Celestica's AI Design Wins Bolster Bull Case, but Concentration and Valuation Risks Persist

What happened

Seeking Alpha published a bullish piece on Celestica citing new AI design wins with AMD Helios and OpenAI Jalapeno, asserting that agentic inference is intensifying networking demand and expanding Celestica's content share across scale-out and scale-up architectures. The article argues that the shift towards scale-up connectivity requiring 10x greater interconnect density, coupled with the 1.6T ramp, will accelerate a mix shift toward higher-value HPS sales and drive operating leverage. These developments are consistent with the company's Q1 2026 performance, which saw revenue surge 53% YoY and CCS segment margin improve to 8.6%. However, the DeepValue master report maintained a WAIT rating at $361.50, citing extreme customer concentration (top customer 35% of revenue) and risks from the $1B capex ramp and utility/grid constraints. The new deal wins do not address those structural risks, and the stock's valuation at 43x P/E leaves little room for error.

Implication

For investors, the AMD Helios and OpenAI wins suggest Celestica is deepening its engagement with key AI customers, which could lead to more durable revenue streams beyond the current hyperscaler capex surge. However, these wins do not negate the concentration risk: the top three customers already accounted for 65% of Q1 2026 revenue, and adding more business from a few large players could increase dependency further. The $1B capex plan for 2026 remains a key swing factor; successful ramp could support the bull case, but any utility or grid delays could cause underutilization and margin compression. At 43x earnings, the stock already prices in sustained execution, so even a minor miss could trigger a sharp de-rating. Therefore, we maintain the WAIT rating and would consider adding on a pullback toward $300 if operational metrics remain stable.

Thesis delta

The new deal wins strengthen the bull case by signaling deeper AI integration and potential for content share gains, but they do not alter the risk factors of concentration, capex execution, or valuation. The thesis remains unchanged: wait for evidence that the capacity ramp is on track and that cash conversion improves before paying up.