IoT products continue to grow smarter, with more robust features and increasing security requirements. In the past, hardware defined and differentiated products. But as products became more complex, so too did the market’s demands. Customers are no longer satisfied with static hardware alone. They expect ongoing value, delivered through software, from the moment the customer activates a device and throughout its lifetime. The software running within defines today’s products — a shift that impacts how OEMs approach development, release cycles, and long-term support across the entire product lifecycle.
In a recent episode of the IoT For All podcast, Northern.tech CTO, Eystein Stenberg, joined host Ryan Chacon to unpack what the IoT shift means in practice for manufacturers. Drawing on experience supporting connected product OEMs across more than 60 countries, alongside insights from Northern.tech’s State of Industrial IoT report, the conversation explores the growing gap between ambition and execution, and what it takes to close it.
Throughout the discussion, Stenberg returns to a consistent theme: while most organizations understand the importance of software, many are still working through how to operationalize it effectively at scale.
AI amplifies existing IoT trends
Rather than being a novel use case, artificial intelligence (AI) features in connected products are expanding across the IoT space. From something as simple as a smart speaker or handheld home device to autonomous vehicles and heavy industrial machinery, AI capabilities are here to stay. As architectures, software, and hardware become more advanced, the proliferation of AI at the edge will grow, further increasing the prominence of AI capabilities in connected products. The key driver is reducing inefficiencies and the need for human operators while opening possibilities.
For OEMs, AI is an accelerator. AI demands faster improvements, updates, and releases. But it's also an extension of what's already been happening in IoT for years; software increasingly defines and differentiates connected products. AI only amplifies that trend, and by extension, the complexity that comes with it.
Time-to-market remains the top challenge
Time-to-market consistently ranks as a top priority for OEMs building IoT products, Stenberg notes. Additionally, product launch delays remain one of the most persistent challenges in the industry. Only 18% of respondents in the 2025 report said they always hit product release milestones. The remaining respondents report hitting deadlines “usually” or “sometimes,” a gap that compounds over time as delayed releases lead to missed market windows, delayed revenue, and engineering teams operating in a reactive mode. According to Northern.tech's 2025 State of Industrial IoT report, the average time from planning to field deployment is 1.9 years, and early data from the 2026 report indicates this is not shortening.
Interestingly, Stenberg shared that early findings also suggest a more nuanced shift. While delays persist and, in some cases, are becoming more pronounced, OEMs are also becoming more realistic about what it takes to launch a connected IoT product successfully. As the industry matures, organizations are factoring in the full scope of requirements, including software complexity, security, infrastructure, and long-term maintainability.
In that sense, longer timelines are not purely a regression. They reflect a more complete understanding of what successful product delivery actually entails, moving beyond assembling components to building robust, scalable systems designed to operate and evolve in the field.
Software cycles cause delays
Nearly half of all product launch delays are software-related, according to the 2025 report. Early data from the 2026 report puts that figure at approximately 55%, reinforcing that software remains the primary source of friction in product delivery. These delays often stem from bugs, QA bottlenecks, deployment issues, and security patching.
As the podcast discusses, this is not a matter of bugs or QA bottlenecks — the challenge is structural. Many OEMs are historically hardware companies, which became software companies by necessity. Internal processes weren't built for continuous software delivery, and adaptation takes time. Understanding the shift is one thing, while executing on it is another. When host Ryan Chacon asked about the role of software in time-to-market, Stenberg pointed out that software now defines the product itself, not just supports it. That shift introduces new layers of complexity across development, deployment, and operations.
Beyond development challenges, OEMs also face infrastructure and organizational constraints. Legacy systems extended beyond their original intent, teams coordinated across multiple layers of the technology stack, and newer components, such as AI models, introduced additional operational overhead.
More software introduces more opportunity, but without the right foundations in place, it also introduces more friction. Understanding the shift is one thing. Executing on it consistently is where most organizations still struggle.
Build versus buy: An uncalculated decision
One of the more revealing disconnects is that faster time-to-market is the top reason OEMs choose third-party software, yet they also still build significant infrastructure in-house. As Stenberg explains, it's more about what a team can build than whether it's a good idea to build it. If a team has the capability to build something, they tend to do so, regardless of whether it is the right long-term choice.
When Chacon pressed on this, Stenberg noted that OEMs do not deliberately make many infrastructure decisions; they evolve over time, extending existing systems, adapting to new products, and gradually accumulating complexity. The result is infrastructure that grows organically, but without a clear architectural or strategic foundation. Over time, this leads to technical debt that may not become visible until it surfaces as a security issue, compliance gap, or scaling limitation.
There is also a hidden cost. About 20% of the effort goes into building software, while 80% goes into maintaining it. When the maintenance burden shifts to infrastructure, like OTA updates or security, it pulls attention and resources away from the work that actually differentiates the product.
Think through the device lifecycle upfront
The challenges discussed in this conversation reflect what many OEMs are already experiencing: Getting a product to market is only part of the problem.
As products become more software-defined, the challenge is how to support them over time — through OTA updates, security, and the right infrastructure required to manage them in the field. OEMs that make deliberate decisions early about what to build, what to rely on, and how to support products in the field are better positioned to avoid unnecessary complexity and focus on what actually differentiates their products.
For a deeper look at these challenges and how leading organizations are approaching them, listen to the full conversation with Eystein Stenberg on the IoT For All podcast.
