A useful ground-level view of what embedded hardware teams are actually dealing with in 2026 — less hype, more design constraint reality.
Chiplets: modularity hits embedded
Chiplet architectures, previously the domain of hyperscale silicon, are becoming relevant for mid-range embedded designs. The key driver: supply chain resilience. A chiplet-based design can swap compute dies between fabs without a full re-spin. For products with 5-7 year production lifecycles, this is a meaningful risk mitigation.
The practical barrier remains packaging cost and complexity. Chiplet designs require advanced packaging (2.5D, 3D stacking) that adds NRE and per-unit cost. The crossover point — where chiplet modularity is worth the packaging premium — is coming down, but it's not yet at mainstream MCU volumes.
AIoT SoMs: the right abstraction for most teams
System-on-Module (SoM) adoption is accelerating for AIoT products. The reasons are straightforward: RF certification complexity, DRAM layout constraints, and power management circuitry are all problems that a SoM solves once and amortizes across multiple customer designs.
For hardware startups especially, the build-vs-buy calculus on a carrier board + SoM versus a fully custom design has shifted decisively toward SoM for anything that isn't a massive volume play.
Localization pressure on component selection
The 2024-2025 tariff environment has created lasting pressure on component sourcing strategies. Teams that previously optimized purely for cost are now building geographic diversity into their approved vendor lists (AVLs) as a standard practice.
Practical implication: BOM analysis now needs to include country-of-origin data alongside availability and pricing. This is an underserved gap in current tooling.
AI tooling compressing NPI cycles
The compounding effect of AI-assisted schematic review, automated BOM analysis, and AI-generated test coverage is starting to show up in NPI timelines. Teams reporting 15-20% reductions in time-to-first-spin aren't getting there from any single tool — it's the cumulative effect of AI assistance at each step removing manual bottlenecks.
The teams seeing the biggest gains are the ones treating AI tooling as a workflow problem, not a point-tool problem.