Been looking back at what happened with AI and technology stocks last year, and there's actually some solid lessons worth revisiting. When the market got shaky in early 2025, a lot of people panic-sold some genuinely strong names. The reality? Buying quality tech during pullbacks has historically been one of the most reliable plays for long-term investors.

The macro backdrop was actually pretty straightforward if you looked past the noise. Earnings were solid and the Fed was signaling rate cuts down the line. AI capex spending was accelerating hard - we're talking hyperscalers projected to deploy roughly $530 billion that year, up from around $400 billion the prior year. Taiwan Semi had already raised its 2026 capex guidance to between $52-56 billion. That kind of spending momentum doesn't just disappear.

I remember watching ServiceNow get absolutely crushed down nearly 50% from its January highs. NOW was the kind of technology stocks that got caught in the broader software selloff, but what people missed was how aggressively the company was integrating AI. They'd deepened their partnership with OpenAI, expanded Claude integration through Anthropic, and were literally positioning themselves as the 'AI control tower for business reinvention.' Meanwhile, the company kept posting 21-24% revenue growth and had over 600 customers with more than $5 million in annual contract value.

The numbers were telling a different story than the chart action. NOW grew GAAP earnings 22% to $1.67 per share in 2025, way up from $0.23 just four years prior. The CEO was buying stock personally, which usually means something. If you'd bought at those depressed levels, the math suggested nearly double upside if it just got back to previous highs.

Then there was Celestica - the pick-and-shovels play in AI infrastructure. CLS had absolutely exploded, up roughly 3,000% over five years. But it pulled back about 25% from its November highs, which actually created an interesting opportunity. The company was printing money - 29% revenue growth to $12.39 billion in 2025, with adjusted earnings up 56%. They were planning to invest a billion dollars in capex in 2026 because demand for AI data center infrastructure kept accelerating.

The thing about technology stocks during these kinds of corrections is they tend to overreact on the downside. CLS was trading 50% below its highs at 30X forward earnings - hardly expensive for a company guiding 37% revenue growth and 46% earnings expansion. The stock had found support at key technical levels, and analysts were mostly bullish.

Looking back now, the broader lesson holds: when macro conditions support earnings growth and you've got real capex tailwinds in AI infrastructure, selective buying into weakness usually works out. Both of these technology stocks had the fundamentals to support recovery, they were just temporarily out of favor.