A Tale of Two Companies in One Stock
Few names in the artificial intelligence space generate as much fascination — or as much confusion — as the so-called neoclouds. These are the companies that have positioned themselves as the picks-and-shovels providers of the AI gold rush, renting out GPU compute to the largest model builders in the world. One company in particular has captured headlines by climbing from $33 to $187 in the span of a few years before settling near $129, with constant news flow propelling its trajectory.
Looking carefully at this kind of business, it is essential to separate three different things: the company itself, the stock, and the underlying financial structure. The company is in the right business at the right moment. Its press releases read like a who's who of the AI economy — Meta, Microsoft, OpenAI — and its product offering, compute capacity built on cutting-edge GPUs and supported by data center availability, is exactly what the market is desperate for. The stock, meanwhile, whips around dramatically with each new announcement, and Wall Street keeps raising price targets; one major bank recently lifted its target from $126 to $155.
But the financial structure is where things get genuinely unique. This business model involves borrowing long-term money to purchase GPUs, then renting those GPUs out for revenue. The central question — the one that separates believers from skeptics — is whether the revenue generated by those GPUs over their useful life will be sufficient to repay the long-term debt used to acquire them. The company itself argues that its GPUs will remain useful for more years than competitors estimate. That is a critical assumption, because if the chips become obsolete faster than expected, the business will be servicing long-term debt on equipment that no longer commands premium rents. Long-term debt and short-term product is a combination that historically does not age well — like fish, this could go stale quickly if the underlying technology cycle moves on before the loans are paid off.
This is why the stock behaves the way it does. It is less a store of value than a store of speculation, in the same way Bitcoin functions. When risk-on sentiment surges and investors want to believe in the AI thesis, the stock races ahead. Big customer announcements and giant backlogs can fuel further enthusiasm. But what is unlikely to materialize anytime soon is earnings. For every dollar that comes in the door, this company is losing roughly $1.30. The interest expense on borrowed money to fund the GPU fleet weighs heavily on profitability, and that is an inherent feature of the business model rather than a temporary inefficiency. It is a high-risk investment, and anyone choosing to own the equity needs to fully understand what kind of risk they are taking on.
The Rotation From Semiconductors to Connectivity
The AI investment narrative has steadily evolved over the past three years. The original story was almost entirely about semiconductors — about the dominant GPU maker and its near-monopoly on AI training silicon. That story is still alive and important, and it has expanded to include other chip designers like AMD, where the focus has shifted toward CPUs as well. But there is now a broader and arguably more interesting frontier: connectivity.
Building modern AI systems is not just about raw compute and power. It is about the ability to scale out, scale up, and scale across — to lash together vast clusters of accelerators and move data between them at extraordinary speeds. This requires optical connectors, transceivers, and most fundamentally, fiber.
Several companies that produce optical components and fiber are quietly emerging as the next phase beneficiaries of the AI buildout. Among them are firms specializing in coherent optical technology, fiber optic networking, applied opto-electronics, and pure-play fiber manufacturers. The category is broad, but the thesis is unified: AI cannot exist without robust fiber connectivity.
Corning's Catbird Seat
One name in particular has rewarded patient investors handsomely. The major fiber optic manufacturer recently broke into new 52-week highs, trading around $183 after touching $195 not long before. Quarterly results have shown double-digit growth, and the bigger picture matters more than the noise around any single quarter's beat-or-miss dynamics.
This company occupies a position in fiber that is structurally analogous to where the dominant GPU maker sat several years ago in chips. They are essentially alone at the high end of their niche. The complexity of manufacturing the fiber required for next-generation AI infrastructure is jaw-dropping — in some ways comparable to the difficulty of building advanced semiconductors — and they are functionally the only company doing it at scale. That market structure has translated directly into strong top-line and bottom-line performance, and it explains why orders are rushing in the door faster than competitors can catch up.
Looking Beyond the Mag Seven
The most crowded research target in the market is the so-called Magnificent Seven. When the entire investing world is focused on the same handful of mega-cap names, the marginal value of any additional analysis on those companies approaches zero. The opportunity for genuine insight lies elsewhere — in companies where dedicated research can actually produce an informational edge over the consensus. Optical infrastructure is one of those places. The businesses are performing, and when businesses perform, the stocks tend to follow.
The Geopolitical Layer Above the Stack
Underneath all of this technological enthusiasm sits a geopolitical reality that dwarfs the quarter-to-quarter movements of any single stock. A high-profile presidential trip to China — accompanied by the chief executives of the largest AI chipmaker and the largest consumer electronics company — carries implications that go far beyond business diplomacy. The new chief executive of the consumer electronics giant in particular has his hands full with the trip's complexity.
Recent global events have changed the calculus considerably. The ongoing war and the disruption to oil markets have had significant effects on China, although China has so far managed to keep oil flowing and has proven to be well prepared for these shocks. But the bigger question concerns military capacity. The United States has used considerable amounts of its military and missile inventory in recent conflicts, and that has direct implications for the country's ability to defend Taiwan should the need arise.
Beijing has repeatedly stated its intent to bring Taiwan back under mainland control, and the leadership in China is plainly thinking about the timeline of that ambition during their lifetime. The military choices made over the past year, and even the past few weeks, affect what options remain on the table for Taiwan. This issue looms much larger than the bilateral relationships of any specific American technology company with China. The semiconductor supply chain, the most critical industrial asset of the AI age, runs straight through Taiwan — and any geopolitical rupture there would reshape the investment landscape overnight.
The Synthesis
What ties all of this together is a clear framework for thinking about where value is being created and where it is being destroyed in the AI buildout. Speculative cash-burning rental businesses face structural challenges that no amount of customer name-dropping can solve. Pick-and-shovel suppliers of irreplaceable physical components — particularly fiber and optics — are cashing in on real demand with real margins. And above it all sits a geopolitical environment in which the entire infrastructure of modern computing depends on the stability of a few key relationships and a few key islands. Investors who keep all three layers in view are best positioned to navigate the years ahead.