Every technology revolution offers a brief window of unreasonable advantage—an era when the bold can seize opportunities that later become ordinary. This is that moment for AI. The cost of cognition is collapsing. The rules haven’t caught up. The field is open to anyone ambitious enough to rewire their work, their organization, or their industry around machine intelligence. But none of these conditions will hold for long.
Like the early internet before paywalls and advertising, or the dawn of ride-sharing before prices rose to match reality, we are living through a period of unsustainable abundance. And history is clear: free rides don’t last.
This is the best of all possible times to embrace AI, and the worst time to hesitate.
1. AI Is Still an Unfair Advantage
AI has already entered the mainstream, but mastery remains uneven. Most people are still using it for trivial tasks—writing professional sounding emails, summarizing meeting notes, or using their chatbot as a digital therapist.
Those who know how to truly leverage AI are operating in another universe entirely. When you train ChatGPT Pulse on your own writing, reports, and meeting transcripts, it’s like waking up each morning to a team of researchers who have curated your private edition of the Harvard Business Review.
Across industries, the same pattern is emerging. Marketing teams are deploying AI agents that can generate, test, and localize thousands of campaign variations overnight—then feed the winning ideas directly into production systems. Supply-chain leaders are using autonomous forecasters that continuously adjust logistics based on weather, pricing, and geopolitical data, eliminating weeks of human analysis. Product teams are training domain-specific copilots that surface customer insights, simulate competitive dynamics, and recommend strategic trade-offs before a meeting even begins.
What once required committees, consultants, and coordination now happens as a real-time conversation between leaders and their digital counterparts—an entire cognitive layer of the enterprise that never sleeps, learns continuously, and compounds advantage for those who know how to direct it.
Right now, there’s still enormous alpha to be captured simply by being better at using AI than your peers. A mid-level consultant, analyst, or creative can plug into these tools and achieve leverage that once required an entire team. They can produce better insights, faster outputs, and more refined products—and sell them at a premium—because the knowledge and implementation gap remains wide.
But that window will close. As AI literacy spreads and models become embedded directly into platforms and workflows, the arbitrage disappears. To add value, you’ll need either deep engineering skill or very specific domain expertise. Everyone else will simply go straight to their own AI agents to get the job done.
2. Someone Else Is Paying for Your Intelligence
AI is absurdly, unsustainably cheap.
For $20 a month—or even for free—you can access sophisticated systems capable of reasoning, summarizing, coding, and designing at levels that would have required teams of specialists and millions in infrastructure only a few years ago.
But remember Uber’s early days, when rides were subsidized to be cheaper than public transport? Or Spotify, when streaming felt like a miracle before licensing costs caught up? Eventually, reality asserted itself. The same will happen here.
AI’s current economics are a gift. OpenAI, Anthropic, and others are still losing money on every query, propped up by investor subsidies and growth ambitions. Someone is paying the bill - just not you.
The energy, compute, and data costs of cognition will ultimately find equilibrium. Prices will rise, or access will be constrained. We are in the VC-subsidized golden age of cheap intelligence—a temporary arbitrage between technological possibility and economic gravity.
If you are a leader, this is your chance to build capability at minimal cost. Soon, you’ll be paying full price for the same cognitive horsepower.
3. Rulemakers Haven’t Caught Up
Regulators are still asleep at the wheel. AI currently sits in a grey zone: too new to be tightly controlled, too powerful to remain that way for long. The early web followed the same pattern. Before copyright enforcement, ad-tracking, and compliance bureaucracy took hold, there was a brief, chaotic explosion of creativity. That was when Google, Amazon, and PayPal were born.
Today’s AI landscape feels similar. The EU has passed its AI Act, and the U.S. is drafting its own frameworks—but enforcement is years behind. China is experimenting with guardrails, yet innovation continues unabated. For now, companies can build, deploy, and adapt with remarkable freedom.
Morgan Stanley’s AI wealth management assistant, trained on hundreds of thousands of internal research reports, was rolled out before clear regulatory guidance existed. The firm gained a first-mover advantage that will be harder to replicate once compliance frameworks tighten.
This window won’t stay open. Once the “regulatory-industrial complex” fully awakens—lawyers, auditors, ethics boards, and oversight committees—AI projects will slow, costs will rise, and freedom to experiment will shrink.
The pioneers will already have built the muscle memory of how to move fast and learn safely. Everyone else will be stuck writing policies. And by then, the biggest and most powerful companies will have captured the regulators—negotiating their own private sandpits with complex, compliance-heavy rules that only they can afford to follow. What begins as public safety will harden into private privilege, locking in incumbents and closing the door on new entrants who arrived just a little too late.
4. Outsize Returns Are Still Possible
Big things start small. In the 1990s, a graduate student project became Google. A scrappy online bookstore became Amazon. A simple payment tool became PayPal.
AI is at the same inflection point. Early movers are already turning modest bets into exponential value.
Harvey, the legal AI startup, began as a simple interface to an open API. Two years later, it’s embedded across global law firms, transforming how attorneys draft, review, and reason about complex documents. Cursor, which started as an AI coding assistant, has quietly become the command center for entire engineering teams—an integrated environment where agents not only write code but understand context, maintain memory, and orchestrate multi-step builds autonomously. And Runway, once an experimental video editor, now powers creative pipelines across media and marketing, collapsing production cycles from weeks to hours.
What these examples share is timing: each turned a narrow use case into a new cognitive infrastructure, capturing the kind of leverage that only exists in the early, chaotic phase of adoption.
Outsize returns always accrue in the early, unruly phase of adoption, before efficiency replaces imagination. The same pattern will play out across every industry—from healthcare and finance to manufacturing and education.
5. Ground Truth Still Exists
The final advantage of the current moment is epistemic: the data that feeds AI models is still relatively pure.
We haven’t yet reached the “hall of mirrors” stage where AI systems are swallowing their own synthetic output. But it’s coming. As AI-generated text, images, and code flood the internet, the risk of “model collapse”, where outputs become increasingly derivative, becomes increasingly likely.
For now, models are still largely trained on a foundation of human-authored books, research, and journalism. They reflect eons of human expertise and craftsmanship. That’s what makes today’s results so surprisingly coherent and useful.
But as the signal-to-noise ratio declines, verifying truth will require more time, tokens, and oversight. Leaders will need to invest in data provenance, curation, and trust architectures.
In other words, cognition is cheap now partly because the world’s knowledge base is still intact. Once that begins to erode, clarity will become a premium commodity.
6. The Window Is Closing
Every advantage we’ve just discussed—unfair knowledge, cheap economics, regulatory freedom, outsize returns, and reliable ground truth—is temporary.
Knowledge will spread. Costs will rise. Regulation will harden. Returns will normalize. Data will decay. The AI era won’t end, of course, it will just become ordinary, bureaucratic, and expensive, like every other mature technology before it.
For now, though, we inhabit a rare moment when a mid-sized company can act with the power of a global enterprise, and a single individual can operate with the leverage of an entire team. The smart leaders are not waiting for “maturity.” They are capturing cognitive territory while it’s still unclaimed.
Those who hesitate in periods like this tend to rationalize their inaction. They say they’re waiting for “best practices,” or for “the hype to settle,” or for “clarity on regulation.” But in doing so, they miss the only window when experimentation is both cheap and high-return.
Clarity will come but only after the opportunity has passed.
As I often say in my talks: the future favors the bold.
