Raspberry Pi has lifted its 2026 profit outlook after a stronger first half, showing that AI demand is now reaching well beyond giant GPU clusters and into low-cost computing hardware.
Raspberry Pi is no longer just a small board for students, hobbyists and weekend builders. The Cambridge company has become a live test of how far the artificial intelligence buildout can travel through the hardware economy, and its latest trading update suggests the answer is further than many investors expected.
The company said trading in the six months ending June 30 has been strong, with unit shipments expected to exceed 4 million and adjusted underlying earnings of at least $38 million. That matters because its official analyst consensus, last updated in May, had full-year 2026 adjusted EBITDA at $42 million. In simple terms, Raspberry Pi is now expecting to make almost a full year of profit in the first half alone.
According to Sharecast, the company said first-half profitability is expected to be materially ahead of last year, helped by higher volumes, a favourable product mix and the use of lower-density DRAM inventory built during 2025. That is a very specific kind of good news. It tells investors that demand is strong, but also that some of the margin strength comes from inventory bought before memory prices became more painful.
The bigger story is not that Raspberry Pi sold more boards. It is where those boards now fit. AI spending is usually discussed through Nvidia chips, hyperscale data centers and cloud capacity. That is still the center of gravity. But more companies are also trying to run useful AI workloads closer to sensors, cameras, robots, factory equipment and local machines.
This is where Raspberry Pi becomes interesting. Its boards are cheap, flexible and familiar to engineers. They are not replacing high-end accelerators, and nobody should confuse a Raspberry Pi with a rack of GPUs. But edge computing does not always need the largest model or the most expensive processor. Sometimes it needs a reliable board that can sit inside a product, collect data, run lightweight inference and connect to a wider system.
That shift is good for companies that sit below the glamorous end of AI infrastructure. A robot that sorts packages, a smart camera in a warehouse or an industrial controller that makes local decisions may never appear in a cloud capex headline. Still, it needs compute. It needs memory. It needs supply. Multiply that across thousands of customers and Raspberry Pi starts to look less like a toy maker and more like a quiet supplier to the next layer of automation.
The company has been careful about its language. It has pointed to robust demand from original equipment manufacturers and other customers, not just enthusiasts. That distinction matters. Hobbyist sales can be cyclical and emotional. OEM demand is usually harder to win, but more valuable once a product is designed around the platform. If a manufacturer builds Raspberry Pi hardware into a device, switching away is not as simple as choosing a different board on a website.
The memory problem cuts both ways
There is a catch, and it is an important one. The same AI boom helping demand is also making memory harder and more expensive to buy. Raspberry Pi has already faced pressure from DRAM and non-volatile memory pricing, and the company said unit economics are expected to moderate in the second half as lower-cost memory inventory is used up.
That is the strange position many hardware companies now face. AI is lifting the revenue opportunity while raising the cost of the components needed to serve it. Large data center buyers have been pulling huge amounts of memory capacity into their own supply chains. Smaller hardware companies can still benefit, but they have less room for error when prices move quickly.
Raspberry Pi says it is confident it can secure the inventory needed for its 2026 production goals. It also expects to use debt facilities where appropriate to make strategic memory purchases. That is a practical move, but it changes the way investors should think about the business. Inventory management is no longer a back-office detail. It is central to whether the company can turn demand into durable earnings.
The stock market noticed. Raspberry Pi shares jumped sharply on Friday after the outlook upgrade, with Investing.com reporting a move of more than 20 percent and an intraday high above 960 pence. That reaction shows how hungry investors remain for AI infrastructure stories outside the obvious US mega-cap names.
For entrepreneurs, there is another lesson here. Markets often reward the companies that make new technology usable, not just the companies that invent the most powerful version of it. Raspberry Pi built its reputation by making computing accessible. Now that same positioning gives it a role in industrial AI, robotics and local automation.
The next question is whether this momentum can survive more expensive memory and a tougher second half. If Raspberry Pi keeps winning OEM customers while protecting margins, it could become one of the more unusual public proxies for the AI buildout. Not because it owns the cloud, but because it helps bring intelligence to the machines outside it.
Also read: AirTrunk makes India the next big test for AI infrastructure • Lightricks splits itself in two as AI costs force a reset • Comfy Desktop makes ComfyUI easier for commercial AI builders
