Bill Miller on Amazon, Bitcoin, and Buying at a Discount to Future Value

Interview with Bill Miller, legendary fund manager and former chair of Legg Mason Capital Management, conducted by William Green in May 2022. Recorded against the backdrop of sharp tech and crypto declines — the ARK Innovation Fund had retraced to March 2020 levels, and the FAANG-heavy funds of James Anderson (Bailey Gifford) and Dennis Lynch (Morgan Stanley) were down more than 50% from their peaks.

Key ideas

Intrinsic value is backward-looking; future value is what you must price. Every piece of information used to value a business derives from the past, but 100% of the value depends on the future. Classical value investing that anchors on today’s price-to-book or price-to-earnings ratios conflates the measurement with the thing being measured.
Frank Knight’s risk vs. uncertainty distinction. Risk has a known probability distribution and can be priced (insurance, casino). Uncertainty — Keynes’s term — is irreducible: no distribution is knowable. Miller is unusually comfortable with Knightian uncertainty, which is why he holds concentrated positions in assets most institutional managers cannot touch.
Bitcoin as insurance against financial catastrophe. Fixed supply, growing demand, and the conspicuous failure of the venture capital community to bet against it (the Silver Blaze argument) make Bitcoin a rational hedge against the collapse of fiat. Stablecoins represent a shadow-banking risk of comparable scale.
Amazon as the canonical case study in future-value investing. Miller bought the IPO, re-entered at $88 in 1998, averaged down through the 2002 lows, used LEAPS in 2012–13, and held 40–50% of his personal portfolio in Amazon at the time of recording. Each entry point looked expensive on present metrics; each proved correct on future ones.
The 4% principle. All S&P 500 returns over time accrue to roughly 4% of publicly traded companies. The question is not whether a business looks cheap today but whether it is one of the 4%.

Context: May 2022

The interview opens with Miller describing the regime change: the Federal Reserve moving from near-zero rates to aggressive tightening. His historical analogy is 1973–74, when the Nifty Fifty high-growth stocks collapsed from extraordinary multiples, and 1982, when 30-year treasuries reached 10% — at which point, with the S&P on a 22× PE and a 6% long-term earnings growth rate, stocks implied roughly the same 9% return as a risk-free government bond, which Miller read as a signal to rotate from equities. The present situation (2022) is different: the economy remains strong and the Fed is cutting into strength, not distress.

The future-value principle

Miller’s core departure from classical value investing: “100% of the information you have to use to value a business is based on the past, but 100% of the value depends on the future.” Intrinsic value — calculated from earnings, book value, cash flows — is a backward-looking construction. Price-to-intrinsic-value ratios tell you nothing about whether the future will resemble the past.

The implication: what looks expensive on present metrics can be cheap at a discount to future value; what looks cheap can be expensive if the business is in secular decline. Bill Miller bought Amazon throughout its apparent overvaluation because he was discounting what Amazon would be worth, not what it was worth on any observable metric.

This is a refinement of the Value Investing framework rather than a rejection of it: Miller still insists on buying below his estimate of value. The difference is that his estimate of value is forward-looking and includes the full distribution of possible futures — which requires genuine comfort with Knightian Uncertainty.

Amazon: a 25-year case study

IPO (1997): Bought on the premise that internet-delivered books would have structural cost advantages and that Bezos was building something genuinely new.
Re-entry at $88 (1998): After it doubled from the IPO price; already expensive on any present metric.
2002 lows: Stock fell to single digits from $100+; Miller averaged down heavily. The decisive signal came in 2002 when Bezos shifted his language in the shareholder letter from “balance sheet” to “customer experience” — a signal of offence, not defence, at the bottom of a crisis. Amazon’s bond yields were 20–30%; Miller reasoned that if the bonds paid off, the equity would be worth multiples.
LEAPS (2012–13): Bought long-dated call options when the stock appeared overvalued on near-term multiples but the optionality on AWS and Prime was not priced in.
At recording: Amazon remained 40–50% of Miller’s personal portfolio despite year-to-date decline of ~30%.
Dilution resolution: Carol Loomis (Fortune) had warned that even if Amazon succeeded, stock option dilution would transfer the gains from shareholders to employees. Miller complained directly to Jeff Bezos; Bezos invited him to present to the board; the board — which believed tech companies could not compete without options — switched to restricted stock. Amazon’s share count moved from roughly 460 million to roughly 480 million over the following 15–20 years (less than 5% dilution), then began buybacks.

Risk and uncertainty (Frank Knight)

Miller draws explicitly on Frank Knight’s 1921 distinction in Risk, Uncertainty and Profit:

Risk: you know the probability distribution of outcomes. You can insure against it; casinos and actuaries price it. This is the domain where rational expected-value calculations work.
Uncertainty: you do not know the distribution. No amount of historical data resolves it. Keynes called this the domain where “we simply do not know.”

Miller’s claim is that he operates comfortably in the uncertainty domain — concentrating in assets like Amazon in 2002 or Bitcoin today, where the probability distribution of outcomes is genuinely unknowable. Most institutional investors cannot tolerate this: their mandates, benchmarks, and client reporting obligations require risk to be bounded and explainable. Miller’s competitive advantage is accepting the discomfort of not knowing.

See Knightian Uncertainty.

Bitcoin

Miller held approximately 50% of his personal portfolio in Bitcoin at the time of recording (down significantly from peak). His thesis:

Supply and demand fundamentals: Bitcoin has a fixed supply schedule; demand is growing. This is a sufficient condition for long-run price appreciation in the absence of a fundamental reason to exit.
Insurance logic: Bitcoin is an insurance policy against catastrophic failure of the fiat financial system. Governments have repeatedly debased their currencies; Bitcoin’s rules cannot be changed by a sovereign. The expected loss if Bitcoin goes to zero is bounded (you paid the premium); the expected gain if fiat breaks down is very large.
The Silver Blaze argument: In Conan Doyle’s story, the key clue is a dog that did not bark in the night — meaning the intruder was known to the dog. Miller applies this to Bitcoin: venture capitalists invest in every asset class imaginable, yet have not systematically bet against Bitcoin by shorting it or funding serious competition. Their silence is informative — they believe Bitcoin will not go to zero.
Stablecoin risk: The growth of stablecoins represents shadow banking at scale — backed assets of uncertain quality, subject to runs. This is a systemic financial risk that makes the insurance value of Bitcoin more salient.

Pragmatist philosophy

Miller’s investment philosophy draws explicitly from William James’s pragmatism:

Three umpires parable: First umpire calls balls and strikes as they are (naive realism). Second umpire calls them as he sees them (perspectivalism). Third umpire — James’s position — says they are nothing until he calls them. Miller aligns with the third: perception structures reality; there is no purely objective vantage point.
Wittgenstein: The limits of language are the limits of one’s world; concepts that cannot be expressed in the dominant grammar of a discourse (e.g. “Bitcoin has value”) will be systematically misunderstood by those inside the grammar.
“On a Certain Blindness in Human Beings” (James, 1899): An essay on the opacity of other people’s inner lives, which Miller reads as a general epistemic caution: most of what is real about another person — or another asset — is invisible to the observer.

China and Baidu

Miller expressed bullishness on Baidu and Chinese internet stocks (citing Munger’s concurrent view that they are “great and cheap”), discounting the political risk as already priced. The Tencent/Alibaba theses are structurally similar to Amazon in his framing: dominant platforms with enormous long-duration optionality, penalised by Western investors for tail political risks.

James Anderson / Bailey Gifford and the 4% principle

Miller recounts a 2020 Johns Hopkins investment committee meeting where he and James Anderson (Bailey Gifford) discussed value vs. growth. Anderson’s framework: all long-run S&P returns concentrate in roughly 4% of publicly traded companies. 96% of companies, over any long period, underperform cash. Therefore:

Diversification to reduce tracking error virtually guarantees underperformance.
The only question worth asking about any potential holding: “Is this one of the 4%?” — and we know what the characteristics of the 4% are.

Anderson announced his retirement in early 2022, after his fund was up 120% over the prior 12 months. Miller: “I wish I’d retired when I was up that much.”

Career retrospective and retirement

Miller announced in January 2022 that Samantha McLemore and his son Bill Miller IV would take over management of Miller Value Partners funds, with Miller becoming a minority owner. Reasons given: regulatory compliance burden (7-day trading prohibitions on overlapping holdings triggered by one fund’s trades while he held the same stocks on margin in his personal account), desire for reading time (Ulysses, Schopenhauer, Brothers Karamazov), and philanthropic commitments.

His own assessment of his career: the 2009–2020 comeback period — where Miller Value Partners was in the top 1% on 1-, 3-, 5-, and 10-year periods simultaneously — was more impressive than the famous 15-year consecutive market-beating run (1991–2005), because achieving top-1% performance over every rolling period is harder than beating the market index in most years.

The 30-year threshold: a Times of London financial journalist told Miller at a dinner ~10 years prior that surviving and outperforming for 30 years is the statistical threshold at which performance is no longer explicable as luck.

Value Investing — Miller’s future-value approach as a refinement of the classical framework
Knightian Uncertainty — the philosophical underpinning of Miller’s concentrated, uncertainty-tolerant style
Compounding — the long-duration, quality-company logic; Amazon as a compounding machine
William Green — interviewer; RWH host