Been watching the market shake out some solid AI plays lately, and honestly, there's some interesting opportunities if you've got a longer time horizon. When the Nasdaq dipped mid-week despite all the noise from geopolitical tensions, the real money was already stepping in to scoop things up. That's the pattern that matters over years, not days.

So here's what's actually moving stocks right now—earnings and interest rates. Nothing revolutionary there, but it's worth zooming in on what's happening beneath the surface. Nvidia's Q4 report basically confirmed what everyone suspected: the AI capex spending isn't slowing down. Taiwan Semi kicked things off in January by raising their 2026 capex guidance to between $52 and $56 billion, crushing their 2025 figure of $40.9 billion. The hyperscalers are throwing roughly $530 billion at capex this year, up from $400 billion last year. That number's probably going higher too.

What caught my eye is how earnings estimates for tech in Q1 2026 jumped to 24% from 18% just a month prior. And get this—15 out of 16 sectors are expected to show year-over-year earnings expansion in 2026. The Fed's still expected to cut rates later this year, which doesn't hurt either.

Let me break down two AI stocks to buy on dips that actually have real fundamentals behind them.

ServiceNow is down nearly 50% from its January highs, which means if it bounces back, you're looking at potential 100% upside from current levels. Wild, right? Most people see AI eating into software companies, but ServiceNow actually adapted early. They've been building AI into their platform for years now. January saw them deepen their partnership with OpenAI to power agentic AI experiences, and they're expanding Claude integration through Anthropic. This isn't a company hoping AI doesn't kill them—they're already building with it.

The numbers back this up. ServiceNow hit $13.28 billion in revenue for 2025, more than doubling from 2021. They had 244 deals over $1 million in Q4 2025, up 40% year-over-year. GAAP earnings grew 22% to $1.67 per share. Analysts are modeling 20% revenue growth for 2026 and 18% for 2027, with adjusted earnings expanding 18% and 20% respectively. The CEO dropped $3 million of his own money into shares recently, which tells you something. Plus they announced another $5 billion buyback program. The stock's actually tripled the broader tech sector since going public in 2012—even with the recent crash. Average price target shows roughly 70% upside from here.

Then there's Celestica, down about 25% from its November peak. Most people have never heard of them, but they're basically the picks-and-shovels play in the AI infrastructure arms race. They design and manufacture the actual hardware—servers, networking switches, data center gear—that the hyperscalers need. Revenue exploded 29% in 2025 to $12.39 billion, more than doubling since 2021. Adjusted earnings jumped 56% last year.

Here's what's interesting: they're projecting 37% revenue growth for 2026 and 39% for 2027, potentially hitting $23.66 billion. Earnings are expected to expand 46% and 43% respectively. They're investing $1 billion in capex this year and funding it internally through operating cash flow. That's confidence. The stock's up roughly 3,000% over five years—crushing the broader tech sector's 100% return. Yes, it pulled back 25%, but it's trading 50% below its highs at 30X forward earnings. That's actually reasonable for a company growing like this.

Both of these AI stocks to buy have real earnings expansion ahead and aren't just riding hype. ServiceNow's integrating AI into enterprise software, Celestica's building the infrastructure everyone needs. When you see this kind of pullback on solid fundamentals, that's typically when patient investors make their moves. The noise clears eventually, but the earnings growth doesn't.