Since 2023 the company has retired more than 35% of its shares. When you cut the denominator by a third, per-share earnings can hit records even as the underlying business gets no bigger, and understanding how much of GM's performance comes from that is the key to reading the quarter honestly.
What actually happened
The headline results were genuinely strong. Adjusted EPS of $3.57 beat the $3.13 consensus, on revenue of $48.03 billion against a $46.56 billion estimate. GM raised full-year adjusted EBIT guidance to $14 billion to $16 billion and adjusted EPS to $12 to $14. The CFO called it the best first-half performance in company history.
But underneath the per-share figures, unit sales fell. GM sold roughly 715,000 vehicles in the US in the second quarter, a 4.2% decline from a year earlier. So the company sold fewer cars and reported record per-share earnings in the same quarter. Those two facts are reconciled by two things: higher profit per vehicle, and far fewer shares.
The buyback math, made explicit
The share reduction is not a footnote; it is a central driver. GM repurchased nearly $3 billion of stock in the first half of 2026, bringing total share retirement since 2023 above 35% of fully diluted shares. In the second quarter alone, $2 billion in buybacks retired roughly 25 million shares, bringing the diluted count down 8% year over year to 893 million.
Here is why that matters mechanically. Earnings per share is total profit divided by share count. A company can raise EPS two ways: earn more money, or shrink the number of shares. Both are real, and they are not the same. Earning more reflects a better business. Shrinking the share count concentrates the same business into fewer hands, which raises each share's slice without the pie getting bigger.
With the diluted share count down 8% year over year, roughly 8 percentage points of GM's per-share earnings growth came from the denominator shrinking, before the business earned an extra dollar. That is not deception, buybacks are a legitimate and often shareholder-friendly use of cash, but it means a reader who sees "record EPS" and infers "the business is booming" is partly misreading a financial-engineering effect as an operational one.
What is genuinely operational, and it is real
The important thing is that GM's quarter was not only buybacks. There is real operational improvement here, and it deserves credit distinct from the share-count effect.
The clearest sign is margins. North American EBIT-adjusted margin climbed to 8.6%, up 2.5 points from a year ago, back inside GM's 8% to 10% target range for the first time since tariffs took effect. That is a genuine profitability recovery, driven by disciplined pricing on trucks and SUVs, lower warranty costs, and narrowing EV losses. GM held more than 42% of the US full-size pickup market, its most profitable segment, more than 10 points ahead of its nearest competitor.
The strategy on display is protecting profit per vehicle rather than chasing volume. GM's guidance raise came even as unit sales declined, which signals a company willing to sell fewer cars at better prices instead of discounting to move metal. In a mature, cyclical industry that is arguably the more disciplined choice, and it is the opposite of how Detroit historically behaved. So the margin story is real and creditable. It simply sits alongside the buyback effect rather than replacing the need to separate the two.
The two overhangs clearing at once
Part of why this quarter looked like an inflection is that two large drags are both receding, and it is worth being precise about their nature because both are somewhat temporary tailwinds becoming absent headwinds.
The first is the EV pullback. GM has now recorded $10.9 billion in EV-related charges since the second half of 2025, with about $7.2 billion in cash cost, of which $4.5 billion is already paid, and management called the material cash costs "substantially" complete. The company deliberately cut EV volumes to stop the losses, which pressures revenue but protects margin, and expects EV losses to improve by $1 billion to $1.5 billion this year. The largest cash pain from the EV reset is moving into the rearview mirror, which flatters the comparison going forward.
The second is tariffs. GM's year-ago EBIT was heavily weighed down by tariff costs, and those are now easing as offsets take hold, though management still expects gross tariff costs of $2.5 billion to $3.5 billion for 2026, roughly $900 million in Q2 with similar exposure expected in Q3 and Q4. Management is leaning into onshoring, bringing US production capacity above 2 million units next year, framed as both reducing tariff exposure and hedging trade risk.
The point about both is that a lot of this quarter's year-over-year improvement is the absence of last year's specific pain, not new growth. That is legitimately good, a company shedding one-time costs and recovering margin is worth more, but it is a normalization story, and normalization does not repeat. Once EV charges are fully behind and tariffs are lapped, the easy year-over-year comparisons stop, and growth has to come from the actual car business.
What deserves weight
A few things are solidly true regardless of how one weighs the buyback effect. GM generates strong free cash flow, guided to $9.5 billion to $11.5 billion for the year, which is what funds both the buybacks and the dividend, and a company that can retire 35% of its shares in three years while investing in its business is generating real cash, not manufacturing an illusion. The buybacks are enabled by genuine profitability, not debt.
The margin recovery is real, the pickup franchise is dominant and profitable, and the pricing discipline is a legitimate strategic strength. None of that is undercut by noting that buybacks also helped EPS. The honest read is additive: strong operations plus aggressive buybacks together produced the record per-share numbers, and both contributed.
The useful mental adjustment is to separate the three engines of GM's per-share earnings and weigh them differently. Real operational improvement, better margins on trucks and SUVs, is durable and repeatable, and deserves the most weight. The clearing of one-time drags, EV charges and tariffs, is real but is normalization rather than growth, and does not repeat once complete. And the shrinking share count is a genuine benefit to remaining shareholders but reflects financial engineering rather than a better business, and should be recognized as such rather than mistaken for booming operations.