Copper to Fiber Part 2
In Part 1 we laid out the case for the copper-to-fiber transition via the physics forcing function, the superlinear scaling of connectivity costs, and the TAM repricing that we believe is still underappreciated. Part 2 is our continuation where we connect the thesis to specific companies, specific timelines, and specific risks.
We profile nine companies across the connectivity value chain, from the switching silicon incumbent that underpins every port shipping today through the photonics pure-plays building the laser and optical switching infrastructure the industry will need by decade-end. For each, we lay out the positioning, the upside case, and the specific dynamics we are tracking to determine whether that upside materializes. We have also run all nine through our proprietary Technical Moat Scorecard using the Hardware/Deeptech weighting framework. The scores range from Exceptional to Moderate, and some of the gaps are revealing.
What became clear as we built the company profiles is that the five major hyperscalers are making distinctly different architecture choices, and those choices have direct implications for supplier exposure. Where NVIDIA is pushing extreme co-design within their vertical stack, we believe the hyperscalers will take a similar approach using supply chain providers to get as close as possible to extreme optomiziation. When it comes to optical specifically, no single connectivity supplier/company has uniform positioning across all five buyers. Google, with TPU v7 specifically, is the furthest along on optical circuit switching. Meta is the most aggressive on co-packaged optics and open networking standards. Microsoft remains the most NVIDIA-dependent. Amazon is pursuing the broadest vertical integration. Understanding which hyperscaler is going which direction, and how those decisions are evolving quarter to quarter, is in our view the most important analytical exercise we have been doing for some time. We map the divergence in detail in part 1.
The positioning framework we developed maps all nine companies across three transition phases, each with different beneficiary profiles. The near-term phase favors companies with volume exposure to the current 800G-to-1.6T ramp and copper extension. The mid-term phase centers on the CPO inflection and the emergence of optical circuit switching beyond a single hyperscaler. The longest-dated phase envisions photonic fabrics and scale-across infrastructure. We also built a positioning map that plots each company on two dimensions: revenue timing and technology breadth, with our moat scores reflected in the visualization.
We close with what we think institutional readers care about most: what breaks the thesis. Six bear cases, each with the specific leading indicators we monitor to detect early evidence. We connect those scenarios directly to company exposure so that a reader who assigns higher probability to, say, copper extension lasting longer than expected can trace that view through to which names benefit and which are impaired. We also identify the four conditions we believe must be demonstrated for the market to modernize the value these names within this cycle and what is a clear TAM expansion for the AI networking category.
SUBSCRIBERS: WHAT’S INSIDE PART TWO
Full data, moat scores, positioning map, and monitoring framework
Section VII | Nine company profiles covering financial snapshots, upside cases, what-to-watch frameworks, and quantified moat scores across differentiated capability, proprietary assets, engineering execution, and scale economics. Includes our updated ZeroFlap Optics thesis for Credo, the Marvell DSP-to-photonics pivot risk, and the read-through from NVIDIA’s $4 billion photonics investment. Several moat scores diverge meaningfully from where we believe consensus is currently positioned.
Section VIII | Twelve open debates whose resolution determines relative winners and losers, covering the cycle-vs-secular question, CPO timing and dilution scenarios through 2028, NPO as a middle path, the linear-drive format competition, OCS disruption potential, ZeroFlap standardization implications for commodity suppliers, the hollow-core fiber opportunity, qualification cycle realism, and the ASP compression versus scarcity persistence divide.
Section IX | Six bear cases with specific leading indicators for each, covering scenarios where copper extension lasts longer than expected, CPO yields disappoint, hyperscaler vertical integration accelerates, and Chinese module houses compress pluggable pricing faster than our base case assumes.
Section X | Our three-phase positioning map across Near-Term Volume, Technology Inflection, and Structural Buildout, with the full Company Exposure Matrix mapping all nine names across scale-up, scale-out, scale-across, AEC/ACC, LPO/LRO, NPO, CPO, and OCS, plus the Leading Indicators Dashboard of nine data points we track quarterly to gauge transition pace and direction.
Section XI & Appendix | The full Technical Moat Scorecard with numerical scores and sub-metric rationale for all nine names, and the complete methodology appendix including our cluster BOM model, power model assumptions, and the key estimates ledger identifying where uncertainty is highest and where a different assumption most changes the investment conclusion.