The Arithmetic of the American Dream

Two People on a Plane

My parents came to America at fifteen. They arrived with a belief so simple it barely qualifies as an idea: that here, effort converts to a better life. Work hard, live honestly, and your children will start from higher ground than you did. That was the whole contract. That was the American Dream — not a promise of ease, but a promise of arithmetic. Effort in, prosperity out.

For them, the contract held. They earned ten times what they would have earned in the country they left. They raised two children into an opportunity they could never have manufactured back home. By any honest accounting, the dream worked.

But I have spent my adult life inside the machinery of American capital — twelve years across technology banking, growth equity, and venture — and I now understand something about my parents' success that they never had the chance to see. They paid for it almost entirely in time. They worked too hard, for too long, because labor was the only asset they knew how to deploy. Nobody ever sat down with them and explained what a tax-advantaged account was. Nobody showed them that a portion of what they earned, invested instead of merely saved, would have compounded into years of their lives returned. Nobody helped them plan an estate, or structure a filing, or think three moves ahead about money — because the people who do that work, the advisors and planners and tax strategists, were built to serve people who already had wealth. My parents weren't the ideal customer. So they were nobody's customer at all.

They didn't lack intelligence or discipline. They lacked access to expertise. And the cost of that gap was not measured in dollars. It was measured in dinners missed, in years of health spent at work, in time with their children that money earned too slowly to buy back.

I am building APEX because I believe that trade — expertise for time, knowledge for years — is the quiet tax on every working family in America. And I believe that for the first time in history, we have the technology to abolish it.

But to see why this matters now — why it is urgent, not merely worthy — you have to look at what happened to the arithmetic.

The Ledger Broke

Start with a single number. In 1980, American workers received about 58 cents of every dollar of national income. Today they receive about 51. Corporate profits' share went the other way — from 7 cents to nearly 12. Seven cents doesn't sound like much until you multiply it across a $30 trillion economy: it works out to roughly $8,000 to $12,000 a year, per worker, every year, that moved from the people who do the work to the people who own the work.

That is the macro version. The household version is worse, because it compounds from both directions. The money coming in grew slowly — and everything a family must buy to live a normal American life grew fast.

Now hold that chart in your mind and look at the other side of the economy — not at what things cost, but at what assets did over the same half century.

The consequences are exactly what you would predict. The Federal Reserve's own accounts show the top 1% of households now holding a record share of the nation's wealth, while the entire bottom half — 65 million families — holds about two and a half cents of every dollar. A generation ago it took roughly twenty years of saving from an ordinary salary to climb from the bottom quarter of the wealth distribution to the top quarter. Now it takes forty. A working lifetime, spent climbing a single ladder.

Here is the part that should unsettle you most. Americans are not confused about any of this. They can read the scoreboard perfectly. Over half of Gen Z now says a full-time job cannot get them to their goals. As of 2020, the single largest bloc of American voters became people who vote while believing their vote changes nothing. Young people pour into sports betting and meme stocks and options — not because they are foolish, but because when the house is priced out of reach, a lottery ticket is a rational instrument. As the economist Kyla Scanlon put it, people are pattern-matching to the environment they're actually in: the downside of American life arrives through prices, and the upside arrives through asset appreciation. If you own assets, the country is working. If you only earn wages, it is working on you.

Building wealth in America has become an active discipline. It requires tax strategy, financial planning, investment management, estate design — the exact suite of services that has always existed, and has almost never been available to the people who need it most. There are not enough experts, the experts are expensive, and the economics of advice have always pointed the professional toward the client who already has millions.

The scarcity of financial expertise is the bottleneck on the American Dream. Hold that thought. Because at this precise moment in history, humanity is spending more money than it has ever spent on anything to make expertise abundant — and aiming almost none of it at this problem.

A Trillion Dollars in Search of a Purpose

I have spent the last several years of my investing career inside the AI buildout — leading investments in AI infrastructure before it was consensus, sitting on boards, watching thousands of companies form and pitch and pivot. And I want to say something that is uncomfortable to say from inside the industry: most of the capital in AI is flowing in the wrong directions, for the wrong reasons.

Hyperscaler capex is headed toward $1.5 trillion a year. Every dollar of it will eventually demand a return on capital, and the current answers for where that return comes from are thin. The fashionable formula says AI success means selling one of three things — compute, tokens, or data. Each has a ceiling built into its foundation.

Compute is a commodity. It is the picks-and-shovels trade of this cycle, and like every picks-and-shovels trade before it, it ends in oversupply and compressed margins. Tokens are cheap — and the honest evidence, including the frontier labs' own published task economics, shows that most tasks current AI completes end-to-end are worth tens of dollars, not thousands. That is why the visible winners are high-volume, low-value automation: support tickets, IT tickets, transcription. Real businesses — but they are harvesting the shallowest water in the economy. And data vendors sit on the wrong side of a wall: built horizontal to maximize their addressable market, they can never hold the vertical depth — the domain context, the vetted experts, the earned judgment — that world-class work in any single field demands. Meanwhile the neolabs acquire nothing but talent, returning zero to the investors and employees who built what was bought.

So where is the value? What is actually missing?

What is missing is everything that happens after the model. Diffusion. Adoption. The human–AI interaction pattern for each craft. The operational redesign of how work flows between person and machine. The reinvention of org structures and incentive structures so that workers are aligned with the systems they train. The transformation of business models — cost structure, throughput, pricing, and ultimately the redefinition of what a client receives for their dollar. The industry has built engines of astonishing power and almost no drivetrain. We know how to make intelligence. We have barely begun to learn how to turn intelligence into outcomes — how to travel the full chain from compute to tokens to completed work to a measured improvement in a real business and a real life.

And here is why nobody does it. Rebuilding an industry around AI destroys the existing market structure, and the people with the power to authorize that destruction are the ones the current structure made powerful. They are at the end of their careers. They are paid on this year's realization rates. They have every incentive to bolt a copilot onto the old machine, issue a press release, and retire before the bill comes due. The consultants and forward-deployed engineers now swarming the Fortune 500 are not the revolution; they are a patch, serving the old structure precisely because the old structure is what pays them.

The companies that will define this era will not sell intelligence to incumbents. They will use intelligence to become what the incumbent should have been — and they will do it in industries where the human stakes are highest. Which brings me back to my parents' kitchen table.

Why Tax, of All Things

Tax is personal to me twice over. My spouse is a tax CPA — the most genuinely helpful person I have ever known, the one every friend of ours calls in March, the kind of professional who treats a client's return as a promise. Through her I learned what that profession really is underneath the stereotype: it is the closest thing an ordinary family ever gets to a financial confessional. Your accountant sees everything — every income source, every life event, every fear about the future — once a year, by law, forever.

And then my mother got cancer. And the question she brought to me was not about treatment. It was about taxes — how to protect the income and wealth she had worked her entire life for, so that as much as possible would pass to my sister and me. I sat with that question for a long time. A woman facing her mortality, and what the American system handed her to think about was the structure of her filings. Nobody had ever helped her plan. There was never anyone whose job it was.

So when I say APEX begins with tax, understand that I did not select it from a market map. But I have also done the market map — obsessively — and everything in it confirms what my life already taught me:

Tax is the profession where the machines can do the most, the workforce is vanishing the fastest, and adoption has barely begun. The most time-consuming work in tax — document extraction, classification, rules-based logic — is precisely what modern AI does best. But the deeper reason tax is the wedge is what it earns. A tax relationship produces the best data asset that exists on a client's financial life, refreshed every year because the government requires it. Do that work with genuine excellence — accurate, fast, humane — and you have earned something no marketing budget can buy: the trust to advise. Trust is the bridge from a $400 return to financial planning, to wealth management, to the full suite of services that my parents never had — for the family that was never anyone's ideal customer. Tax is how you earn the right.

And there is no fear of running out of demand. This is Jevons' paradox applied to expertise: when the cost of world-class financial guidance collapses, consumption of it explodes. Tens of millions of American households who have never once had a real advisor become, for the first time, servable. America minted more new millionaires last year than the rest of the planet combined. The question of our generation is whether the next wave includes the nurse, the contractor, the teacher, the immigrant with a small business — or only the people who already own the index.

The Firm We Are Building

APEX is not a software company selling tools to accountants, and it is not an accounting firm sprinkling AI on old workflows. It is a third thing, built from a blank page: an AI-native financial services institution that delivers the work itself — tax, then planning, then wealth — and improves with every client it serves. It runs on a small set of principles I will not compromise:

The AI system is in service on day one. Not piloted, not "explored" — deployed in live delivery with humans in the loop, inside software and infrastructure built to capture data through the activity of the work itself. A firm that adopts AI later inherits its old habits; a firm born with it inherits nothing.

Every workflow is dissected to its atoms and rebuilt. We do not automate the old process; we ask what the optimal division of labor between human and machine actually is for each atomic task — and we let that division evolve as the models improve, quarter after quarter, forever.

A language model never computes the final number. Deterministic calculation engines produce every figure that touches a client's return. AI reads, classifies, drafts, flags, and reasons; verified arithmetic and encoded law decide. In a domain where trust is the entire product, this is not a technical preference. It is a moral one.

Judgment is the moat. Every time one of our professionals reviews, corrects, or approves the system's work, that judgment call is captured. The result is an adjudicated dataset of expert decisions — the full trace of how excellent tax and financial work actually gets done — that compounds with every client and every return. Models are becoming commodities. A decade of accumulated professional judgment, structured for learning, is not.

Humans operate at the top of their license. As the machine absorbs the mechanical, our people rise to the judgment, the relationship, the moments that matter — the conversation with a mother facing cancer, which no model should ever hold alone. Low-skill work commoditizes; rare, complex, human work commands a premium. We are building the firm where professionals do the most valuable version of their craft that has ever existed.

I know what I bring to this and what I don't, so let me state the founder's case plainly. I have the context — a decade-plus and thousands of companies deep in how AI value actually accrues, married to a household that lives the tax profession. I have the conviction and the track record of backing the non-obvious before it was consensus, when the market was small and the crowd was elsewhere. I have the creativity to design an institution that doesn't yet exist rather than optimize one that shouldn't. And I intend to build a culture with an unfashionable ordering: client first, employee first — in the deep belief that if you get those two right, the returns are a consequence, not a target. What's left is courage, and courage is a choice. I've made it.

Who This Is For

Manifestos are usually written for everyone, which means they are written for no one. This one is written for four specific people, and if you are one of them, you will know before you finish the sentence.

To all four, one sentence, which is also our test for every hire and every decision: do something that would make your parents proud. Not impressed — proud. There is a difference, and everyone who has ever watched their parents work too hard knows exactly what it is.

And I will be honest with you, because a manifesto that hides the cost is an advertisement. This will be harder than the alternatives. Services businesses are unforgiving; trust is earned by the return, every return; the compounding is slow before it is sudden. If you want the fastest personal path to a windfall, the casino economy is right over there, and it is very busy. This is the other thing — the long thing. The thing you tell your grandchildren you built.