Every market cycle has a moment when people’s focus subtly moves from the loudest company in the room to the one quietly supplying the room. Something similar appears to be occurring at the moment with a particular class of hardware companies that, up until recently, the majority of investors couldn’t have named on a good day.
Located in the industrial park city of Yokneam, SolidRun is a small Israeli company that manufactures single-board computers and what the industry refers to as system-on-module solutions. These are dense, palm-sized engineering marvels, the kind of hardware that seldom makes headlines but finds its way into everything from border security systems to telecom infrastructure to edge nodes that send AI decisions to factories and hospitals. Customers are not the company’s target market. The majority of people won’t ever hold one of their products. However, as of early 2026, Wall Street’s more astute observers are discreetly circling precisely this type of business.
COMPANY PROFILE
| Company Name | SolidRun |
| Founded | 2010 |
| Headquarters | Yokneam, Israel |
| Sector | Embedded Computing / Edge Hardware |
| Key Products | System-on-Module (SOM), Single-Board Computers (SBC), Network Edge Platforms |
| Target Markets | AI edge inference, industrial IoT, telecom, smart energy, defense |
| Notable Partners | NXP Semiconductors, Marvell Technology, Arm |
| Funding Stage | Privately held; strategic investment rounds |
| Industry Position | Leading developer of high-performance embedded computing hardware for AI workloads at the edge |
| Further Reading | TechPowerUp embedded hardware coverage |
Even though the hardware is complex, the reason isn’t. With cloud providers expected to invest close to $600 billion in data center infrastructure in a single year due to the acceleration of AI spending, the focus has gradually shifted to where all that intelligence will ultimately end up. It is one thing to train enormous models in massive data centers. It is a completely different matter to get those models to function at the edge, in real environments, with real latency constraints. A different type of hardware is required for that second issue. Ruggedized, purpose-built, smaller, and using less power. For years, SolidRun and a few rivals have been quietly honing their craft.
In the world of tech investing, Mark Twain’s famous quote about pick-and-shovel merchants and gold rushes has practically become a cliché. However, because it remains true, it continues to be applied. Sitting quietly behind the headlines about Nvidia and TSMC, ASML, a Dutch manufacturer of lithography machines, has a near-total monopoly on the equipment required by every advanced chip manufacturer, making it perhaps the most perfect example in recent memory. Although it’s still too early to say with certainty whether the economics will be as clear, micro-hardware companies may occupy a similar structural position in the next phase of AI deployment.
Investors are now paying more attention to the physical layer as a result of the early 2026 software selloff. When software stocks plummeted due to concerns about AI disruption, with loan prices for software-company debt declining significantly but unevenly, some investors began to focus on a more straightforward question: which businesses actually produce the materials that AI uses, regardless of which software succeeds? After all, the arrival of a more intelligent chatbot does not render hardware obsolete. If anything, more intelligent AI necessitates it.
The most recent platforms from SolidRun are based on Marvell and NXP processors, which are intended for network edge and industrial compute workloads that cannot wait for a round trip to a data center. This is a practical argument that doesn’t rely on hype: energy utilities managing distributed grids, manufacturers updating assembly lines, and logistics companies tracking goods in real time all require computing that operates in environments where cloud latency is just unacceptable. It’s difficult to ignore the fact that this market was expanding steadily prior to the discussion of agentic AI, and that it is currently accelerating in ways that feel more structural than speculative.
None of this indicates that micro-hardware is suddenly a crowded market or that SolidRun is going to become a household name. The businesses in this sector are still small, frequently privately held, and operate in areas of the market that are unfamiliar to the majority of retail investors. However, there’s a genuine sense that the physical infrastructure of AI is finally receiving the scrutiny it probably always deserved, especially after widespread software volatility upends the easy narratives. The factories are operating. The hardware is on its way. As usual, the more difficult question is whether the valuations follow.
