The Limits of Artificial Intelligence

---

• AI is advancing rapidly, but turning ideas into real-world systems is more challenging than most people expect.
• Energy constraints, physical environments, and weak infrastructure slow adoption far more than software innovation alone.
• This blog explains why AI progress may be slower and more uneven than headlines suggest, and where realistic opportunities are likely to appear.

The most important question about AI might not be what it can do, but what stands in the way of scaling it reliably in the real world.

Artificial intelligence feels omnipresent as new tools appear weekly and productivity demos go viral. Trillions of dollars in market value are now tied to the promise that machines will soon think, reason, and act at human levels. That confidence masks an uncomfortable truth.

The story of AI right now is less about capability and more about constraint. Less about what is possible in theory and more about what is difficult in practice. Less about intelligence itself and more about the messy systems intelligence depends on.

History suggests this matters. Every transformative technology looks inevitable in hindsight, but halting and fragile while it’s unfolding. Railroads needed steel, electricity needed grids, and the internet needed protocols and trust. None arrived fully formed.

Artificial general intelligence will probably be no different. An edge will belong not to those who assume seamless progress, but to those who understand the bottlenecks that shape it.

The Last Economy Built on Labor Scarcity

Every major economic era has revolved around scarcity: land, capital, labor, and information. The industrial age reorganized labor, and the digital age amplified it. But both assumed something fundamental: human effort remained limited and valuable.

AI challenges that assumption. In theory, once intelligence becomes cheap and scalable, labor stops being scarce. Tasks that once required training, time, and coordination become abundant, an idea fuels the boldest forecasts. It also explains the unease beneath them.

If labor is no longer scarce, the rules of wages, careers, and value creation must change. But this shift doesn’t happen overnight. It happens only when intelligence is reliable, affordable, and deployable at scale. That is where reality intrudes.

Constraint #1: Energy Is the Real Bottleneck

Intelligence runs on power, not ideas. AI feels digital, weightless, and infinite, but it’s none of those things. Every query requires computation, every computation requires electricity, and every scale-up magnifies that demand.

The arithmetic is simple: Advanced AI systems require enormous and stable power supplies. Data centers already strain regional grids. New capacity takes years to plan, permit, and build. Efficiency gains help, but they do not eliminate the underlying need for more energy.

This is not a software problem. It’s an infrastructure problem.

History shows that infrastructure rarely moves at the pace of ambition. It moves at the pace of physics, regulation, and capital cycles. Until energy becomes more abundant, reliable, and resilient, AI’s theoretical capabilities remain economically constrained.

This creates an inversion most investors overlook. The opportunity is not only in smarter algorithms. It’s in the less glamorous systems that make intelligence usable at scale.

Constraint #2: The Physical World Does Not Run on Code

Digital tasks are easy to automate because they live in clean environments. Physical tasks do not. Factories, farms, construction sites, and supply chains are filled with edge cases: Objects fall, materials vary, conditions change, humans adapt instinctively, and machines struggle.

This is not a temporary limitation. It’s a structural one. Even more limiting is the state of data. Many foundational industries operate on fragmented, incomplete, or unusable information. Automation requires clean inputs before it can produce clean outputs and, without that foundation, intelligence has nothing to learn from.

The result is a paradox. The sectors that dominate employment and economic activity are typically the least prepared for full automation. They are also the most essential.

This suggests a fairly long transition period where human skill, judgment, and adaptability remain central. The future may be automated, but the path there is uneven, slow, and full of friction.

Constraint #3: Intelligence Without Coordination Creates Noise

This is why smarter systems can still fail. Early AI tools feel powerful in isolation. The problems emerge when they must work together. Memory breaks across platforms. Context disappears between systems. Trust erodes when outputs conflict. Workflows become clunky instead of seamless.

Coordination is the hidden cost of intelligence. It’s also the least solved.

In virtually every technological shift, standards matter more than breakthroughs. That includes rail gauges, electrical frequencies, and internet protocols. Without shared rules, progress fragments.

AI lacks its equivalent of a universal adapter. Until systems can reliably communicate, remember, and align, intelligence remains impressive but inefficient. This matters because economic value comes not from isolated brilliance, but from integration. Businesses run on workflows and real, long-term cash flow, not demos.

Why This Feels Like a Bubble and Why It Is Not

Both things can be true. Periods of hype often coincide with real transformation. Excess capital chases possibility, expectations overshoot reality, and disappointment follows.

That does not invalidate the underlying shift. It resets it. The likely path forward is not a straight line. It’s a cycle as investment surges, constraints surface, capital retreats, infrastructure catches up, and progress resumes.

Those who confuse volatility with failure miss the pattern. Those who mistake inevitability for immediacy make a different error. Another edge belongs to patience disguised as skepticism.

A Glimpse of the End State

This is why human traits may become more valuable, not less. If intelligence becomes abundant, what remains scarce? Trust, meaning, creativity, and belonging.

When outputs are automated, authenticity gains value. When information is infinite, discernment matters. When efficiency dominates, human connection stands out. This does not mean utopia. It means differentiation.

Some people will opt out, seeking autonomy and self-sufficiency. Others will double down on expression, art, leadership, and embodied experience. Still others will struggle with dislocation, searching for purpose in a system that no longer rewards them as it once did.

These are tendencies. They are already visible in small ways. The lesson is subtle but important: Technology does not eliminate human value. It simply rearranges it.

The Opportunities Hidden in the Transition

Most people look for breakthroughs. History rewards those who fix bottlenecks, such as energy infrastructure, grid resilience, data normalization, skilled trade enablement, human coordination tools, trust systems, and real world logistics.

These are not moonshots. They are foundations. They lack the romance of general intelligence, but they enable it. They also persist regardless of timelines. Even if the future arrives slower than expected, these investments remain useful. That is the essence of sound risk management: Favor paths where multiple futures still pay off.

The Long View

The greatest investment mistakes tend to emerge from perceived certainty. The greatest opportunities come from understanding limits and humility. 

AI will reshape the world, but not all at once, not evenly, and not without resistance. Progress will be gated by energy, data, coordination, and human adaptation. Those constraints will define who benefits, when, and how. The future is not written by intelligence alone. It’s written by the systems that support it and the people who learn to live alongside it.

A key risk is not that machines become too powerful. It’s that we misunderstand the pace and shape of the change. And history is unforgiving to those who confuse possibility with inevitability.

---

This is based on an episode of Top Traders Unplugged, a bi-weekly podcast with the most interesting and experienced investors, economists, traders and thought leaders in the world. Sign up for our Newsletteror subscribe on your preferred podcast platformso that you don't miss out on future episodes.