Fujikura’s AI Boom: The Century-Old Japanese Company Behind a 161% Profit Surge

The most important AI company you may never think about

The artificial-intelligence boom is usually told through the companies selling processors, cloud capacity or consumer software. Fujikura offers a different vantage point: the industrial connections that make those systems work. On November 10, the Tokyo-based manufacturer reported ¥867.433 billion in revenue for April–September 2026, a 55.2% rise, and ¥235.013 billion in operating profit, 160.6% above the same period a year earlier.

That profit increase means operating earnings reached roughly 2.61 times their prior-year level. It is an extraordinary acceleration for a company whose roots predate modern electronics by more than a century. But the deeper story is not simply that demand is high. It is that the physical network inside a computing facility has become a strategic piece of industrial infrastructure.

Follow the money to Information & Communications

Fujikura attributed much of its performance to demand associated with expanding data-center investment. Its Information & Communications segment posted about ¥581.5 billion in six-month sales, up 91.6%, and roughly ¥217.7 billion in segment operating profit, up 194.8%. Those segment profits are equivalent to approximately 93% of the group's reported operating profit, although segment reporting and consolidated accounting include adjustments and should not be treated as perfectly interchangeable.

The company also identified orders for optical components fulfilled on short lead times. That detail matters. When infrastructure buyers need sophisticated connections quickly, availability, manufacturing yield and delivery capability may be valuable alongside the underlying component. The release does not disclose customer-by-customer sales, pricing terms or product-specific margins, so it would be a mistake to assign the entire increase to any one global technology company.

What the AI data center actually needs

A cluster of AI accelerators is not an isolated collection of chips. Information must move among machines, racks and switching systems. As facilities scale, links and their installation become complex constraints: they need to move large amounts of data, fit into limited spaces and remain serviceable when something breaks.

Optical fiber is part of that system, but the complete commercial opportunity extends beyond raw fiber length. Cable architecture, connectors, high-density assemblies, component tolerances and reliable deployment all matter. Fujikura's earnings suggest that customers are purchasing connection capability under demanding operating schedules, not merely buying more meters of glass. This is an interpretation of the business mix, not a disclosed breakdown of individual customer orders.

A founder who saw electricity coming

Fujikura traces its founding to February 1885, when Zenpachi Fujikura began manufacturing copper and silk-wrapped wire. According to the company's Japanese history, he saw an early public demonstration of electric lighting and recognized an opportunity to redirect experience in braided materials toward electrical insulation and coverings. Fujikura Electric Wire Co. was established in March 1910.

This origin is more revealing than a simple corporate birthday. The company did not begin by predicting the internet. It began by applying a practical manufacturing skill to an emerging infrastructure need. Repeating that type of technological adaptation—changing the materials and techniques while solving the problem of connection—became an enduring theme in its history.

From copper to glass: the optical-fiber transition

Fujikura's historical account describes optical-fiber development and field trials in Japan through the 1970s. Tests associated with the former Nippon Telegraph and Telephone Public Corporation progressed from laboratory integration in 1976 to field evaluation in 1978. The firm's cable approach was standardized in the program in 1979, and a graded-index fiber cable entered full-scale adoption in 1982.

Joining the individual fibers was another difficult engineering problem. Alignment and fusion splicing must minimize loss at each connection. Fujikura's records date a core-alignment fusion splicer advance to 1985. The history matters to modern AI infrastructure because connecting a high-density optical network is both a materials challenge and a manufacturing-and-installation challenge. Neither disappears when a facility buys faster processors.

A remarkable margin—and the danger of concentration

Based on the disclosed figures, Fujikura's consolidated operating margin for the six months reached approximately 27.1%, versus roughly 16.1% a year earlier. Information & Communications generated around two-thirds of reported group sales, while contributing a much larger share of operating earnings before reconciliation adjustments.

That concentration is not automatically a weakness. Scale, specialized capabilities and product mix can create substantial advantages. Yet the same concentration makes future results sensitive to purchasing cycles, capacity additions, price competition and changes in network architecture. A strong six-month margin demonstrates realized profitability; it does not establish a permanent margin for every future AI investment cycle.

Not every Fujikura business is booming

Fujikura is more than an optical-connectivity vendor. Its portfolio also includes electronics, automotive, energy and property businesses. The electronics segment, for example, recorded about ¥80.9 billion in sales, down 6.2%, and approximately ¥4.6 billion in segment operating profit, down 7.1%, as the mix of smartphone-related products changed.

The contrast matters. Calling the entire group an AI pure play obscures what its other units manufacture and what drives their results. Global manufacturers often contain several different business cycles at once. Understanding Fujikura requires recognizing both the exceptional momentum in communications and the areas that do not share it.

The forecast paradox: higher sales, lower expected profit

A headline about 160.6% operating-profit growth misses an important twist in the same announcement. Fujikura lifted its full-year revenue projection from ¥1.755 trillion to ¥1.768 trillion, yet trimmed its operating-profit forecast from ¥432 billion to ¥429 billion and its ordinary-profit outlook from ¥453 billion to ¥449 billion. Its projection for profit attributable to owners of the parent remained ¥326 billion.

Fujikura explained that its stronger communications outlook supported higher projected revenue, while the revision of its second-half assumed exchange rate from ¥160 to ¥156 per US dollar, among other effects, lowered expected profit. Actual foreign-exchange exposure is more complicated than converting one sales figure at a spot rate. The lesson is that demand strength, currency assumptions and profit forecasts can move in different directions at the same time.

The engineering and commercial questions ahead

Four tests will help determine whether this performance represents a durable shift. Can Fujikura secure orders through successive data-center construction phases? Can its optical components and delivery capabilities command attractive margins? Can manufacturing and quality systems expand without compromising reliability? And how vulnerable are results to customer concentration, exchange rates and new competing capacity?

Public results do not disclose the full mix of products, customer contracts or repeat-order rates needed to answer every question. Similarly, the size of spending on AI processors is not a direct forecast for Fujikura's revenue. What matters is the architecture that customers choose and the business Fujikura actually wins within it.

A century-old industrial answer to a new question

The scale of the AI buildout can distract from the industrial history beneath it. The companies supplying the enabling infrastructure may have spent decades mastering technologies that became essential only when a new use case arrived. Fujikura's move from electrical wire to optical connectivity illustrates that long preparation.

Its forecast of ¥429 billion in full-year operating profit is still a forecast, not a completed outcome. Future reports will reveal whether the company can turn a remarkable surge in demand into sustained competitive advantage. For now, its results provide a useful corrective to the familiar story that the AI economy is built exclusively by chip designers and software developers. Much of it is built by the people who know how to connect things.