Celestica looks misread on this selloff. A stock that just pushed its 2026 outlook to $19.0 billion in revenue from $17.0 billion and adjusted EPS to $10.15 from $8.75 is not acting like a business with a broken growth engine. The market is punishing CLS for lofty expectations, not deteriorating fundamentals. We think that distinction matters because the AI networking story is no longer theoretical; it is already flowing through revenue growth, product availability, and named partner programs.
The cleanest reason to stay bullish is that management already did the hard part: it raised the bar. Q1 2026 revenue came in at $4.05 billion, and the company said stronger Q2 customer demand plus better visibility for the rest of 2026 supported that much higher full-year outlook. That is not a token guide-up. Moving annual revenue guidance by $2.0 billion and adjusted EPS by $1.40 in one shot says demand is exceeding what Celestica itself expected just a quarter ago.
The operating trend underneath that guide is even stronger than the headline selloff suggests. Q1 revenue grew 53% year over year, while market revenue jumped 69% on data center networking demand and an AI/ML compute program with one hyperscaler customer. The broader TickerSpark Score backs that up: CLS posts an 82 overall score, with a perfect 100 in Growth and 100 in Momentum. Even on trailing figures, revenue growth is 30.7% and EPS growth is 103.0%, which is exactly what a real AI infrastructure winner should look like when the cycle is working.
The second reason the thesis still holds is that Celestica's AI pipeline is getting tangible, not more promotional. The DS6000-series 1.6TbE switches are now available to order for initial customers, giving the market an actual product milestone instead of a vague roadmap. Celestica also partnered with AMD on the Helios rack-scale AI platform, where it is handling the R&D, design, and manufacturing of scale-up networking switches. That combination matters because it ties CLS to both current hyperscaler demand and the next wave of AI cluster buildouts, while the broader Ethernet market is still expanding fast enough to support the story.
The obvious knock is valuation. At 48.5 times trailing earnings and 35.57 times EV/EBITDA, CLS is not cheap, and a business with an 11.6% gross margin does not get much room for execution mistakes when the market has already priced in a lot of future success. Bulls also have to accept that customer concentration is real here; hyperscaler spending can be lumpy, and one pause in AI capex would hit sentiment hard.
That said, expensive stocks can still be mispriced when growth is outrunning the multiple, and CLS has a 0.35 PEG alongside 103.0% EPS growth. The peer backdrop helps too: KEYS trades at a higher P/E despite much slower 8.0% revenue growth, while NOK carries a far weaker growth profile with a richer earnings multiple than many investors would expect. The risk is not that Celestica lacks a business case; the risk is that the market has become impatient with a company that is still executing.
What matters now is whether Celestica keeps converting AI enthusiasm into reported numbers, and the setup still favors that outcome. The stock remains above its 50-day and 200-day moving averages, has outperformed the technology sector by 9.8 percentage points year to date, and still carries a Buy consensus with 17 buys against zero sells. We would treat this drop as a reset inside an intact uptrend, not a reason to abandon the name.
What would change our mind is straightforward: a slowdown in hyperscaler-driven networking demand or a failure to support that newly raised 2026 outlook at the next report around late July. Short of that, the current weakness looks like the market fading a winner because expectations got too hot, even though the business keeps giving it reasons not to. We stay bullish on CLS because the numbers still say the AI networking trade is alive.
