The fact that Nokia is making financial headlines once more is almost poetic; it’s not because it’s releasing a durable phone or making it through another quarterly loss, but rather because it’s emerging as one of the more subtly intriguing AI infrastructure plays of the last 12 months. Most observers were taken aback when Nvidia committed $1 billion to the Finnish telecom company last October, purchasing shares at a price of $6.01. Nokia. The business your parents linked to an unbreakable phone and a snake game. Nokia, that is.
And yet, here we are. The current price of Nokia’s stock is close to $15, which is almost twice as much as Nvidia paid for its share. At the very least, such a move shouldn’t be made solely on the basis of sentiment. Whether or not the stock increased is not a worthwhile question. It’s whether something genuine lies beneath it.
Nvidia’s investment was not motivated by nostalgia or charity. AI-RAN technology, which integrates Nvidia’s GPUs and Arc-Pro computing platform directly into Nokia’s radio access network software, was the main focus of the agreement. To put it simply, instead of sending everything back through centralized data centers, the idea is to enable mobile operators to perform AI processing at the network edge, closer to where data is actually generated. This could end up being the most popular model for 6G networks. It’s also possible that the buildout is more expensive and takes longer than anyone is currently estimating.
The way Nokia is managing the shift is actually fascinating. Instead of giving up on its decades-old telecom infrastructure, which would have been costly and probably stupid, the company is adding AI capabilities to what it already has. radio access networks, IP routing infrastructure, and optical transport systems. These are not legacy liabilities that are being discreetly recorded. They are being put to new uses. According to reports, Nokia has an optical networking backlog of about EUR 1 billion, and its AI-related business is expanding by almost 50% annually. That business is not stumbling into relevance. That’s more intentional.
With the completion of the Infinera acquisition in early 2025, Nokia’s optical networking portfolio gained significant depth. AI training clusters rely on these high-bandwidth systems to transfer data between processors at scale. As this develops, it seems as though Nokia quietly built the precise infrastructure needed for the current AI moment for years. It’s really difficult to determine whether that’s due to lucky timing or strategic foresight.
The insider purchasing adds yet another level of mystery. Despite the stock’s sharp increase, Nokia’s own directors and executives have been buying shares. In contrast to insiders buying at the bottom, insiders buying after a run-up indicates that they think there is still more to come, not just that they were fortunate. Nothing is guaranteed by that. However, it’s a useful piece of information.
Data center expansions, hyperscaler capital expenditure announcements in the tens of billions, and Nvidia’s own story have all contributed to the narrative of the larger AI infrastructure race. It feels like a scene change when Nokia enters that discussion from the telecom side. It serves as a reminder that AI eventually needs to reach people via networks, spectrum, and the unglamorous plumbing of mobile infrastructure—it doesn’t just exist in warehouses full of GPU clusters.
It’s difficult to ignore the fact that the businesses that are most likely to be significant in five years aren’t always the ones that are currently generating the most attention. No noise is coming from Nokia. Insiders are merely purchasing stock. Sometimes paying attention is sufficient.⁖※
