General Motors' record EPS is engineered by a shrinking share count, not a growing business

---

On April 28, 2026, General Motors reported a first quarter that the headlines called a blowout. Adjusted earnings per share of $3.70 beat the prior-year quarter's $2.78 by a third, and trampled the Street's $2.64 consensus by better than 40%. Management raised full-year EBIT-adjusted guidance by $500 million and lifted the adjusted-EPS range to $11.50–$13.50. The stock did what stocks do when a cyclical names a bigger number. And yet, line by line, the quarter tells a quieter and more uncomfortable story than the beat implies. Revenue did not grow; it fell, to $43.6 billion from $44.0 billion the year before. The single largest swing factor in the margin was not a product, a price, or a cost program — it was an accounting adjustment tied to a Supreme Court tariff ruling. And the EPS that everyone cheered was lifted as much by the shares GM has retired as by anything the company actually built and sold.

This is not an accusation of fraud. GM's numbers are audited, its disclosures are thorough, and its operators are competent. The forensic question is narrower and more dangerous to a bull: how much of the per-share growth that the market is paying up for is organic — a structurally better business — and how much is engineered — the same flat earnings divided across a deliberately smaller share count, dressed in a one-time tariff windfall and the absence of last year's write-downs. The answer, line by line, is that the engineering is doing most of the work.

The denominator does the heavy lifting

Start with the arithmetic the headline depends on. GM ended 2025 with roughly 904 million shares outstanding, down from about 995 million at the end of 2024, and roughly 1.2 billion at the end of 2023. Stretch the lens and the picture is even starker: the share count has fallen from about 1.54 billion in 2015 to roughly 0.90 billion in 2025 — a reduction north of 40% in a decade. In 2025 alone, $6 billion of repurchases lowered the share count by double digits. In the first quarter of 2026, GM bought back another $800 million of stock at roughly $75 a share, and still had $5.5 billion left on its authorization, with a fresh $6 billion program approved by the board for 2026.

Now run the counterfactual. Earnings per share is a fraction: profit on top, shares on the bottom. If you shrink the bottom by double digits year after year, EPS rises even when the numerator — actual net profit — is flat or falling. That is precisely the regime GM is in. Net income attributable to stockholders for Q1 2026 was $2.6 billion. The company raised its adjusted earnings guidance, but simultaneously lowered its full-year net-income-to-stockholders forecast to $9.9–$11.4 billion because of special charges. Read those two facts together and the trick is visible in plain sight: the per-share number that investors anchor on is climbing while the GAAP dollars the business throws off are not. The growth lives in the denominator.

This is the denominator illusion in its purest automotive form. A buyback is not value creation; it is a financing decision that redistributes a fixed pie across fewer slices. It can be a perfectly rational use of cash — and for a mature, cash-generative manufacturer it often is — but it is categorically not the same thing as selling more cars at better margins. The market, pricing GM on a forward EPS that the buyback mechanically inflates, is paying an operating multiple for a financial-engineering result. When the buyback slows — and it must, because cash is finite and a softening cycle will demand it elsewhere — the EPS tailwind reverses into a headwind, and the underlying flatness is exposed at exactly the wrong moment.

A one-time tariff ruling masquerading as margin

The second pillar of the beat is even more fragile than the buyback, because it cannot recur. GM's reported Q1 EBIT-adjusted margin carried roughly a 1.5-point benefit from a tariff accounting adjustment tied to a recent Supreme Court decision on IEEPA tariffs — in cash terms, on the order of a $500 million favorable adjustment in the quarter. That is the difference between a good headline and an ordinary one. Management was candid that the underlying business still absorbed a full quarter of tariff costs; the windfall was an accounting true-up, not an operating improvement.

Investors should treat a margin point sourced from a court ruling the way they treat a tax settlement or an insurance recovery: as below-the-line noise, not above-the-line earning power. GM itself guided that gross tariff costs for the full year would still land in the $2.5–$3.5 billion range even after the favorable ruling lowered the expected burden. In other words, the tariff story is a net negative for the year that happened to deliver a one-time positive in a single quarter. Annualize the operating run-rate without the legal windfall, and the margin that thrilled the Street compresses. The quality-of-earnings problem is not that GM lied — it disclosed all of this — it is that the headline number the algorithms traded on bundled a non-recurring legal adjustment into "adjusted" profitability and let the beat speak for itself.

The write-down cycle that never quite ends

Bulls will say the worst is behind GM — that Cruise is closed, the China joint-venture impairments are taken, and the EV strategy has been right-sized. The disclosures suggest the cleanup is still underway. Recall the scale of what came before: in the fourth quarter of 2024 and the period around it, GM absorbed on the order of $4 billion in non-cash restructuring and impairment charges tied to its China joint ventures, plus roughly $500 million connected to shuttering Cruise — the robotaxi unit into which it had poured more than $10 billion since 2016. Those charges helped drive a multi-billion-dollar GAAP loss in that quarter. The Cruise shutdown is real and the promised savings — roughly $1 billion a year — are real. But the cash and the credibility spent to learn that lesson are gone.

And the restructuring did not stop with the 2024–2025 cleanup. In Q1 2026 — the very quarter of the celebrated beat — GM booked an additional $1.1 billion in EV-related restructuring charges, primarily from cancelling supplier contracts. That is the tell. A business that had truly finished re-sizing its EV ambitions would not still be writing seven-hundred-million-and-up checks to unwind supplier commitments in the most recent quarter. The pattern is a multi-year sequence of "one-time" charges that, taken together, stop looking one-time and start looking structural: a company repeatedly over-committing to a future — autonomy, China, electrification at scale — and repeatedly paying to retreat from it. Each retreat is framed as discipline. Cumulatively, it is the sound of strategic bets not working, financed by the cash flow of the truck franchise.

The cyclical engine priced as if it were secular

So where does the actual money come from? Trucks and SUVs, sold in North America, powered by internal combustion. GM's North American operations led the quarter, with adjusted earnings up double digits year over year to roughly $3.66 billion. That is the engine. It is a very good engine — the full-size pickup and large-SUV franchise is among the most profitable vehicle businesses in the world. But it is also among the most cyclical, and it is being asked to carry a valuation premium built on the hope of a smooth EV transition that the same company keeps writing down.

Here is the forensic discomfort. The earnings that fund the buyback, that absorb the tariff costs, that paper over the EV losses, all flow from a product category exquisitely sensitive to the US consumer, to interest rates, to fuel prices, and to a freight-and-housing cycle that drives commercial truck demand. The thesis suggested that GM's earnings are propped by full-size ICE trucks and SUVs into a softening US market — and that is precisely the risk. When pickup affordability strains under high financing costs, when fleet buyers pull back, the most profitable mix-rich units are the first to soften, and the operating leverage that flatters good quarters runs violently in reverse. A cyclical business at the top of its cash-return program, leaning on its most cycle-exposed product line, is the definition of cyclical-priced-as-secular. The buyback that looks brilliant at the peak becomes the capital you wish you had kept at the trough.

China: profitable again, but a fraction of what it was

GM's China story is a case study in how a "return to profitability" can still be a structural retreat. Equity income from China was about $100 million in Q1 2026 — excluding a plant-sale gain — which management framed as the sixth consecutive profitable quarter. That is genuine progress after the brutal joint-venture impairments. But context matters: $100 million a quarter is a rounding error against what China once contributed to GM, back when the country was a pillar of the global growth story rather than a market the company is methodically shrinking into. The JVs were restructured precisely because GM was losing the volume war to domestic Chinese EV makers. Profitable-but-tiny is better than loss-making-and-large, but it is not growth — it is managed decline with a positive sign in front of it. An investor pricing GM on a China re-acceleration is pricing a market GM has, in practical terms, conceded.

The "adjusted" wedge that keeps widening

There is a second number worth watching: the growing distance between what GM calls adjusted earnings and what lands on the GAAP bottom line. In Q1 2026, adjusted EPS was $3.70 while diluted GAAP EPS was $2.82 — a wedge of roughly 88 cents a share, or close to a quarter of the adjusted figure, excluded as "special items." For the full year, the divergence is even more telling: GM raised adjusted earnings guidance to $13.5–$15.5 billion while simultaneously lowering its GAAP net-income-to-stockholders forecast to $9.9–$11.4 billion. Read those two ranges side by side and the message is unambiguous — management expects the special charges to keep coming, and expects investors to keep looking past them. That is the textbook definition of a quality-of-earnings problem. When the adjustments are genuinely one-time, excluding them clarifies the picture. When they recur year after year — China impairments, Cruise wind-down, EV supplier cancellations, plant rationalizations — the "adjusted" number stops being a cleaner view of the business and starts being a more flattering one. A bull is paying the multiple on the top range; a forensic eye underwrites the bottom. The wider that wedge grows, the more the headline EPS becomes a managed narrative rather than a measured result, and the more the burden of proof shifts onto GM to show that the charges it keeps labeling "special" are not, in fact, the ordinary cost of a strategy that keeps not working.

A capital-return clock running against the cycle

There is a timing problem buried in the capital-return story that deserves its own scrutiny. GM is executing its most aggressive buyback program at a point in the cycle when truck demand, financing costs, and consumer health are arguably as good as they will get. A disciplined allocator buys its own stock when the business is cheap and the cycle is early; GM is buying when the stock is cheap but the cycle is late. Those are not the same trade. Deploying $6 billion a year to retire shares feels prudent against a single-digit P/E, but it also drains the optionality a cyclical most needs heading into a downturn — the dry powder to fund a counter-cyclical product push, to absorb a demand air-pocket, or to buy back stock at a genuine trough rather than a plateau. If the US consumer softens as the thesis suggests, GM will have spent its peak-cycle cash retiring stock at $75 that it might have repurchased far cheaper a year later, while the EPS tailwind it manufactured evaporates exactly when it is needed most. The buyback that looks like conviction today can look like mistimed financial engineering in hindsight — and the clock on that judgment is already running.

What the bulls genuinely get right

The bear case has limits, and intellectual honesty requires naming them precisely. GM is genuinely well run, and several pillars of the bull thesis are not spin.

First, the cash generation is real and the balance sheet is strong. GM ended the quarter with roughly $19 billion in cash and a fully funded buyback authorization. This is not a distressed company papering over a hole; it is a profitable industrial returning enormous capital. A buyback executed at a single-digit P/E on a genuinely cash-generative business is accretive — retiring stock at $75 when the business throws off the cash it does is defensible capital allocation, not a gimmick, and bulls are right that per-share value compounds when shares are cheap.

Second, the EV losses really are shrinking. GM's EV losses narrowed by several hundred million dollars versus the prior-year quarter on better volumes, manufacturing efficiencies, and lower fixed costs. Management deserves credit for refusing to chase unprofitable EV share and for rightsizing rather than subsidizing forever. The discipline that produces ugly "one-time" charges is the same discipline that stops the bleeding — and that is the bullish read of the very write-downs the bear case flags.

Third, the truck franchise is a durable, high-margin moat that competitors have spent decades failing to dislodge. North American pickup and SUV profitability is structural, brand-driven, and dealer-reinforced; it is not going away next quarter. And the Supreme Court tariff ruling, however non-recurring as a margin item, genuinely lowers GM's forward cost burden — the windfall is one-time, but the relief is ongoing. Finally, the stock is cheap on conventional metrics, and a cheap, cash-returning cyclical can outperform for a long time before the cycle turns. None of that is fabricated. The bear case is not that GM is a bad company. It is that the price embeds a turnaround narrative the numbers do not yet support.

The quality-of-earnings ledger

Pull the threads together into a single ledger and the picture sharpens. The headline beat rested on three legs: a shrinking share count, a one-time tariff true-up, and the absence of the prior year's catastrophic write-downs. The first is financial engineering. The second cannot recur. The third is a base-effect flattery — comparing against a quarter that was deliberately blown up by Cruise and China charges makes almost any subsequent quarter look like growth.

Now subtract them. Strip the buyback, and EPS growth largely evaporates because net income to stockholders is guided down, not up. Strip the 1.5-point tariff adjustment, and the margin beat narrows toward consensus. Strip the favorable comparison, and you are left with revenue that fell year over year and an EV unit still writing billion-dollar restructuring checks. What remains is a high-quality, deeply cyclical ICE business doing roughly what it did last year, returning capital aggressively into the back half of a long expansion. That is a fine business. It is not a growth story, and the gap between those two descriptions is the asymmetry a short-seller cares about.

The asymmetry, priced

The danger in GM is not that any single quarter is fake — it is that the market keeps extrapolating engineered per-share growth as if it were operating momentum. The setup is asymmetric in the bear's favor for a specific reason: the very levers flattering today's number all reverse together at the wrong time. A softening US consumer hits truck mix. A truck-mix hit cuts the cash that funds the buyback. A smaller buyback removes the EPS tailwind. A removed tailwind exposes the flat numerator. And a recession-wary market re-rates a cyclical down, not up, exactly as the denominator stops shrinking. Each lever is benign in isolation; correlated, they compound. That is how a stock that looks cheap on trailing engineered EPS can be expensive on normalized, mid-cycle, buyback-neutral earnings — and how a "beat-and-raise" can mark a top rather than a breakout.

The kicker

GM did not lie about a single number in its first quarter — every figure here came from its own disclosures and the coverage of them. The company simply let the market do what markets reliably do with a 40% beat: cheer the fraction without auditing the denominator. The forensic case is not that the trucks do not sell or that the cash is not real. It is that a flat-to-down industrial business, leaning on its most cyclical product line into a softening consumer, financed a record per-share number with retired stock and a one-time court ruling, and the market paid an operating multiple for a financing result.

When the buyback that flatters the numerator finally slows into a softening truck cycle, the same arithmetic that manufactured the beat will manufacture the miss — and there will be no Supreme Court ruling left to round the margin back up.