Bitcoin Is Better Money

Official reserve managers spent two decades building dollar positions. USD-denominated reserve assets rose from about $1T in 2000 to a peak near $3.7T in 2025. Gold sat below that line the entire time, worth roughly $0.25T in 2000 and still only about $1.6T at its 2012 high, before falling back toward $1.0T by 2015.

The relationship inverts at the end of the series. Gold reserve assets reach about $3T in 2025 and a forecast $5.8T in 2026, while dollar reserves ease to $3.4T. The 2026 values are estimated forecasts rather than reported data, and the final move in the gold line is steep enough that the forecast carries most of the crossing.

What matters is not the precise level but the direction of official demand. Central banks accumulating gold are buying an asset no other government can freeze, dilute, or refuse to settle. That is a reserve preference expressed through balance sheets rather than through speeches, and it is the same property that draws large individual holders to Bitcoin. If sovereign balance sheets are already repricing neutrality, the open question for private capital is what else that repricing has yet to touch.

Bitcoin's annualized issuance rate falls to 0.8 percent in 2026, with 20.1 million of the eventual 21 million coins already mined. The mechanism is fixed in the protocol: the block subsidy started at 50 BTC and halves every 210,000 blocks, which works out to roughly four years. The chart marks each step, from 50 BTC through 25, 12.5, 6.25, and the current 3.125, then forward to 1.5625 in 2028, 0.78125 in 2032, and 0.390625 in 2036. By 2040 the cumulative supply curve is visually indistinguishable from the 21 million ceiling. Everything to the right of today is a projection and should be read as one. It holds as long as the consensus rules remain unchanged. What separates it from a forecast of gold output or central bank policy is that no estimate of demand, price, or political will enters the calculation. The path is arithmetic. For someone deploying capital, this removes new supply as a variable worth modeling. At 0.8 percent, halving again in 2028, issuance is no longer a seller of any consequence. Price discovery from here is a negotiation among existing holders rather than a contest between buyers and a supply response. Assets whose supply cannot answer a rising price behave differently, and this schedule is the reason to expect that behavior.

The table grades gold, fiat currency, and Bitcoin against the eight properties traditionally used to define money. Bitcoin earns an A on six of them: portable, fungible, verifiable, divisible, scarce, and censorship resistant. It takes a B on durability, matching gold. Its one D is established history, and that is the row where gold holds the only A on the board. Gold's profile is close to the mirror image. It grades A on fungibility and established history, B on durability, scarcity, and verifiability, but D on portability and divisibility, the two properties that made it impractical as a settlement medium once commerce began moving faster than freight. Fiat currency grades B on portability, fungibility, verifiability, and divisibility, C on durability and established history, D on censorship resistance, and F on scarcity, the only failing grade anywhere in the table. The letter grades are qualitative Early Riders analysis rather than a measured index. What the layout makes clear is where the real disagreement sits. Almost nobody argues that Bitcoin is less divisible or less verifiable than the alternatives. The debate reduces to one row: track record. That is also the only property that resolves through the passage of time rather than through engineering or policy, which means the case gets cheaper to defend every year the network keeps running.

As of July 2026, bitcoin's monthly average price was $63,073, some 45% below the August 2025 peak of $114,868. Network hashrate over the same month averaged 897.9 EH/s, only 21% below its November 2025 record of 1,136.9 EH/s. Price fell roughly twice as far as the capacity securing the chain.

That gap is structural rather than coincidental. Mining capacity is bought with committed capital: machines are ordered months ahead, power contracts run for years, and a facility that is already built keeps running as long as revenue covers marginal electricity cost. Price can reprice in an afternoon; a hosting agreement cannot. The result is a ratchet. Hashrate rises quickly when margins expand and gives back only a fraction of the gain when they compress, which is why the trough after each drawdown has sat well above the prior cycle's peak.

For anyone holding the asset, this is the cleanest available read on operator conviction. The people with the least liquid, most irreversible exposure to bitcoin's future price are not withdrawing capacity at anything like the rate the market has withdrawn its bid.

The market repriced the asset. The people who build the infrastructure largely did not.

The framing treats Bitcoin as a claim on the demand for storing value rather than as a technology bet. Early Riders sizes seven pools: real estate at $370T, bonds at $318T, the stock market at $135T, money at $129T, gold at $22T, fine art at $20T, and cars and other collectibles at $7T. That comes to about $1,000T of assets held at least partly because they are expected to hold value over time.

The assumed capture rates carry the argument, and they are estimates rather than observations. Gold is set at 50%, on the view that Bitcoin's monetary properties make it a close substitute. Bonds and money take 30% each. Stocks and real estate take 15%, since most of their value is productive rather than monetary. Art and collectibles take 5%. The result is $222T of full potential value, with bonds at $95.4T and real estate at $55.5T supplying the largest pieces.

That $222T works out to $10.6M per Bitcoin, roughly 170X the $1.3T market value in August 2026. Nothing in the table attaches a date. What it does is convert an abstract thesis into explicit assumptions a reader can argue with line by line, which is the only honest way to size a position this asymmetric.

Fidelity plotted 25 assets and strategies on annualized return against annualized risk using weekly data from 2020 to 2023. Twenty-four of them form a tight cluster: none returned more than about 20% a year, and none carried more than about 30% annualized risk. The S&P 500 anchors the top of that group at roughly 20% return and 21% risk. Gold sits near 8% return at about 15% risk. Long Treasury is the worst performer at about -13%, with China near -7%.

Bitcoin sits alone in the upper right at about 58% annualized return and about 68% annualized risk, nearly three times the S&P 500's return at more than three times its volatility. The underlying data is Fidelity's, drawn from Bloomberg, Haver Analytics and FactSet as of 10/29/2023.

An asset positioned that far from the cluster cannot be fitted into an existing sleeve. Grouping it with equities understates its volatility, and grouping it with venture understates its liquidity. The practical consequence is that position size, rather than entry price, dominates the outcome. A weight small enough to survive a drawdown of this magnitude still moves the return of the whole portfolio, which is the case for sizing Bitcoin as its own category and holding it across cycles.

A flagship iPhone cost roughly 50,616 bitcoin at the iPhone 5's launch in 2012. By the iPhone 17 Pro in 2025 the same purchase took about 5 bitcoin, a decline of more than ten thousand times in thirteen years.

The gold line is what turns this from a curiosity into an argument. Measured in troy ounces the same phone cost 0.48 ounces in 2012 and 0.46 in 2025, with a brief rise to 0.89 at the iPhone X, whose launch price jumped while gold was relatively cheap. Against a monetary metal that has held its role for four thousand years, a flagship phone costs almost exactly what it always did. The product did not become dramatically cheaper to build and gold did not become dramatically weaker. What changed is the unit.

This is the denominator argument in its simplest available form. Priced in dollars, consumer electronics look roughly flat to modestly more expensive, which is what most people actually experience. Priced in a money whose supply cannot be expanded, the same goods collapse in cost, because thirteen years of manufacturing and semiconductor progress show up in the price instead of being absorbed by the currency. For anyone deciding what to hold, the question is not whether phones got cheaper. It is which unit was telling the truth.

From January 2020 through August 2026, Bitcoin returned about 590% on a cumulative total-return basis. Over the same months SPY and GLD each delivered roughly 150%, while BIL and BND finished close to flat, with the bond fund spending long stretches of 2022 and 2023 below its starting point. No other liquid asset on the chart came within a factor of three.

That 590% is a post-drawdown figure. Bitcoin peaked above 1,100% in 2025 and gave back roughly half of the cumulative gain by mid-2026, and it still ends the period several times ahead of equities and gold. The asset that set the return bar did not clear it smoothly, and the 2021 to 2023 stretch shows the same pattern at smaller scale, with the cumulative line falling from above 500% to below 100%.

Two consequences follow for anyone holding or deploying capital. The near-flat BIL line is a direct measure of what sitting in cash-like instruments cost over these six and a half years. The path, meanwhile, is the real constraint: a position sized so that a 50% give-back forces a sale turns the best performer of the period into a realized loss. Selection was the easy part of this trade.

Measured against bitcoin rather than dollars, the four standard portfolio building blocks have all lost ground since October 2020. SPY sits about 58% behind, gold about 62%, Treasury bills about 79% and aggregate bonds about 83%. More telling than the endpoints is that not one of the four lines has touched zero in nearly six years. The closest any came was late 2022, at the bottom of the last bitcoin drawdown, when SPY had narrowed the gap to roughly 17%. Even that was a deficit. The series then widened again through 2024 and 2025 to about 80% or worse for all four before bitcoin's own 2026 decline pulled equities and gold back toward 60%. Returns are total return from Yahoo Finance adjusted close, indexed to October 11, 2020, with the comparison built by Early Riders. The use of this is not to argue that bonds were a mistake in isolation. It is to establish what the opportunity cost has actually been. If bitcoin is the benchmark a portfolio has to beat, then six years of data say nothing conventional has beaten it, including through a period that contained two severe bitcoin drawdowns. Anyone underwriting a new position should be asked to clear that bar explicitly.

Deflating the Coinbase Bitcoin price by CPI strips out the part of the gain that is only dollar debasement, and what is left is still large. Measured in constant dollars indexed to February 2012, Bitcoin has multiplied real purchasing power 197 times since January 2015, with the series reading 43,715 as of August 2026. The level matters less than the shape underneath it. On a log scale, each cycle low prints above the one before it: the 2018 and 2019 trough held above 3,000 in real terms, the 2022 and 2023 trough held above 10,000, and the pullback from the 2025 high has left the series near 43,700. Repeated drawdowns of well over half the price occurred inside this window without resetting the floor.

For anyone holding Bitcoin, the inflation-adjusted view answers the common objection that the entire move is a nominal illusion produced by an expanding money supply. It is not, at least not over eleven years of monthly data. For anyone deploying capital, the rising sequence of lows argues for sizing a position to survive a full cycle rather than trading around each drawdown. The risk to underwrite is a decline you can hold through, not the loss of the trend itself.

This is a counterfactual, not a record of holdings. It models what would have happened if Apple, Alphabet, and Dell had each directed 1% of quarterly free cash flow into Bitcoin from Q2 2020 through Q2 2024, with series values read from the published Early Riders figure. The combined position reaches about 160,000 BTC by Q2 2024: roughly 88,000 for Alphabet, 59,000 for Apple, and 13,000 for Dell. That combined stack would rank second only to Strategy among corporate holders.

The proportion is what makes the exercise worth running. One percent of free cash flow is inside the rounding error of a treasury policy at companies of this size, well below what any of them spends on buybacks in a single quarter, and it required no change in operating strategy or capital structure. The accumulation is also front-loaded, with the combined line clearing 50,000 BTC by early 2021 while prices were still low.

For anyone advising a balance sheet with recurring free cash flow, the lesson is that meaningful Bitcoin exposure never required a large commitment, only an early and consistent one. The companies that did nothing did not avoid a decision; they made one, and this chart prices it.