In the current era of AI concept stocks being popular, even the Raspberry Pi, a favorite of "makers," has jumped on the bandwagon.
Raspberry Pi, listed in London, saw its share price surge by as much as 42% earlier in the day. This rally not only ended Raspberry Pi's months-long decline in share price but also prompted the market to re-evaluate the potential value of this small single-board computer (SBC) in the AI era.
There are two main driving forces behind this: first, the company's CEO, Eben Upton, personally entered the market to "buy at the bottom," and second, rumors began circulating online that the Raspberry Pi would become the best platform for running low-cost AI agents.
The CEO made a public statement through his actions: personally buying his own company's stock.
While the 40% surge might seem like a major acquisition, London traders say there were no concrete positive company announcements. The only substantial news is that regulatory filings show Raspberry Pi CEO Eben Upton earlier purchased approximately £13224 (about NT$54) worth of his own company's shares at around 282 pence per share.
Although this amount is not huge for a listed company, it has a very strong "significant meaning", symbolizing that the CEO is optimistic about the future development of his company's stock price.
Raspberry Pi's stock price briefly fell below its initial listing price of 280 pence in early February, and was down 50% from its peak a year earlier. Eben Upton's entry at this time undoubtedly conveyed to the market management's confidence in the company's value and successfully stopped investors' panic selling.
A new trend is gaining traction on social media: running AI agent services like "OpenClaw" on a Raspberry Pi.
Recently, on social media platforms like X and Reddit, an increasing number of developers and tech enthusiasts have been discussing how to run AI agent services using a Raspberry Pi, with one frequently mentioned project being..."OpenClaw".
For example, rumors have started circulating on the X platform that many buyers are frantically hoarding Raspberry Pis because it's more cost-effective to spend tens of dollars on a Raspberry Pi than to spend over $500 to buy an Apple Mac mini as an AI agent service execution platform.
This claim quickly gained traction in the tech community. Meanwhile, as open-source models such as Llama 3 and Mistral become increasingly lightweight, coupled with the improved performance and relatively low power consumption of the Raspberry Pi 5, this credit card-sized computer is indeed capable of becoming a personal AI agent server that never shuts down.
Official Response and Potential Concerns
In response to the dramatic fluctuations in its stock price, Raspberry Pi's official response was quite low-key and formulaic: "Apart from the information that has already been made public, the company has no other news."
From a fundamental perspective, Raspberry Pi stated in January that its core earnings for 2025 would exceed expectations. However, they also issued a warning, pointing out that the outlook for 2026 remains uncertain, primarily due to fluctuations in memory supply and prices. This coincides with recent reports that AI servers are taking over global memory production capacity, leading to a shortage of memory in consumer electronics.
Analysis of viewpoints
This surge in Raspberry Pi prices reflects the trend of AI development moving from "cloud training" to "edge inference".
In the past, when discussing AI, the focus was on large-scale computing power training using tens of thousands of H100 GPUs. However, with the rise of AI agents, users now need a terminal device that can maintain a long-term connection, consumes little power, and can handle simple tasks (such as reading emails, controlling home appliances, and answering simple questions).
Although the Raspberry Pi 5 does not have a powerful NPU (Neural Processing Unit), it can indeed be transformed into an AI edge computing host at a very low cost through expansion kits such as AI HAT+ (e.g., connecting to a Hailo-8L accelerator).
