The Art of Foresight

Two chip companies saw the same future coming. One spent twenty years building for it. The other kept optimising the business it already had. Then a third party changed everything.

This article is part of The Art of Leadership series, where I cover twelve principles of leadership that my research found underpin superior company performance. Read the full series here.

In March 2026, Nvidia’s CEO stood on stage and forecast one trillion dollars of AI computing demand by 2027. The market barely blinked, because by then, Nvidia’s dominance of AI infrastructure was simply a fact.

But here is what most people forget. Nvidia did not stumble into that position when ChatGPT launched. It had been building toward it for nearly two decades.

Back in 2004, Nvidia began investing in a software platform called CUDA that enabled its graphics chips to handle general-purpose computing. For years, this looked like a strange bet. CUDA cost money, added complexity, and served a market that barely existed. Analysts questioned why a company that made gaming graphics cards was pouring resources into scientific and parallel computing. The answer was that Nvidia’s leadership had made a judgment about where computing was heading, and they organised around it.

Twenty years later, when the AI era arrived, every large language model on earth needed exactly the kind of parallel computing that Nvidia had spent two decades perfecting. The bet did not just pay off. It made Nvidia one of the most valuable companies in history.

Now consider Intel.

Intel had every advantage Nvidia lacked. It was larger, richer, and utterly dominant in the market that mattered most for decades: the CPU. It had the money, the talent, and the manufacturing capability to build anything it wanted. And across the same twenty-year period, it made a series of decisions that look, in hindsight, like a masterclass in the absence of foresight.

Intel declined to build the chip for Apple’s iPhone, closing the door on the mobile computing era. Its then-CEO considered acquiring Nvidia for around $20 billion; the board balked. It had the opportunity to invest in OpenAI and passed, prioritising short-term returns over an uncertain future. Each decision was individually defensible, yet together they left Intel cutting 15,000 jobs as revenue fell 30%, its stock down from over $60 to under $20.

Two companies in the same industry facing the same transformative force. One built for the future. The other optimised the present.

And then something happened that makes this story far more interesting.

The third player

In August 2025, the US government converted $8.9 billion of CHIPS Act funding into 433.3 million Intel shares at $20.47 each, taking a 9.9% stake in a company that had just missed the defining technology shift of its era. At the time, it was widely mocked. Critics called it a bailout of a firm that had lost the AI race.

But look at what the government was actually doing.

It was not looking at Intel’s AI chip roadmap. It was looking at a different structural change entirely: the fact that the world’s most advanced semiconductor manufacturing had become dangerously concentrated in Taiwan, ninety miles from a country that has declared reunification inevitable. As the Center for Strategic and International Studies put it, Intel was the linchpin of American efforts to reduce dependency on chips manufactured abroad and regain leadership in semiconductor manufacturing technology. The deal even included a golden share giving Washington veto power over any future sale of Intel’s foundry business.

The market was valuing Intel on whether it could compete with Nvidia in AI chips. The government was valuing it on whether America could make advanced chips on its own soil.

Intel’s stock had surged over 190% for the year, its best performance on record. Q1 revenue rose 7% to $13.6 billion, data centre revenue jumped 22%, and earnings came in at $0.29 per share against a consensus of roughly one cent. Apple agreed to manufacture some of its chips at Intel’s foundry. Nvidia itself took a $5 billion equity stake. The government’s $8.9 billion position was worth more than $46 billion, an unrealised paper gain of over $35 billion.

This is important: three parties looked at the same company. The market saw a firm that had lost the AI race. Intel’s own leadership had spent two decades protecting a profitable present. And the US government saw a structural change nobody else was pricing, geopolitical concentration of chip manufacturing, and organised around it while everyone else was arguing about GPUs.

That is what foresight looks like. And it is available to anyone who learns to look properly. By the end of this article, you will understand why most leaders are bad at this, and you will have a practical method for getting better.

Let’s get into it.

Foresight is not prediction

Let me clear up the most common misunderstanding first. Foresight is not prediction.

Prediction is trying to guess what will happen: who wins the election, where interest rates land, which product takes off. Prediction is mostly a mug’s game, and leaders who stake their strategy on specific predictions usually get burned.

Foresight is different. Foresight is pattern recognition applied to structural changes that are already in motion but have not yet impacted in your results. Nvidia did not predict ChatGPT two decades ago. Few people did. What Nvidia’s leadership saw was a structural direction: computing was becoming more parallel, more data-intensive, and more demanding of exactly the capabilities their chips provided. They did not need to know the specific application that would prove them right. They needed to recognise the direction of travel and commit to it.

The same logic applies to the Intel stake. Washington did not predict that Apple would sign a foundry deal or that Q1 earnings would beat by 2,000%. It recognised a structural change, the securitisation of semiconductor supply chains, and acted on it before the market had priced it in.

This is the crucial distinction. You do not need a crystal ball. You need the discipline to identify the deep currents that are already flowing in your own industry, and the courage to build for where they lead, even when the destination is not yet visible in your results. Frankly, that courage is rarer than the insight, because the insight costs nothing and the commitment costs everything.

Remember: the structural changes that will reshape your industry over the next decade are almost always visible today, if you know how to look. They are just not yet urgent, which is precisely why most firms ignore them.

Why most leaders are bad at foresight

If foresight is so valuable, why is it so rare? Three reasons, and you will recognise all of them.

The present is noisy and the future is quiet. This quarter’s numbers demand attention. The competitor’s new product demands a response. The customer complaint requires resolution. The structural shift that will matter in ten years makes no noise at all, so it loses every fight for the leadership team’s attention. Intel’s CPU business was enormously profitable right up until it wasn’t. That profitability was exactly what made it so hard to look up and see the shift coming.

Foresight requires acting before the evidence is conclusive. By the time a structural change is obvious enough that the data proves it, the window to build a leading position has usually closed. Nvidia committed to parallel computing when the evidence was thin. If it had waited for proof, it would have been just another company scrambling to catch up when AI arrived. The US government bought Intel at $20.47 when the consensus was that Intel was finished. Both acted before the evidence was conclusive. That is not recklessness. That is the entire point.

Success is the enemy of foresight. This is the cruellest pattern, and it may be the one that applies most to you. The more dominant you are in your current business, the harder it is to invest in the thing that will replace it. Intel’s decades of CPU dominance did not help it see the future. They actively blinded it, because every instinct in the organisation was tuned to protect and extend the business that was already winning. As one analysis of Intel’s decline put it, moving beyond your historical advantage is hardest precisely when you are still profitable and change seems slow. But that is exactly when you have the resources to prepare for what comes next.

The foresight discipline

Here is the practical part. Foresight is not a personality trait that some leaders have and others lack. It is a discipline you can build. Here is how.

Separate structural changes from noise. Not everything that is happening is a structural change. Most of what fills the business press is noise: temporary trends, hype cycles, and things that will not matter in three years. A structural change has three properties. It has been building for years. It crosses multiple industries. And it is driven by deep forces, technological, demographic, economic, or regulatory, that are not going to reverse. Apply those three tests, and most of the noise falls away, leaving the handful of changes actually worth building around. I have written about these three tests in more detail.

Look where nobody else is looking. Everyone in the chip industry was looking at AI performance. The US government was looking at a map. The most valuable foresight rarely comes from analysing the same variables as your competitors more skilfully. It comes from noticing a change that sits outside the frame everyone else is using. If your foresight process only examines the factors your industry already tracks, you will see what your competitors see, at roughly the same time they see it.

Look where the researchers are looking. One of the most reliable early signals of a structural change is academic and scientific research. AI research was growing exponentially for a decade before the business world noticed. The deep currents show up in laboratories and journals years before they show up in revenue. If you want to see the future of your industry, look at what the smartest people in adjacent fields are working on now.

Ask the ten-year question, deliberately and often. Most leadership teams never ask “what will be true in ten years that is not true today?” They are too busy with this quarter. Build the question into your rhythm. Once a quarter, force the leadership team to lift its eyes from the present and debate the structural changes reshaping the environment. I have described how to build this into a repeatable foresight process.

Invest before it is comfortable. Foresight without commitment is just observation. The whole point is to act before the evidence is conclusive, which means allocating real capital and talent to a structural change while it still feels early. This is the hardest part, because it means diverting resources from the profitable present to an uncertain future. But that discomfort is the price of a leading position. Nvidia paid it for twenty years. Washington paid it in a single transaction that the entire market thought was foolish.

What foresight is not

You might be thinking, “This sounds like an argument for chasing every shiny new trend.” It is actually the opposite.

Foresight is disciplined, not fashionable. It means committing to a small number of deep structural changes and building around them for years, not lurching toward whatever is trending this month. The metaverse was fashionable in 2021. Firms that chased it were not exercising foresight; they were following hype. The difference between a structural change and a fad is exactly what the discipline is designed to detect.

Nvidia’s foresight was not that it jumped on every emerging technology. It was that it identified one deep structural direction, parallel computing, and committed to it with a consistency that lasted two decades. That is what foresight looks like in practice: not restlessness, but disciplined, patient conviction about where the world is heading.

What next?

The Intel story has a final lesson.

Intel is now recovering. Under a new CEO, with government backing, an Apple foundry deal, and an Nvidia partnership, the company that missed mobile and missed AI is up 190% and building chips on American soil. But notice who created that recovery. It was not the twenty years of internal strategy. It was an outside party that saw a structural change Intel’s own leadership had not organised around, and acted on it.

If your firm is not exercising foresight, someone else will exercise it on your behalf. A regulator. An investor. An acquirer. A government. And you will spend the next decade executing someone else’s judgement about your future.

The structural changes that will define your industry in 2036 are visible right now. They are operating in the background, they are not yet urgent, and they are losing the daily fight for your attention against this quarter’s demands. That is exactly why building around them is a source of advantage: because almost everyone else is ignoring them too.

Ask yourself the question Intel failed to ask, and Nvidia never stopped asking: what is changing in the world that will reshape my industry over the next decade, and am I building for it, or just optimising what I already have?

You still have time to ask it. Ask it today.