Q2 earnings season “breaks records” — but the real 2H map is the AI-and-guide-down mix Cisco-style

FactSet’s blended S&P 500 earnings growth for Q2 2026 landed at 50.4% year over year, a level that immediately frames the whole season as “the best on record.”

But the distribution matters more than the average. FactSet also shows that once you exclude Alphabet and Amazon.com—two mega-caps that can swing EPS through large surprise effects—the blended growth rate falls to 32.0%. That gap is the first warning that “record” is not the same thing as “broad and durable.”

The market reaction you saw across the strongest AI-adjacent names wasn’t “beat = go up.” Instead, investors repeatedly punished companies when next-quarter guardrails implied margins or demand would normalize sooner than bulls expected.

A useful way to read the tape is this chain: Beat-and-selloff tells you the market needed more than the quarter. Guide-down (or any downgrade in the direction of expectations) tells you the marginal buyer is underwriting 2H risk. And because AI buildouts are capital- and timing-sensitive, that underwriting tends to show up in forward revenue/EPS ranges and in segment mix (data center, networking refresh, storage intensity).

AMD reported Q2 2026 revenue of $11.5B, up 50% year over year, with Data Center revenue of $6.7B (+107% YoY), indicating the earnings engine is AI/data-center concentrated. However, a beat alone doesn’t change what investors paid for: they still have to underwrite sequential progress through the next reported quarter.

Cisco similarly posted Q2 FY2026 results that beat the season’s mood—Q2 revenue of $15.3B (+10% YoY) and GAAP EPS of $0.80 (+31% YoY). The market’s attention shifted to forward guardrails: Cisco’s Q3 FY2026 and FY2026 guidance frameworks become the “how much of the beat is one-time” test.

Applied Materials delivered Q2 FY2026 revenue of $7.91B (+11% YoY) and GAAP EPS of $3.51 (+33% YoY), showing the chip-capex tool channel is still participating. The “durability” question is visible in the next-quarter framing: Applied Materials guided Q3 FY2026 revenue around $8.95B (±$0.5B) and non-GAAP EPS around $3.36 (±$0.20). That’s the type of forward range investors use to decide whether the AI buildout rhythm stays intact.

Seagate similarly reported fiscal Q2 2026 revenue of $2.83B with GAAP diluted EPS of $2.60 and guided fiscal Q3 2026 revenue to about $2.90B (±$0.1B) and non-GAAP diluted EPS to about $3.40 (±$0.20). When storage demand acts like a “downstream amplifier,” investors care less about one quarter’s bounce and more about whether next-quarter targets reflect normalization.

Q2 earnings season ending with a ~50% S&P 500 growth print makes for an easy headline. But the investor-grade insight is that concentration—visible even in FactSet’s ex-Alphabet/Amazon adjustment—means the “record” is not broad-based by default.

In the AI cluster, the more consistent signal is the combination of a strong quarter plus forward guardrails that still leave room for doubt. That’s why the next step for 2H positioning is not chasing the beat; it’s modeling whether the forward EPS and revenue ranges embed durability or normalization pressure.

In other words: the quarter says “the cycle is on,” and the guide says “how fast it cools”.