AI data centers now consume roughly 27 tons of copper per megawatt for power delivery, cooling, and interconnects. The U.S. added about 11,300 megawatts of data center capacity last year alone. That surge helped create a projected global copper deficit of 320,000 tons in 2026.
Copper prices climbed, lifting shares of miners such as Freeport-McMoRan (NYSE:FCX | FCX Price Prediction). Yet the very speed AI demands exposes copper’s limits. Electrical resistance caps reach, drives up power use, and generates heat that data centers can no longer afford at scale. Optical photonics — transmitting data as light — sidesteps those constraints. It delivers higher bandwidth over longer distances with lower energy and less heat. That shift opens a multi-billion-dollar window for companies that can deliver both copper and optical solutions.
Copper Hits the Wall as AI Scales
Let’s look at the numbers. Short-reach copper interconnects still dominate inside AI racks because they are cheap and reliable today. But as clusters grow to hundreds of thousands of GPUs, the physics no longer work. Copper cables lose signal integrity beyond a few meters at 800G and 1.6T speeds. Power consumption per bit rises sharply. Cooling demands explode.
According to S&P Global’s January analysis of copper use in data centers, substitution options remain minimal in dense server environments — aluminum cables simply take up too much space and conduct heat poorly. Investors who rode the copper wave via copper miners saw gains from the metal-price spike, but the long-term fix lies in light, not more metal.
Credo Delivers Speed Today — and Preps for Tomorrow
Credo Technology Group (NASDAQ:CRDO) started by solving the immediate problem. The company designs high-speed SerDes chips and Active Electrical Cables (AECs) that extend reliable copper reach inside AI clusters while cutting power roughly in half compared with passive alternatives. Demand exploded.
According to Credo’s March earnings release, third-quarter fiscal 2026 revenue reached $407 million — up 201.5% year over year and 51.9% sequentially. That growth outpaced the broader semiconductor sector and reflected hyperscalers snapping up AECs for massive GPU clusters. The stock reflected the enthusiasm: it climbed to an intraday high of $213.80 in early December. Then reality set in.
Copper-versus-optics debates, broader market rotation, and profit-taking drove shares down more than 50% to a low near $87.81 by late March 2026. Smart investors who bought the dip have since watched the stock more than double from those lows as the AI buildout continued.
Here’s what the numbers tell us about Credo’s position. Trailing 12-month earnings per share stand at roughly $1.92. At recent prices around $160, the shares trade near 84 times trailing earnings — steep, but typical for a company posting triple-digit revenue growth. Cash and short-term investments ended the quarter at $1.3 billion, giving the company ample runway to invest without dilution pressure.
DustPhotonics Deal Positions Credo for the Optical Upside
Last week, Credo announced it would acquire DustPhotonics, a leader in silicon photonics photonic integrated circuits, for $750 million in cash plus about 0.92 million shares and up to 3.21 million additional shares tied to milestones. The deal is expected to close in the second quarter.
DustPhotonics creates a vertically integrated stack: Credo’s SerDes and DSP chips now combine with in-house silicon photonics for optical engines, transceivers, and system-level solutions. Management expects the move to drive fiscal 2027 revenue growth above 75% year over year and to become accretive to non-GAAP earnings per share that year. Optical revenue alone could top $500 million in fiscal 2027, expanding Credo’s addressable market into longer-reach interconnects where copper simply cannot compete.
Granted, competition exists. Established optics suppliers already ship pluggable transceivers and are pushing co-packaged optics. Integration always carries execution risk, and the stock’s valuation leaves little room for missteps. That said, Credo’s proven track record scaling AECs gives it credibility with the same hyperscale customers who will buy its optical products.
The acquisition does not abandon copper — it complements it. Short-reach racks stay copper-based while optics take over the fabric that connects them.
Key Takeaway
Credo Technology built a leadership position in copper-based AI connectivity and just spent $750 million-plus to own the optical layer that solves copper’s limits. The stock has already proved it can double from sharp sell-offs when growth reaccelerates. If the company hits its fiscal 2027 targets and captures even a modest slice of the expanding optical TAM, the upward trajectory can continue.
Savvy investors should expect volatility, watch the DustPhotonics integration, and treat this as a multi-year AI infrastructure bet rather than a short-term trade. The copper crunch is real. Credo just bought the flashlight to see past it.