The Treasure Map Is Already Published: Why the Greatest Companies of 2031 Are Buried in Nvidia’s Portfolio, the Hyperscalers’ Venture Arms, and the Dreamers They Are Funding
By Brian French, Former Portfolio Manager and Investment Analyst A companion piece to the SLIC Framework and “The Imagination Economy”
The Thesis in One Paragraph
Part One argued that the cost of turning an idea into a digital business has collapsed, and that the next five years will produce more wealth from imagination than any period in history. Part Two asks the obvious follow-up question: where are those companies right now? The answer is that most of them are not hidden at all.
They are sitting, in plain sight, in the investment portfolios of the companies that can see the future first. Nvidia has quietly assembled a venture portfolio that by recent counts runs to roughly 260 companies. Alphabet operates three separate venture vehicles. Microsoft, Amazon, and the venture capital industry have collectively written checks into thousands of startups whose only asset is an imagination and a team.
The headline names in those portfolios, the frontier AI labs and the rocket company, are already priced as giants and are not where the extraordinary returns will come from. The extraordinary returns are in the smaller, stranger, less-covered companies further down the list: the autonomous-driving startups, the AI-for-science labs, the photonics and networking dreamers, the vertical-AI companies solving problems no one in finance has heard of. And the speed at which one of them can go from nothing to a trillion dollars of value has compressed, by the forces described in Part One, to something like twenty-four months. We are in the first inning, and the scorecard has already been printed.
Part One: Why the Incumbents See the Future First
There is a reason Nvidia’s venture portfolio is the most interesting document in technology, and it has nothing to do with Nvidia’s stock.
Nvidia sells the tool that every AI company needs. That means Nvidia sees the order book of the entire artificial-intelligence economy before anyone else does. It knows which startups are buying compute, how much, how fast their consumption is growing, and what they are building with it. When Nvidia writes a check into a startup, it is not making a blind bet the way a traditional venture fund does. It is investing in a company whose demand curve it has already observed from the inside.
This is a structural information advantage that has no precedent. The railroad barons could not see which towns would grow before laying track. The oil majors could not see which refineries would be profitable before building them.
Nvidia can see, in real time, which of its customers are pulling away from the pack, and it can invest in them before the public market has any idea they exist. Brian French
The hyperscalers have a version of the same advantage. Alphabet, through GV, CapitalG, and Gradient Ventures, sees which startups are consuming its cloud, which are building on its models, and which are generating traffic through its search and advertising systems. Microsoft’s M12 fund and its direct investments see the same through Azure and GitHub. Amazon sees it through AWS. These are not venture funds guessing at the future. They are distribution platforms watching the future arrive through their own pipes and buying a piece of it at the door.
Traditional venture capital, for its part, has been doing this for seventy years without the pipes, and its best firms have developed a pattern-recognition capability that compensates. The point is not that any one of these sources is infallible. The point is that between Nvidia’s order book, the hyperscalers’ usage data, and the venture industry’s pattern recognition, the companies that will define 2031 have already been identified, funded, and named. The map exists. It is simply not being read by the people who buy public stocks.
Part Two: Forget the Headliners
The first thing an investor does when reading these portfolios is to look for the famous names, and the first mistake is to stop there.
Nvidia’s investments in the frontier AI labs and its reported participation in the aerospace company that dominates the imagination narrative are the entries that make headlines. They are also the entries that are already priced as if they had won. A frontier lab valued in the hundreds of billions, or a rocket company valued in the same range, may well double or triple. But a company already worth $300 billion cannot go up a hundredfold; the math of market capitalization does not allow it.
The great returns of the Imagination Economy will not come from the companies that are already giants. They will come from the companies that are currently worth $1 billion, or $5 billion, or $20 billion, and that will be worth a trillion.
This is the same lesson the SLIC framework draws from Amazon and Apple. The extraordinary return was not available to the investor who bought Amazon at $500 billion. It was available to the investor who bought it at $5 billion, when it was a money-losing bookstore with a founder’s letter.
The portfolios of Nvidia and the hyperscalers are full of companies at exactly that stage today. They are, almost by definition, the companies no one is writing about, because the companies everyone is writing about have already been discovered.
So set the headliners aside. The treasure is further down the list.
Part Three: The Speed Problem, or Why Twenty-Four Months Is Not Crazy
The claim that a company can go from zero to a trillion dollars in value in two years would have been absurd in any previous era. It is worth explaining why it is not absurd now.
Start with what has already happened. Nvidia itself crossed $1 trillion in market capitalization in mid-2023. It crossed $2 trillion roughly eight months later, $3 trillion a few months after that, and $4 trillion by mid-2025. It added more than $2 trillion of value in about twelve months, which is more than the total market capitalization of every company in the world except a handful. The value did not come from a new product; it came from the market re-imagining the addressable market for AI compute.
Now look at private companies. The leading frontier AI lab was reported at a valuation of roughly $30 billion in early 2023, about $160 billion in late 2024, about $300 billion in early 2025, and about $500 billion in a secondary transaction later that year. That is a roughly fifteenfold increase in a little over two years for a company that did not exist as a commercial entity a decade earlier. A second lab, founded in 2023, was reportedly valued at around $50 billion by the end of 2024 and multiples of that by late 2025. A third, founded by a departing co-founder of the first, reportedly raised at a valuation above $30 billion before releasing any product at all. A fourth, formed by former lab executives, reportedly raised a seed round at a $12 billion valuation.
Then look at Nvidia’s own portfolio. One of its earlier investments, a company that began as a cryptocurrency mining operation and pivoted to renting GPUs, went public in March 2025 at a valuation around $23 billion and, within a few months, was reported to be worth several times that. Nvidia had invested when the company was worth a small fraction of its IPO price. An AI-native code editor, not a Nvidia investment but a pure Imagination Economy company, reportedly went from launch to a valuation approaching $30 billion in about two years.
None of these is a trillion yet. But the rate is the point. Fifteenfold in two years, from a $30 billion base, is $450 billion. Fifteenfold in two years from a $70 billion base is a trillion. The base valuations of the leading private companies are now high enough that a single additional cycle of the kind already observed puts them at a trillion. And the forces from Part One, the operating leverage of digital assets, the speed of social distribution, and the compression of building time by AI, mean that the next cycle can be faster than the last.
There is also a structural reason the private market can now produce a trillion-dollar company before the public market ever sees it. The pools of capital available to private companies have grown enormous: sovereign funds, the hyperscalers’ balance sheets, Nvidia’s own cash flow, crossover investors, and secondary markets.
A company no longer needs an IPO to raise tens of billions or to give early employees liquidity. It can stay private, compound in private, and arrive on the public market, if it ever does, already worth more than most of the S&P 500. The public investor who waits for the IPO will be buying at the end of the ride.
Part Four: The Treasure Trove
What follows is not a recommendation list. It is a description of the categories in which the smaller, less-covered portfolio companies of Nvidia, the hyperscalers, and the venture industry are concentrated, with illustrative examples drawn from public reporting. Every one of these should be scored on the SLIC framework: nearly all of them have negligible C, high L (the capital is flowing to them), maximum I, and a rising S.
Autonomy. For seventy years, software could think but not act. That barrier is breaking, and the first place it is breaking at scale is on the road. Nvidia has invested in autonomous-driving companies pursuing end-to-end learned driving rather than hand-coded rules, and in autonomous trucking. Alphabet’s own autonomous-vehicle unit has reportedly been valued at around $45 billion and is operating paid rides in multiple cities. The addressable market is global transportation of people and freight, a market measured in the trillions of dollars annually, in which the largest cost is the human driver. A company that removes that cost reliably and safely will be worth more than every automaker combined. That is the imagination narrative, and it is already funded.
AI for science and medicine. The largest imagination narrative in the portfolios is not about chatbots. It is about compressing the time it takes to discover a drug, a material, or a physical law. Nvidia has invested in AI-driven drug discovery companies. Alphabet has spun out a drug-design lab built on the protein-folding breakthrough that won a Nobel Prize. Startups in the portfolios are working on AI-designed proteins, AI-driven clinical trials, and AI agents that handle patient communication for health systems.
The addressable market is the entire pharmaceutical industry plus the entire healthcare-delivery industry. A company that reliably cuts drug-discovery time in half would be the most valuable company in the world, and there are several in the portfolios trying.
The infrastructure dreamers. Beneath the AI models is a layer of physics that most investors never see: the chips, the interconnects, the optical links that move data between them, the cooling, and the power. Nvidia’s portfolio includes companies building optical interconnects to replace copper, companies re-architecting the network fabric inside data centers, and companies building specialized inference chips.
These names are boring to the general public and therefore invisible to the S pillar. That is exactly why they are interesting. A company that solves the interconnect bottleneck for the next generation of AI clusters has an addressable market equal to the entire AI capital-expenditure budget of the world, which is measured in the hundreds of billions of dollars per year. It will not trend on social media until after it has won.
Vertical AI and agents. The first wave of AI products was horizontal: a chatbot for everyone. The next wave is vertical: an AI lawyer, an AI radiologist, an AI accountant, an AI customer-service department, an AI software engineer. Several companies in the hyperscaler and venture portfolios have reportedly reached hundreds of millions of dollars in annualized revenue within two years of launch by picking one profession and automating it deeply.
The economics are the Imagination Economy in its purest form: near-100% gross margins, a product that improves every quarter as the models improve, and a customer who pays because the alternative is a human salary. The total addressable market of vertical AI is global white-collar labor, which is measured in the tens of trillions of dollars annually.
Creative tools. Video generation, image generation, music generation, and game creation are the consumer face of the Imagination Economy. Nvidia and the hyperscalers have invested in the leading video-generation and creative-tool companies. These are the businesses that will be loved on social media first, because their output is social-media content. They will have the highest S scores of any category, and the S score will arrive before the C score.
Sovereign and regional AI. Nvidia and the hyperscalers have invested in AI labs in Europe, Japan, the Middle East, and elsewhere, on the thesis that every major economy will want a frontier model it controls. These companies have a captive addressable market, their national governments and enterprises, and a liquidity pillar backed by sovereign capital. They are less likely to reach a trillion, but more likely to reach fifty billion with lower risk.
Across every category, notice the pattern. The company’s name is obscure. Its product is technical. Its revenue is small or zero. Its addressable market is described in trillions. And it has already been found, funded, and observed by the one investor in the world who can see its demand curve from the inside. Brian French
Part Five: How to Read the Map
The investor who wants to find the great companies of 2031 does not need proprietary access. The portfolios are public in outline. Nvidia’s venture arm and its direct investments are reported in the financial press and tracked by the private-market databases. Alphabet’s venture units publish their portfolios. The venture industry announces its rounds. The information is available; what is scarce is the discipline to read it.
Three rules for reading it.
Rule 1: Weight the check-writer’s information advantage. A Nvidia investment in a compute-hungry startup carries more signal than a generalist venture investment in the same company, because Nvidia has seen the demand. An Alphabet investment in a company building on Google’s cloud carries more signal than one in a company that does not touch Google’s infrastructure. Follow the investor who could see the order book.
Rule 2: Ignore the top of the list. The famous names are priced. Start reading from the bottom, at the companies with the least coverage and the strangest products. Score each one on SLIC. The ones with a maximum I score, an absurd TAM, and no social coverage at all are the candidates. The Imagination Economy rewards the investor who reads the footnotes.
Rule 3: Watch the S pillar for the turn. The moment a portfolio company begins appearing in the influencer ecosystem, the moment the YouTube channels and the podcasts and the subreddits discover it, its S score jumps and its price follows. The investor who identified the company on I alone, before the S turn, captures the re-rating. The investor who arrives after the S turn is buying at the same price as everyone else.
Part Six: The Honest Caveats
The dreamers’ portfolio is not a free lunch, and the SLIC framework’s interaction rules apply with full force.
Most of these companies will fail. A portfolio of 260 startups is a portfolio in which perhaps 200 will be worth nothing, 50 will return modest multiples, and a handful will produce all of the returns. This is the mathematics of venture capital and it does not change because the check-writer is Nvidia. Concentration in any single portfolio company is a bet on a coin that lands on zero most of the time.
The valuations already embed I and S, with no C. A lab worth $30 billion before a product, or an autonomy company worth tens of billions before meaningful revenue, is priced entirely on imagination and liquidity. If either pillar turns, the price falls by 80% or more, as the 2021 cohort of story stocks demonstrated in 2022. Rule 4 of the SLIC framework applies: I without C requires a time limit, and the investor must know in advance the date by which cash flows must appear.
Some of the liquidity is circular. Skeptics have noted, correctly, that when Nvidia invests in a company that then uses the money to buy Nvidia’s chips, some of the reported demand is Nvidia’s own capital coming back to it. The same is true when a hyperscaler invests in a lab that then rents the hyperscaler’s cloud. This does not mean the companies are worthless; it means their revenue should be examined for how much of it is independent of their investors. The L pillar is real, but the investor should know where the liquidity is coming from.
Access is unequal. Most of these companies are private, and most private rounds are closed to the public. Participation is possible through the public shares of the check-writers themselves, through the growing set of publicly traded vehicles that hold private technology companies, through secondary-market platforms, and, for accredited investors, through venture funds. Each has its own costs, fees, and risks. None of this is a recommendation to use any of them; it is an observation that the asset class exists and that the index investor holds none of it.
The First Inning, Revisited
Part One ended with the observation that we are in the first inning of the Imagination Economy. Part Two adds the detail that the lineup card has already been posted.
The companies that will be worth a trillion dollars in 2031 are, with high probability, already incorporated. Most of them have already raised money. Many of them appear in the investment portfolios of Nvidia, Alphabet, Microsoft, Amazon, and the venture firms that have been reading the same signals. Their names are obscure, their products are technical, their revenues are small, and their addressable markets are described in units that make traditional analysts uncomfortable.
That discomfort is the opportunity. The single-factor model, the one that discounts cash flows at a CAPM rate and calls the result intrinsic value, assigns these companies a value of approximately nothing. The SLIC framework assigns them a score, decomposes the score into what is known and what is imagined, and applies a time limit to the imagined part. Neither framework can tell you which of the 260 will win. But only one of them can tell you that the 260 are worth looking at.
The world moves on. The map is published. Read the bottom of the list.
This document is provided for educational and analytical purposes only. It does not constitute investment advice, an offer to buy or sell securities, or a guarantee of any investment outcome. Portfolio counts, valuations, and company figures are drawn from public reporting, are approximate, and change rapidly; readers should verify them against primary sources before relying on them. Private-company investments carry a substantial risk of total loss. Readers should consult a qualified financial advisor before making investment decisions.
