GM's two earnings numbers, and the stock priced in the gap between them

The advice, as Mary Barra retells it, is almost anti-climactic. Over lunches at a local restaurant, Warren Buffett asked her whether she would do anything differently if General Motors were a private company that did not have to report earnings every quarter. She said no. He said "great" — by which he meant, in her telling, that he wanted her running the company for the long term, not for the next print. She is "forever grateful," she told Fortune this week.

That one question turns out to be the right lens for a stock that has been behaving in a curious way. GM's shares have risen some 56 per cent over the past year and now sit close to their 52-week high of around $88. Yet the company carries the valuation signature of a beaten-up value name, not a winner. The explanation for the mismatch is not buried in a strategy deck. It sits in a single quarter's results, where two earnings figures moved in opposite directions.

Two earnings numbers, one quarter

GM's second quarter, reported in July, is a small study in the difference between reported and adjusted accounting. On the official, GAAP basis — the rules that include one-off items — net income fell 31 per cent to $1.3 billion. On the adjusted basis, the company's preferred view with those one-offs stripped out, earnings per share came in at $3.57, well ahead of the $3.19 analysts expected, and full-year guidance was raised for the second time in 2026. North American profits, the core of the business, jumped 43 per cent.

The two figures are not two opinions. They are two measures of the same quarter, and the gap between them is an electric-vehicle write-down, a charge for a bet being re-timed as American consumers slow their purchase of EVs. The GAAP number falls because it absorbs the cost; the adjusted number rises because the underlying trucks and SUVs keep earning. Which of the two an investor believes is, in practice, the whole of the case.

Backtest: Implement a long-only strategy for GM over the past 2 years. Entry: Price breaks above the 20-day Donchian high. Exit: Price closes below the 200-day SMA, or after 20 trading days, or TP +8%, SL -4%.

The multiple measures the noise

This is where the two multiples belong. The trailing multiple — the share price divided by the earnings of the past year on that official basis — is around 38 to 43, an expensive figure by any automaker's standards. The forward multiple, the price divided by the earnings analysts expect for 2026, sits in the mid-single to low-6s, a cheap one. The same share, the same price, two ratios roughly a factor of six apart in magnitude.

The distance is not a valuation puzzle. It is the accounting noise, made visible. A large EV-related impairment booked in the fourth quarter of last year left GM with a reported net loss of $3.3 billion in that period, which drags the trailing denominator down and the trailing multiple up. The forward multiple, looking past the impairment to a year in which North American profits are recovering, shows the cheap version. If the first-half adjusted run-rate holds for the full year — a large if — the stock yields on the order of 16 per cent in earnings, an unusually rich figure for a share trading at its high. An AInvest aggregate signal assigns the stock a buy rating behind a fundamental score of 9.58, though the contributing analysts and methodology behind that aggregate are not disclosed.

What "run for the long term" actually is

Barra's Buffett line is not a slogan; it is a description of how the capital is being handled. In 2025 GM retired roughly 212 million of its own shares for about $11.7 billion and authorised a fresh $6 billion buyback, while lifting the quarterly dividend. Every share retired makes the surviving ones own a larger slice of the cash flow, which is why the multiple on the remaining equity can look generous even as the price climbs.

The strategy behind that cash is one of patience where the market wanted speed. The EV transition, Barra says, will take decades in the United States, not the aggressive timetable some rivals assumed; in 2025 GM sold more than 700,000 vehicles under $35,000. Europe has been sold, its Opel arm handed to the French group, leaving GM to export rather than fight a loss-making war. China, the harder test, has been restructured, and is — by Barra's account — the only Western carmaker there still gaining share and back in profit. A defence unit selling a squad vehicle to the Army has added roughly $700 million of revenue. The through-line is the same: hold the high-margin truck and SUV core, fund it from its own cash, and let the EV and self-driving stories re-time themselves rather than be bled to hit a forecast.

The trouble is that the low forward multiple is not the market being naive; it is the market being paid for a risk it sees. The EV charge was not a one-off in a quiet year — there was another in the second quarter of 2026. Tariffs, a possible turn in the truck cycle, and a China exposure that has already forced a restructuring are all real, and the reported GAAP line is where each of them would first show up. A capital-rich, share-reducing automaker is a satisfying machine to own, until the transition costs it keeps deferring stop being deferrable.

Barra's answer to Buffett was honest: she would not change her behaviour for the quarter. The investor's question is harder, because the investor cannot stop being exposed to it. The reported number will keep falling on whatever is being re-timed, and the adjusted number will keep rising on the trucks. GM is a company whose management says it is run for the long term, and whose stock is priced — in the high trailing multiple and the low forward one — for the seam between the two. The shares are not cheap because the business is weak. They are cheap because the accounting can keep surprising, and the entire case rests on believing the adjusted line is the durable one.