Financial markets were dominated by AI, tariffs, data centers, “Liberation Day,” and Fed independence throughout 2025. Investors also noticed (and bought) gold. Lots of it.

In 2025, gold prices rose roughly 64%, finishing the year at about $4,550 per ounce, noted a story by Reuters. That return bested nearly all developed and emerging market equity indices, fixed income, and even digital assets (bitcoin declined by about 6% over the same period), according to Kiplinger. Headlines celebrating gold’s resurgence revived a familiar narrative: gold as the ultimate store of value—steady, timeless, and immune to AI.

The Warren Buffett Gold Cube Thought Experiment

This enthusiasm is not new. Warren Buffett addressed this exact debate 15 years ago in Berkshire Hathaway’s 2011 shareholder letter, where he introduced one of the most quietly devastating thought experiments in modern finance. No charts. No regressions. Just arithmetic—and a very large cube.

At the time (year-end 2011), Buffett noted that all the gold ever mined, approximately 170,000 metric tons, could be melted into a cube measuring approximately 68 feet on each side. At then-prevailing prices of roughly $1,750 per ounce, that cube was worth approximately $9.6 trillion.

Buffett labeled this Pile A.

He then asked readers to imagine Pile B: using the same $9.6 trillion to purchase all U.S. farmland, plus 16 ExxonMobils, with roughly $1 trillion left over (presumably for transaction costs, or perhaps a storage unit). His point was not that gold could not rise in price but that productive assets create or do something while you own them, whereas gold’s primary contribution is to exist and remain gold.

What Has Happened Since?

Fast forward to year-end 2025. Gold’s price climbed to approximately $4,550 per ounce, pushing Buffett’s original gold cube to a value of roughly $24.4 trillion. That equates to a total return of about 154%, or approximately 6.9% annualized over 14 years.

A respectable outcome for an inert metal!

Revisiting Pile B, however, highlights the difference between appreciation and compounding.

• Farmland: The United States contains approximately 880 million acres of farmland. At an average value of about $4,350 per acre in 2025, total U.S. farmland is valued at approximately $3.8 trillion (figures from the U.S. Department of Agriculture). This reflects significant appreciation since 2011.
  More significantly this figure excludes the compounding effect over the last 15 years from the income generated from crops, rents, or the occasional surprise soybean boom, and the ability to deploy that cash into more farmland, or other productive pursuits. Buffett estimated that at $200 billion per year, which, assuming a 3% growth rate, would have added another $3.4 trillion.
• ExxonMobil: Its market capitalization increased from approximately $238 billion at the end of 2011 (data thanks to Macrotrends) to about $525 billion at the end of 2025, according to CompaniesMarketCap. Sixteen ExxonMobils would therefore have appreciated from roughly $3.8 trillion to approximately $8.4 trillion, a 221% jump. More quietly, but no less importantly, ExxonMobil produced cash. From 2012 through 2025, the company paid approximately $260 – $270 billion in cumulative dividends, according to Macrotrends. Scaled across 16 equivalent holdings, that amounts to roughly $4.2 – $4.3 trillion in cash distributions. These arrived regularly, without requiring anyone to visit a pawn shop.
• Cash: While Buffett claimed we would have $1 trillion in cash left over, he was a bit off. Using 880 million acres (instead of his 400 million), and the true average price according to the USDA of $2,980, the total cost of all U.S. farmland would have been around $2.6 trillion. Buffett was willing to pay up for farmland, around $4.8 trillion. So in our case, instead of just having $1 trillion we would have actually had $3.2 trillion. Now we’re no fools, we would have put that in 20-year Treasuries yielding around 4.1%. Which means our pile of cash would have kicked out an additional ~$130 billion each year. The value of the Treasuries would still be around par, so we can ignore any principal change.

Putting It Together

By the end of 2025, a reasonable estimate of Pile B’s value includes:

• Farmland market value: ~$3.8 trillion
• Farmland income: ~$3.4 trillion
• ExxonMobil equity value (16×): ~$8.4 trillion
• Cumulative Exxon dividends (16×): ~$4.2 trillion
• Cumulative Cash Yield: ~$1.8 trillion
• Cash: $3.2 trillion

That totals approximately $24.8 trillion, excluding reinvestment of dividends, farmland income, fixed income yield, and productivity improvements over time.

By contrast, the value of the gold cube of approximately $24.4 trillion was powered entirely by price appreciation, most of which happened in 2025 (at year’s end 2024, that cube of gold would have been worth around $14.5 trillion).

The Growing Cube

Over the same period, continued mining expanded the total above-ground gold stock to approximately 218,000 metric tons, according to the World Gold Council. If all gold ever mined today were formed into a single cube, it would measure about 73 feet on each side. At $4,550 per ounce, that cube would be worth roughly $31.9 trillion.

Gold’s ascent has been impressive. Structurally, however, it remains unchanged. It produces no income, throws off no cash flow, and politely declines to reinvest on your behalf. As Buffett put it, “If you own one ounce of gold for an eternity, you will still own one ounce of gold at its end.”

The long-term return is also constrained, at least in part, by production economics. Industry estimates place all-in sustaining costs (AISC) for gold miners at roughly $1,500–$1,600 per ounce in 2025, according to the gold council, meaning higher prices tend to summon additional supply. Gold has many virtues, but scarcity responds poorly to incentives.

For comparison, in January, the estimated average mining cost per bitcoin was about $100,000 compared with a price of roughly $90,000, according to MacroMicro. Even in digital assets, it turns out, costs still matter.

What Could Today’s Gold Pile Buy?

Buffett’s original thought experiment worked because it reframed portfolio construction in terms of opportunity cost, not sentiment. Updating the exercise shows how large the numbers have become.

At the end of 2025, the world’s above-ground gold stock, worth approximately $31.9 trillion, could theoretically purchase one of the following:

• All U.S. farmland nearly eight times (or one Greenland?), valued at roughly $3.8 trillion
• More than half of all publicly traded U.S. equities, with total market value near $55 trillion (data from S&P Dow Jones)
• Nearly the entire U.S. Treasury market outstanding, excluding intra-government holdings, totaling approximately $27 trillion (figure from the U.S. Treasury Department)

As before, these transactions are hypothetical. The point is not feasibility; it is perspective. Owning the gold cube itself would still produce nothing. Converting it into productive assets would change its nature entirely.

Gold may shine. But it still does not compound.

Epilogue

Warren Buffett derided owning gold and laughed all the way to retirement. Berkshire Hathaway stock traded at approximately $120,500 per share at year-end 2010 (and closer to $122,000 when the 2011 letter was published). By year-end 2025, it traded at roughly $753,900 per share.

That represents a total return of 625%, achieved without dividends, and a compound annual growth rate of approximately 13% over 15 years, outpacing both Pile A and Pile B.

Berkshire also ended 2025 with roughly $385 billion in cash, which Buffett has described, in effect, as patience with a balance sheet.