The sign on your wall is lying to you.
You know the one: “Good, Fast, Cheap. Pick Two.”? For seventy years, this “Iron Triangle” hasn’t just been a project management tool; it’s been the ultimate corporate insurance policy. It was the valid excuse for every missed deadline, every bloated budget, and every mediocre delivery. It was a law of business physics. The comforting certainty that allowed us to shrug our shoulders when things went wrong.
I’m here to tell you that the law has been repealed. The triangle has collapsed. And if you’re still using it to set expectations, you aren’t being “realistic”—you’re becoming obsolete. I’ve seen this sign, or some variation of it, in countless offices throughout my career. From my early days building ISV channels to my current work leading marketing across North-West Europe, this triangle has been the unspoken contract between what clients want and what’s actually possible. It gave us permission to be slow. Permission to be expensive. Permission to deliver less than extraordinary.
Those permissions have been revoked.
The Anatomy of the Impossible
Before we get into the *how*, let’s understand the *why*. Why couldn’t these three things coexist in the first place? Because the Iron Triangle was anchored in the limitations of human biology. Humans have a “Cognitive Tax.” We get tired. We need sleep to synthesize ideas. We have “off days.” When you hire someone to do something, you’re paying for their time, their expertise, and their capacity. That capacity is brutally limited. A brilliant designer can only produce so many concepts in a day. A skilled developer can only write so many lines of quality code before exhaustion sets in. A seasoned strategist needs time to think, to process, to connect dots that aren’t obvious at first glance.
To get something Good, you had to buy a human’s focus—the rarest commodity on earth. To get it Fast, you had to buy multiple humans’ focus, which is expensive and leads to “too many cooks” syndrome. To get it Cheap, you either hired less experienced people (not good) or you stretched timelines to allow for iterative improvement (not fast).
We weren’t fighting bad management. We were fighting the biological ceiling of the human brain. The triangle wasn’t some arbitrary rule invented by consultants to justify higher fees. It was a direct consequence of human limitation. And for decades, it held firm.
Until now.
The AI Breach
Let me walk you through how generative AI is systematically dismantling each pillar of the Iron Triangle.
Fast: Processing vs. Thinking
Here’s a fundamental truth about human creativity: it takes time. Not because we’re lazy, but because the cognitive process of synthesizing information, making connections, and generating novel ideas requires cycles of thought. We need to sleep on things. We need to step away and come back with fresh eyes. We need coffee breaks and shower thoughts and those random moments of clarity that strike at 2 AM.
Generative AI doesn’t have any of that. It doesn’t need days to brainstorm. It doesn’t require multiple rounds of internal review before sharing a first draft. It doesn’t get tired, doesn’t need weekends off, and definitely doesn’t lose momentum after lunch.
What used to take a team three days of workshop sessions and two weeks of refinement can now happen in seconds. Or minutes, at max. And I’m not talking about mediocre output that needs heavy editing. I’m talking about genuinely useful starting points that compress the entire creative timeline.
This isn’t about replacing human thought. It’s about redefining the relationship between question and answer, between brief and draft, between idea and execution. The speed isn’t incremental. It’s exponential.
Good: The Quality Renaissance
The skeptics are still laughing at 2023. They’re still making jokes about AI-generated hands with six fingers and “hallucinations.” Stop laughing. The “Quality Ceiling” isn’t just rising; it’s shattering.
Look at the SWE-bench—a brutal test of real-world coding capability. In 2023, AI solved 4.4% of problems. In 2024? 71.7%. That isn’t an “improvement.” That is a species-level leap in capability. When the difference between the AI models in the world shrinks to less than 6%, you aren’t looking at a niche tool; you’re looking at a commodity of excellence.
The data is unambiguous. According to the Chatbot Arena Leaderboard, the Elo score difference between top-tier and tenth-ranked models narrowed from 11.9% in early 2024 to just 5.4% by early 2025. The gap between the top two models collapsed from 4.9% in 2023 to just 0.7% in 2024. As the Stanford AI Index report noted: “The frontier is increasingly competitive — and increasingly crowded.”
Code became AI’s first true “killer use case” as models reached economically meaningful performance—with Anthropic’s Sonnet 3.5 triggering the category’s initial breakout in mid-2024. Adoption followed soon after; 50% of developers now use AI coding tools daily (65% in top-quartile organizations).
Teams report 15%+ velocity gains as they’ve adopted AI tools across the software development lifecycle. This is not hype. This is not theoretical. This is developers—the most skeptical profession on the planet—fundamentally changing how they work.
The “AI produces mediocre work” argument is increasingly becoming a relic of 2023 thinking. The gap between “human-grade” and “AI-grade” is closing so fast it’s becoming negligible for most business outputs.
Cheap: The Economics of Abundance
And here’s where things get really interesting.
Sam Altman captured this phenomenon with a striking phrase: “Intelligence too cheap to meter.” And the data backs him up spectacularly.
“You can see this in the token cost from GPT-4 in early 2023 to GPT-4o in mid-2024, where the price per token dropped about 150x in that time period. Moore’s law changed the world at 2x every 18 months; this is unbelievably stronger.”
Let that sink in. Moore’s Law—the economic engine that powered the entire digital revolution—operated at 2x performance improvement every 18 months. AI cost reduction is operating at 10x every 12 months. This isn’t an incremental shift. This is an entirely new category of exponential change.
To put that in perspective: If the price of a $50,000 luxury car dropped at the same rate as AI tokens, that car would now cost $178.
New AI startups like DeepSeek are offering similar AI tools at even lower prices than OpenAI. DeepSeek-V3 is priced at $0.14 per million input tokens and $0.28 per million output tokens, compared to OpenAI o1 Preview’s $15.00 and $60.00, respectively.
Token pricing has become significantly more competitive in 2025, but the differences between providers can still make or break your budget. Major AI providers now offer various tiers to suit different use cases. For lighter tasks, GPT-4o mini offers an economical alternative at just $0.15 per million input tokens and $0.60 for output.
But if energy infrastructure improves while AI companies continue prioritizing efficiency, we can expect the cost of tokens to drop further. We could even be looking at a future where AI is as cheap as a Google search.
When the marginal cost of a world-class strategy, a comprehensive market analysis, or a stunning piece of creative work approaches the price of the electricity used to generate it, the word “Cheap” no longer does it justice. We are entering the era of Economic Abundance.
Once a model is trained, it doesn’t ask for a raise. It doesn’t require benefits. It doesn’t need six months of onboarding. The more you use it, the cheaper per-unit it becomes.
This isn’t coming. It’s here.
The Collapse in Practice
So what happens when you combine speed measured in seconds, quality that rivals human expertise, and costs approaching negligible?
You get the death of the Iron Triangle.
Enterprise AI adoption accelerated dramatically in 2024-2025. Implementation extended beyond pilot programs into production deployments across core business functions. 78% of organizations now use AI in at least one business function, up from 55% in 2023. Generative AI adoption more than doubled in a year, rising from 33% in 2023 to 71% in 2024.
This isn’t hype. This is adoption at a speed we’ve rarely seen with any technology. The demand side tells a clear story: broad adoption, real revenue, and productivity gains at scale, signaling a boom versus a bubble. Enterprise AI has surged from $1.7B to $37B since 2023, now capturing 6% of the global SaaS market and growing faster than any software category in history.
Industry analysts forecast the AI market will expand at a compound annual growth rate (CAGR) of 35.9% from 2025 to 2030, potentially reaching $1.81 trillion by the end of the decade. This growth rate surpasses both the cloud computing boom of the 2010s and the mobile app economy of the early 2010s.
The economics are simply too compelling to ignore. Companies that moved early into GenAI adoption report $3.70 in value for every dollar invested, with top performers achieving $10.30 returns per dollar.
Consider the efficiency gains: In 2022, the smallest AI model registering a score higher than 60% on the MMLU benchmark was PaLM, with 540 billion parameters. By 2024, Microsoft’s Phi-3-mini, with just 3.8 billion parameters, achieved the same threshold. This represents a 142-fold reduction in just over two years.
Smaller models. Better performance. Lower costs. The trends are all pointing in the same direction.
The New Constraint: From Execution to Taste
So here’s where I want to pivot the conversation. If Good, Fast, and Cheap are now the baseline—if the Iron Triangle has truly collapsed—what becomes the new differentiator?
The differentiator shifts from Execution to Taste.
In a world where anyone can press a button and get a 90th-percentile output, the “worker” is dead. The “Editor-in-Chief” is the new king. The new scarcity isn’t the ability to *do* the work; it’s the wisdom to know *which* work is worth doing. We are moving from a world of “How” to a world of “What.”
Leaders must now rethink workflows from first principles, asking fundamental questions about whether certain processes and roles should exist in an era of abundant intelligence.
When everyone has access to AI that can produce quality work at speed for minimal cost, the winner is no longer the one with the biggest budget or the fastest team. The winner is the one with the best idea. The clearest vision. The sharpest strategic thinking. The ability to recognize a good output from a mediocre one. The creativity to imagine possibilities that haven’t been done before.
If you can’t tell the difference between “good” and “great” without a committee, you are in trouble.
As Altman envisions: “Anyone in 2035 should be able to marshal the intellectual capacity equivalent to everyone in 2025; everyone should have access to unlimited genius to direct however they can imagine.”
This is simultaneously exhilarating and terrifying.
Exhilarating because the barriers to entry are crumbling. Anyone with a good idea now has access to execution capabilities that were previously reserved for well-funded enterprises. A startup can produce marketing materials that rival those of Fortune 500 companies. A solo entrepreneur can build prototypes that would have required entire teams. The sharp jump in AI coding tools from $550M to $4B in 2025 reflects a shift in capability: Models can now interpret entire codebases and execute multi-step tasks.
Terrifying because the old excuses no longer work. “We don’t have the budget” loses its weight when quality AI is nearly free. “We don’t have the time” falls flat when AI can deliver in minutes. “We can’t find the talent” becomes hollow when AI can replicate specialized skills at scale.
The Danger of Waiting
The Iron Triangle wasn’t just a project management tool; it was a security blanket for the slow and the expensive. It provided a valid excuse for the friction of human limitation.
That blanket has been pulled away.
Organizations are beginning to explore opportunities with AI agents—systems based on foundation models capable of acting in the real world, planning and executing multiple steps in a workflow. Twenty-three percent of respondents report their organizations are scaling an agentic AI system somewhere in their enterprises, and an additional 39 percent say they have begun experimenting with AI agents.
2025 is the year AI stopped being a “tool” you use and started being a “utility” you plug into. The gap between the companies embracing this and the ones “waiting for the tech to mature” is no longer a gap—it’s a canyon.
Every month you wait, the gap between you and your AI-embracing competitors widens. Not because they’re spending more money than you. Not because they have bigger teams. But because they’ve realized that the old rules no longer apply—and they’re acting accordingly.
In 2025, 31% of the use cases studied reached full production, which is double the amount compared to 2024. But expectations that AI would cut costs and boost productivity are underdelivering—not because the technology isn’t ready, but because organizations haven’t restructured their thinking around abundance instead of scarcity.
Most organizations are still navigating the transition from experimentation to scaled deployment. The experience of the highest-performing companies suggests a path forward. These organizations stand out for thinking beyond incremental efficiency gains: They treat AI as a catalyst to transform their organizations, redesigning workflows and accelerating innovation.
Stop asking what AI can do for your budget. Start asking what your business looks like when intelligence costs nothing and execution takes no time.
In a world where you can finally have all three—Good, Fast, and Cheap—the only thing standing between you and a “perfect” output isn’t time or money. It’s your own imagination. Your willingness to experiment. Your courage to question whether the processes you’ve relied on for decades still make sense.
The question is no longer “What can we afford to do?” It’s “Now that we can do anything, what is actually worth doing?”
