Q3 Earnings Season Kicks Off: AI Is Still Early, and Short-Term Bad News Is a Buying Opportunity

On September 30, Q3 earnings season officially got underway, with a wave of companies set to report through October. One pattern is already clear: the companies actually delivering earnings surprises are, overwhelmingly, tied to the AI theme. Over this stretch, the names beating expectations have been heavily concentrated in the AI hardware supply chain.

1. Nasdaq Hits New Highs, With Memory and Semis as the Real Engine

On September 30, the Nasdaq surged over 1% intraday, driven by the strongest earnings performers: memory, semiconductors, and chip names. After the close that day, Micron reported results that beat expectations — and gave guidance worth flagging in its own right: the company expects memory and storage supply-demand conditions in fiscal 2027 and 2028 to be even tighter than in 2026.

That guidance lines up closely with what analysts have been modeling. UBS's supply-demand analysis has shown AI-driven demand for HBM, server DRAM, and enterprise NAND growing far faster than manufacturers can add capacity, with the structural shortage expected to persist at least through mid-2028. Goldman Sachs estimates put the 2026 DRAM supply-demand gap at around 4.9%, with the HBM gap running as high as 50%–60% — and cloud providers have already locked in supply contracts for Q1 2027, willing to pay above current prices to secure them.

From CPUs to DRAM to NAND, nearly the entire compute hardware chain is running in a state of supply shortage, and the root cause is structural: a new fab takes two to three years from groundbreaking to production, which simply can't keep pace with the explosive growth in compute demand from model training and inference. That's why my own read is: over the next three years, earnings for the leading names across this chain are likely to stay spectacular — a genuine supply shortage is about as solid a foundation for pricing power and earnings leverage as it gets.

2. OpenAI's AGI Moment: A New Engine on the Software Side

Beyond the hardware-side supply crunch, there's a new growth engine worth noting on the software side too. In early September, OpenAI released its next flagship model, GPT-6 Astra, with president Greg Brockman declaring "welcome to the AGI era" at the launch. The model posted near-saturated scores across several of the hardest benchmark tests out there.

To me, this news represents more than just another model release — it's a new software-side narrative that can keep driving compute demand, and it's genuinely just getting started. And I'd argue the real scale of future compute demand goes well beyond chatbot-style software applications — it extends into the physical world: medical foundation models, autonomous driving, humanoid robotics, and eventually large-scale manufacturing lines. The compute consumption at that scale could end up dwarfing what today's software applications require.

Seen this way, AI is still genuinely in an early stage of its development. The hardware supply crunch has only just begun, the AGI narrative on the software side is still in its early innings, and the real-world physical applications haven't even gotten started — there's still a very long runway ahead.

3. Short-Term Volatility Is Just Headline Noise; the Long-Term Trend Points Up

Over the long run, markets trend upward and keep making new highs. Short-term volatility mostly comes from headline-driven disruptions — the ongoing U.S.-Iran conflict, swings in energy prices, U.S.-China trade friction, and back-and-forth shifts in Fed rate hike expectations. These factors genuinely move short-term sentiment and capital flows, but they don't actually change the underlying supply-demand logic or industry trajectory behind the AI theme.

4. Through a Swing Trader's Lens: Bad News Is a Buying Opportunity

Back to my own trading framework: from a swing trading perspective, any short-term bad-news headline is, at its core, just pushing a fundamentally sound company's stock price temporarily below its true value — and that is exactly the kind of moment worth considering as an entry point. This is the same standard I've kept coming back to in earlier posts: what matters in a pullback is whether it comes with real damage to the underlying fundamentals, not whether sentiment got spooked. As long as the supply-demand logic behind the AI theme holds and earnings keep confirming it, every round of headline-driven selling is, if anything, building up energy for the next peak.

Closing Note

This post records my own thinking on the current Q3 earnings season and where the AI theme stands. It reflects a personal view and is not investment advice. Market conditions move fast, and any actual decisions should be grounded in your own ongoing research and independent judgment.