Morgan Stanley prediction isn't its size, but its unexamined assumption: that AI development will continue to follow scaling laws—bigger models, more data, more compute. What if the next breakthrough comes from efficiency, not scale? What if Mamba, or state-space models, or neuromorphic chips achieve similar performance with 100x less energy? Then $10 trillion becomes stranded capital, and the investors who bet on it are left holding empty server racks.
I've spent the last year facilitating a digital salon called 'Conscious Algorithms,' where philosophers, AI ethicists, and blockchain developers discuss the soul of autonomous agents. One recurring theme: the most overlooked variable in AI infrastructure is utilization. Data centers today run at 20–30% average utilization. That's not an engineering problem—it's a design problem. We're building for peak demand, not sustained need. The $10 trillion forecast assumes we will find applications that consume every last flop. But if history teaches us anything, it's that the actual utility of general-purpose compute rarely matches the hype.
Trust is not encrypted; it is woven.
This brings me to the ethical dimension. As I wrote in my 2017 manifesto 'The Moral Architecture of Trust,' the concentration of computational power is a species-level risk. $10 trillion of centralized AI infrastructure would give three or four corporations the ability to define truth, influence elections, and automate labor decisions for billions. The blockchain community has spent years arguing about TPS and finality—but this is the real finality. If we don't build decentralized alternatives now, the window closes.
I've seen what happens when capital overwhelms conscience. In the wake of Terra/Luna's collapse in 2022, I withdrew from social media for six weeks to document the trauma of retail investors. The silence was deafening. The same silence now surrounds the $10 trillion prediction. No one is asking the hard questions: Who will own this compute? Will they be accountable to anyone? And if the system fails, who absorbs the loss?
Silence is the loudest indicator of systemic rot.
From my time working with women in blockchain through 'Women of the Chain,' I've learned that inclusive structures build resilience. Homogeneous decision-making—whether in a boardroom or a compute cluster—produces fragile systems. The $10 trillion forecast assumes a monoculture of thought: that scaling is the only path, that centralization is inevitable, that capital should flow north. But I've seen the power of small, diverse networks. The 30 women I mentored in 2023 are now building compliance tools, educational platforms, and local DeFi solutions. They don't need $10 trillion. They need access, fairness, and a level playing field.
Takeaway: The choice is ours.
The Morgan Stanley prediction is a mirror reflecting our collective priorities. If we accept it as destiny, we build a world where AI is owned by the few, and the many pay the price. But if we read it as a warning—a signal that the current trajectory is unsustainable—we have a chance to redirect capital toward infrastructure that heals rather than extracts. The code compiles either way. The question is whether it will also heal.
I'll leave you with this: in 2024, I contributed to ASIC's ethical governance guidelines for tokenized assets. The regulators asked me to define 'fairness.' I told them it's not a line of code. It's a practice. The same applies to AI. The $10 trillion will be spent—by someone. The only question is whether we'll spend it on building walls, or on weaving trust.
