Applied Optoelectronics Slumps 10%, Lumentum Dives 9%, Coherent Drops 6% as AI Optics Profit-Takers Strike

Applied Optoelectronics, Lumentum Holdings, and Coherent declined during Monday trading without a specific negative catalyst. Selling appeared concentrated in the most extended names after a powerful multi-month run. Lumentum had large year-to-date and one-year gains, along with a strong one-month advance. Coherent also showed substantial gains over multiple timeframes and a one-month run. Applied Optoelectronics rose sharply year-to-date. Technical indicators showed overbought conditions, with RSI levels near or above typical exhaustion thresholds. Optical components have been central to AI data center buildouts, driven by demand for high-speed interconnects and the shift from 400G to 800G transceivers, with 1.6T anticipated. Company results reflected strong datacenter and communications revenue growth and optimistic guidance.

"There's no fresh negative catalyst tied to any single name. Instead, momentum traders appear to be cashing out across the AI optical components complex after a powerful, multi-month run. The selling is concentrated in the most extended names, which is exactly where profit-taking tends to bite first."

"The setup heading into Monday was extreme. Lumentum stock entered the session up 163% year to date (YTD) and 1,148% over one year, with a 18% one-month gain already in the books. Coherent stock came in up 107.2% YTD and 385.2% over the past year, with a 24.1% one-month run. Applied Optoelectronics stock had the sharpest ascent of the group, sitting up 377% YTD through Friday's close."

"The technicals were flashing exhaustion. Coherent's 14-day RSI hit 71.4 on May 14 before retreating, while Applied Optoelectronics' RSI peaked at 69.05 on May 13. Both readings signaled overbought conditions just as the rally ran out of fuel."

"Optical components have become the picks-and-shovels play of the AI data center buildout. Hyperscalers training large models need enormous quantities of high-speed interconnects to shuttle data between GPU clusters, and the migration from 400G to 800G transceivers, with 1.6T on deck, has been the focal point."

Read at 24/7 Wall St.