The Q2 2026 earnings season presented an impressive surface tableau: headline earnings-per-share growth was strong across major equity indices, with both the S&P 500 and the Stoxx 600 recording substantial year-over-year gains. Aggregate blended earnings growth for the S&P 500 reached 50.4%, its highest level since Q2 2021 19, while European companies recorded 23% EPS growth 3.
Yet the proper object of inquiry is not the nominal beat rate alone. The utility of EPS growth depends upon its source. Across the reporting season, per-share gains reflected a combination of genuine operating improvement, margin expansion, share repurchases, dilution effects, and one-time accounting items. The central distinction is therefore between earnings growth that demonstrates durable productive capacity and growth that is principally the consequence of financial or accounting mechanics.
The breadth and composition of the earnings beats
The breadth of the outperformance was notable. By early August, 88% of reporting S&P 500 companies had exceeded EPS estimates 3, compared with a five-year average of 78% 19. In Europe, 63% of Stoxx 600 companies beat EPS expectations 3. Conservative guidance contributed to this result, but the evidence also indicates genuine operating leverage in several industries.
ON Semiconductor provides a useful example. Adjusted EPS increased 40% year over year 17, approximately four times the company’s 9% revenue growth 17. Such divergence is consistent with meaningful margin expansion rather than mere top-line momentum. EMCOR Group likewise increased diluted EPS by 34.8% while revenue rose 19.8% 7, and Fortinet’s non-GAAP EPS grew 41% 6. In these cases, moderate revenue expansion was translated into substantially greater per-share earnings through cost discipline, pricing power, or both.
The method of difference, however, requires that such examples be compared with cases in which EPS improved without a corresponding increase in underlying operating capacity. Ziff Davis’s adjusted EPS rose 13.2% even as adjusted EBITDA declined 3.7% 15. The discrepancy was explained entirely by a 13% reduction in diluted shares following aggressive buybacks 15. Tyler Technologies offers a similar, though less pronounced, illustration: EPS increased 5.8% while net income was nearly flat 5. Chemed converted high-single-digit revenue growth into more than 40% EPS growth through repurchases 4.
These examples do not establish that buybacks are inherently unproductive. Repurchasing shares can be an efficient allocation of surplus capital when a company’s intrinsic value exceeds its market valuation. But when the denominator is reduced while operating earnings stagnate or contract, the resulting EPS improvement is less persuasive as evidence of productive progress. It may instead represent a temporary enhancement of the per-share statistic, vulnerable to mean reversion if core operating momentum does not recover 15.
The converse effect was visible where companies issued new equity. Pattern’s adjusted EPS declined 17% year over year despite 16% net-income growth, because its diluted share count doubled following an IPO 13. Intel’s planned $15 billion equity raise threatened approximately 3% dilution 18. These cases demonstrate that capital-structure decisions can materially alter the distribution of earnings among shareholders, even when aggregate profitability moves in the opposite direction.
The problem of adjusted EPS
Adjusted EPS introduces a further difficulty: the reported measure may conceal economically significant costs. Eli Lilly’s reported EPS included a $3.03 charge for acquired in-process research and development, leaving the unadjusted result materially weaker 11. Pfizer’s adjusted EPS remained flat despite 3% revenue growth 8. Novo Nordisk reported a 20% decline in diluted EPS, while adjusted EPS increased 5%; the 30% difference was attributed to a non-recurring impairment 12.
One-time investment gains also affected the aggregate picture. Amazon’s large EPS beat was partly attributed to unrealized gains from its Anthropic investment 1, while Uber’s headline EPS growth benefited from a $1.6 billion equity revaluation 10. When Alphabet’s and Amazon’s investment gains were excluded, S&P 500 blended earnings growth fell from 50.4% to 32.0% 19, and the aggregate earnings surprise declined from 29.2% to 10.9% 19.
The implication is methodological rather than merely semantic. An adjusted figure may be useful when it isolates recurring operating performance, but it may also become an instrument for removing inconvenient evidence. The analyst must therefore ascertain whether an excluded item is genuinely exceptional or simply a recurring consequence of the company’s chosen business model and capital allocation.
Market reactions and the limits of the beat signal
Price action further weakened any simple interpretation of the EPS beat. EMCOR shares rose 17.5% after its result 9, demonstrating that a credible earnings improvement can be rewarded decisively. Tesla, by contrast, fell 15% despite strong earnings, as its spending plans overshadowed the reported result 20. AMD shares also sold off after a 246% increase in non-GAAP EPS, likely reflecting a “sell the news” response to elevated expectations 14.
The statistical relationship between EPS and market value is consequently powerful but not uniform. For LQ45 Indonesian stocks, EPS was the strongest statistical determinant of prices 2. More broadly, EPS remains a dominant fundamental driver in many markets, with a robust statistical relationship 2 and an intuitive connection to shareholder value creation 2. Nevertheless, the market evaluates earnings against expectations, capital requirements, valuation, and the durability of the underlying business—not in isolation.
Implications for NVIDIA
No direct NVIDIA claims are included in this cluster, so the conclusions must be applied by analogy rather than presented as company-specific findings. That analogy is nevertheless material. In an environment of elevated EPS expectations and widespread reliance on repurchases, NVIDIA’s future reports will be assessed not only by the magnitude of any beat but by the quality of the per-share growth.
The relevant syllogism is straightforward. If EPS expands primarily because revenue and operating income are increasing, the result indicates expanding productive capacity. If EPS expands chiefly because the share count is falling or because one-time gains are being recognized, the result is less durable. NVIDIA’s operating leverage and pricing power would therefore be viewed more favorably if its EPS growth materially exceeded revenue growth through genuine margin expansion, as in the ON Semiconductor example 17. By contrast, growth driven by aggressive repurchases or temporary accounting benefits would invite the same methodological skepticism applied to other companies in the cluster.
The market’s response to Tesla and AMD supplies a second caution. Even a substantial earnings beat may be discounted when capital spending, future resource requirements, or already elevated expectations alter the investor’s assessment of intrinsic value. Capital intensity represents a necessary tension between present sacrifice and future improvement; the decisive question is whether the expenditure produces sufficient future utility to justify its cost.
Conclusion
The Q2 2026 season delivered exceptional headline EPS growth and unusually high beat rates, but the quality of those results varied widely. Buybacks and one-time gains inflated many apparent beats, and adjusted S&P 500 earnings growth was 18 percentage points lower after investment gains were removed 19. Share-count management thus remains a double-edged instrument: repurchases can materially increase EPS and support prices in the short term, while dilution from equity issuance can reverse those per-share gains 16.
The proper analytical response is not to reject EPS as a signal, but to deconstruct it. Reported and adjusted figures can diverge substantially, as demonstrated by Eli Lilly and Novo Nordisk 11,12. For NVIDIA and other high-growth companies, the probability of a favorable market tendency will depend less upon the existence of an EPS beat than upon its demonstrated source. Durable revenue expansion, operating leverage, and disciplined capital allocation constitute evidence of genuine economic progress. Financial engineering and transient gains may improve the nominal statistic, but they provide a weaker foundation for long-term intrinsic value.