S&P 500 companies earn 20% stated return on equity but only 12-13% return on capital due to high levels of debt in the capital structure; additionally, much of capital allocation at index companies goes to share buybacks at high multiples, which shrinks equity without creating proportional gains, reducing true reinvestment returns.

If Meredith's firm had simply stuck with the portfolio of holdings from when it started Semper Augustus in the late 1990s without making any changes, results would have been a disaster, proving that buy-and-hold forever is incompatible with long-term outperformance.

Meredith's portfolio trading at 12x earnings at the end of Q1 2025 (up from 10x at end of 2024) has generated 42% returns in 2024 and was up 12-13% in the first two months of 2025, then lost money for three weeks straight in late February, and recovered to be up 6-8% by the time of this recording, while the market experienced the same drawdown but has recovered to be up only 4-5% for the year.

Meredith holds Berkshire in his taxable accounts and relies on never paying a dime of tax through a combination of: never receiving dividends (preferring buybacks), counting on the government not changing the tax code, and counting on receiving a cost basis step-up at death—so he opposes regular dividends but would accept a special dividend.

Meredith's firm practices low turnover (15-20% annually) but uses that activity advantageously by forcing themselves to sell something whenever they buy something, trimming the most expensive corners of the portfolio to buy the cheapest corners, which is a critical discipline that generates alpha.

In periods of high debt and secularly high valuations, value investors must be significantly more active with understanding of valuation and intrinsic value to generate returns, as the passive buy-and-hold approach becomes riskier.

If Buffett had owned the Magnificent Seven in meaningful allocation from 2011 onward, returns would have been excellent, but they weren't expensive when purchased—Apple traded at 10x, Microsoft at single-digit multiples, Google at reasonable valuation—demonstrating that initial purchase price is determinative of long-term returns.

Berkshire's insurance operations are being less aggressive in writing new business (particularly reinsurance and some primary lines) due to late-cycle conditions and excess competing capital, and the company has accordingly reduced underwriting activity to match reduced capital deployment capacity.

Revenues accruing to hyperscalers from AI in many cases are coming from capital investments the companies are making in each other, lending each other money, and borrowing against chips as collateral, rather than genuine external customer adoption, representing artificial internal circulation of capital that masks the lack of external market demand.

The hyperscaler companies have acknowledged among themselves that they probably shouldn't all be spending this much money and don't need five large language models, but each is terrified of being left out and losing competitive position to competitors, creating a prisoners' dilemma dynamic that forces overinvestment.

Apple's current valuation of 35x earnings is unjustified because the company can grow its top line only 6-7% annually and lacks the ability to expand margins much beyond current levels, meaning the stock should not command such a premium multiple given these growth constraints.

Investors deploying capital for new clients should follow the Hemingway principle from 'The Sun Also Rises'—'gradually at first and then suddenly'—by not fully investing capital on day one but rather deploying gradually, allowing the opportunity to buy as valuations evolve.

The only advantage the current hyperscalers have over 1999 fiber investors is the sheer magnitude of their existing profitable operations and balance sheet strength, which means they cannot go bankrupt in the near term, but this does not protect shareholders from significant stock price declines if margins compress and capital returns disappoint.

In the late 1990s, most investors expected future returns of 15-20% per year based on prior 15-20 year performance, but instead experienced losses for the better part of the next 13-14 years, demonstrating how recent experience dictates expectations and causes investors to extrapolate favorable conditions forward.