Every transformational technology starts the same way: as something most investors overlook. Not because they're careless, but because the early evidence doesn't look like evidence.
It looks like noise. A curiosity in a research paper. A line item buried in an earnings call transcript. A startup solving a problem nobody's quantified yet.
This is Stage 1 of a paradigm shift, the Footnote. It's where asymmetric opportunities are born.
The Pattern Hiding in Plain Sight
In 1995, a Newsweek columnist wrote a piece titled "The Internet? Bah!" He argued that no online database would ever replace a daily newspaper, that e-commerce was a fantasy, and that the whole thing was overhyped.
It was a reasonable position at the time. There were roughly 16 million internet users worldwide, less than 0.4% of the global population. Amazon had just launched and was losing money selling books out of a garage.
Was the columnist simply a luddite? Anything but.
Clifford Stoll was a PhD, teacher, astronomer, the system administrator at Lawrence Berkeley National Lab, and he even captured a KGB hacker! He knew his way around computers—far better than most.
But Stoll was pattern-matching against the wrong frame.
He considered online shopping a dead end due to the lack of human contact. He called digital newspapers "baloney" because reading on a screen was an "unpleasant chore."
These were all correct observations. And that's what makes this so hard.
In 1995, screens were terrible. Online payments were untrustworthy. The internet was slow, ugly, and unpleasant to use. Every data point Stoll cited was accurate.
What he didn’t account for was the technology improving exponentially while human behavior adapting to meet it halfway. Today, global e-commerce sales exceed $6 trillion.
That gap—between what a technology looks like and what it becomes—is the defining feature of the Footnote stage.
And it recurs with striking regularity.
Why Smart Money Gets This Wrong
The Footnote persists because it exploits the very heuristics that normally make investors successful.
"Follow the revenue."
By 2013, Tesla had been public for three years. The Model S was on the road, winning Motor Trend's Car of the Year and proving that an electric vehicle could be desirable—even aspirational.
But Wall Street wasn't buying the story.
On paper, their skepticism made sense. Tesla delivered about 22,000 cars that year and lost $74 million. General Motors delivered 9.7 million vehicles and posted $3.8 billion in net income.
Every sensible screen—P/E, EV/EBITDA, revenue growth per dollar of capital employed—pointed you toward GM. Tesla was the most-shorted stock on the Nasdaq.
The shorts had a coherent thesis: Tesla was a niche luxury automaker burning cash, dependent on regulatory credits, with no credible path to mass-market production. Every traditional valuation framework validated that view.
Tesla wouldn't post a sustained profitable quarter until Q3 2018, when it finally ramped Model 3 production and reported $312 million in GAAP net income. Between 2013 and early 2020—the entire Footnote window—the stock hovered around $7 to $20 (split-adjusted).
By January 2021, it was above $280.
To capture that return, it wasn’t enough to follow the revenue. We needed to follow the physics: the falling cost curves of lithium-ion batteries, the thermodynamic superiority of electric drivetrains, the regulatory trajectory toward decarbonization.
None of that shows up on a traditional screen.
"Invest in what you understand."
Buffett's axiom is powerful, but it has a blind spot: frontier technologies are misunderstood by definition. They don't fit existing categories.
Consider Nvidia circa September 2012.
That month, a team at the University of Toronto—Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton—entered a neural network called AlexNet into the ImageNet image-recognition competition.
This model, to put it bluntly, obliterated the field.
AlexNet cut the error rate by more than 10 percentage points versus the runner-up. It instantly made all other approaches obsolete.
The key detail: AlexNet was trained on two Nvidia GTX 580 GPUs, gaming graphics cards that cost $500 apiece, using Nvidia's CUDA programming platform.
Jensen Huang noticed immediately. He began repositioning the company's roadmap around this insight—that GPUs could become the central processor for a new class of AI workloads.
Wall Street didn't notice. To analysts, Nvidia was still a gaming chip company. They benchmarked it against AMD's graphics division. They modeled revenue based on PC shipment cycles and console refresh timelines.
Through 2014 and 2015, you could read dozens of sell-side reports on Nvidia without finding a single mention of deep learning. Data center revenue, the segment that would eventually become Nvidia's dominant business, was barely broken out. It was a rounding error inside a $4 billion gaming company. The stock traded at around $0.30 to $0.40 (split-adjusted) through most of 2013 and 2014.
Since then, Nvidia has delivered returns north of 450x. The workload that didn't exist in 2012 turned out to be the most important computing platform of the next decade.
"Wait for proof."
This is the costliest heuristic of all. Not because it's wrong, but because by the time proof arrives in a paradigm shift, the asymmetry is gone.
When Amazon Web Services launched in 2006, cloud computing was a footnote even within Amazon itself. AWS was a side project. Jeff Bezos had noticed that Amazon's internal infrastructure teams were spending months provisioning servers for new projects, and he intuited that if Amazon's own engineers needed elastic, on-demand computing, other companies would too.
The company didn't break out AWS revenue until 2015. When it finally did, the number was $7.9 billion, growing 70% year-over-year.
For nearly a decade, AWS grew inside Amazon's financial statements without its own line item. By the time investors had standalone "proof"—a clean, auditable revenue line to underwrite—Amazon's stock had already risen more than 1,000% from AWS's launch.
What the Footnote Actually Looks Like
If you study the early coverage of technologies that later redefined their sectors, a few signatures keep recurring.
The technology solves a problem that hasn't been widely named yet.
Before the iPhone, nobody walked around saying, "I wish I had a pocket computer with a multi-touch display and a software ecosystem built by third-party developers." That wasn't a felt need. The Blackberry seemed fine. Nokia shipped 133 million phones in Q4 2007. Apple sold 1.4 million iPhones by its first quarter after launch—a solid start, but nothing that screamed paradigm shift.
Then Steve Ballmer settled the matter for Microsoft. In a 2007 interview, the then-CEO of the world's largest software company laughed at the device and said there was "no chance" the iPhone would gain significant market share. His reasoning: the phone cost $500, had no keyboard, and wouldn't appeal to business customers.
Every point was defensible. But every point was irrelevant, because Apple wasn't competing for the existing smartphone market. It was creating a new one.
The early market is too small to register.
CRISPR Therapeutics went public in October 2016 at $14 per share. The company had zero revenue and zero approved products. Its core technology—CRISPR-Cas9 gene editing, based on the 2012 Science paper by Jennifer Doudna and Emmanuelle Charpentier—had been demonstrated in labs but never in humans.
Major pharmaceutical companies weren't allocating meaningful R&D budgets to gene editing. There was nothing to model, nothing to screen for. The entire gene editing therapeutics sector was a collection of pre-clinical companies running on venture capital and hope.
The stock drifted between $11 and $20 for most of its first two years as a public company.
On December 8, 2023—roughly seven years after that IPO—the FDA approved Casgevy, developed by CRISPR Therapeutics and Vertex Pharmaceuticals, as the first CRISPR-based gene-editing therapy ever cleared in the United States.
The gene editing therapeutics market is now biotech’s fastest growing segment.
Incumbents dismiss it with confidence.
When the CEO of a trillion-dollar company tells you a new technology isn't a threat, it's worth asking a simple question: Does this person have a structural incentive to believe that?
The answer is almost always yes.
Incumbents are optimized around the existing paradigm. Their revenue models, their org charts, their capital allocation frameworks, their compensation structures—everything is built to extract value from the current way things work. Acknowledging a paradigm shift threatens their market share and their identity.
So when Ballmer laughed at the iPhone, he was defending a worldview in which Microsoft's software licensing model was the right way to capture mobile computing value. When legacy media dismissed the internet, they were defending a worldview in which editorial curation and printing presses were indispensable.
Each of these worldviews was internally coherent. Each was also wrong.
The dismissal itself is the signal. Pay attention when powerful people confidently explain why something won't work.
The early data is ambiguous—and that ambiguity is the opportunity.
Paradigm shifts never arrive with clean dashboards and consensus buy ratings. They arrive with contradictory signals, small sample sizes, and heated debates about whether the underlying technology even works.
This is a feature, not a bug. Ambiguity is what keeps an asset mispriced.
If the data were clean, institutions would already be in. If the thesis were obvious, it would already be priced. The mess is the opportunity, not an obstacle to it.
The Misunderstanding Premium
In the Footnote stage, the gap between current perception and future reality is at its widest. That gap is where excess returns live.
We call it the Misunderstanding Premium, the implicit discount the market applies to a technology it doesn’t yet fully understand.
Structurally, this premium will always exist for frontier technologies.
Sell-side research coverage is driven by market capitalization. If a company is too small, nobody writes about it. Institutional mandates require liquidity thresholds. If the float is thin, the large funds can't buy. Index inclusion depends on established metrics. If the company doesn't meet the criteria, the passive flows never arrive.
Every layer of the capital markets machinery is optimized to identify things that have already become important. It is, by design, late to things that are becoming important.
As a result, there's a persistent, structural edge for investors willing to engage with technologies before they're legible to the consensus.
The numbers support this.
Hendrik Bessembinder, an economist at Arizona State University, analyzed the returns of over 26,000 publicly listed U.S. stocks from 1926 to 2019. His finding was extraordinary: just 4% of listed companies accounted for all net shareholder wealth creation above Treasury bills—roughly $47.4 trillion.
The remaining 96% of stocks collectively matched risk-free returns. More than half of all individual stocks actually lost money over their lifetimes.
The implication is clear. Almost all long-term equity returns come from a tiny number of outliers. And those outliers, overwhelmingly, were companies riding paradigm shifts that most investors dismissed when the evidence was still ambiguous.
A Simple Mental Model
Of course, we don’t want to chase every speculative idea with a whitepaper and a ticker symbol.
Most technologies that look like footnotes stay footnotes. They fail. That's the base rate, and ignoring it is how people blow up.
Our stance is subtler: if the market confidently categorizes a technology as irrelevant, unproven, or "too early"—yet the underlying science or engineering is sound—we lean in rather than tune out.
That discomfort is where paradigm shifts begin.