The Next Wave of AI: Building for the Edge in a Changing Semiconductor Market
Artificial intelligence is increasingly being incorporated into products that are being designed, developed, and shipped today. Avnet's latest Insights Survey found that more than half of engineers surveyed are already shipping products with AI-enabled functionality, a significant increase from the prior year. As AI moves into embedded and industrial applications, success depends on more than processor selection, model performance, or software frameworks. Engineers must also consider component availability, manufacturability, product lifecycle requirements, and long-term support to ensure AI-enabled products can be built, deployed, and sustained at scale.
The conversation is no longer about whether AI belongs in the product. Instead, engineering and procurement teams must determine how to design Edge AI systems that can be successfully built, scaled, and sustained in production.
The next wave of market growth
The market signals surrounding Edge AI are becoming increasingly difficult to ignore. Semiconductor revenue grew by roughly 20% in 2024, but many engineers and sourcing teams did not experience anything close to 20% growth in their day-to-day business. The reason is that a significant portion of that growth was concentrated in devices supporting hyperscaler AI infrastructure, which grew approximately 66% year-over-year, while many traditional high-volume market segments remained flat or declined.
Why does this matter to engineering teams? Because the semiconductor market is now operating as two markets simultaneously. One is being driven by AI training and inference infrastructure in the world's largest data centers. The other is the broader market where most embedded and industrial designs compete for components, manufacturing capacity and engineering resources.
DRAM + Flash + MPU Compute + GPU + AI accelerators vs Served Semiconductors;
Data source: Avnet estimate based on industry data – June 26
The encouraging news is that the mass market is recovering. Avnet's market intelligence—derived from both industry analyst sources and transactional data across approximately 13 million SKUs—indicates that served markets have entered a rebound phase in 2025, with sustained expansion forecast through 2028. The three-year CAGR is continuing to expand well over 8% for the overall component category, driven by recovery across industrial and consumer applications.
For engineers, this recovery creates both opportunity and competition. Demand is returning, AI workloads are expanding, and capacity constraints can emerge quickly in adjacent categories as investment continues flowing toward AI infrastructure.
Understanding where Edge AI fits into the broader AI adoption curve is equally important. The industry is progressing in waves:
- Hyperscale AI: The first wave was dominated by the Magnificent Seven and other hyperscalers building massive AI training and inference infrastructure.
- AI-Capable Devices: The second wave is driving adoption of AI PCs, smartphones and intelligent endpoint devices, helping accelerate hardware refresh cycles.
- Edge AI: The third wave is now emerging across industrial, medical, transportation, automation, security and IoT applications, where inference workloads are deployed directly onto embedded systems in the field.
This third wave is where most embedded engineers will create value. Edge AI brings intelligence closer to the source of data, reducing latency, improving privacy and enabling real-time decision-making. But it also introduces new design and supply-chain considerations that must be addressed early.
Designing for the realities of Edge AI
As AI moves from experimentation to production, the engineering organizations that succeed will be those that treat supply-chain planning, manufacturability and lifecycle management as core design disciplines rather than downstream activities.
Engage procurement and supply-chain stakeholders early. The greatest flexibility exists at the beginning of the design cycle. Memory, processors, connectivity modules and power-management devices may all be influenced by broader AI demand patterns. Engineers who understand those dynamics early gain more flexibility later.
Design for flexibility. Edge AI applications often depend on technologies that may experience changing availability or allocation pressures. Whenever practical, qualify alternative architectures and avoid unnecessary reliance on a single component path. Design flexibility can become supply-chain resilience.
Build forecasts that can be defended. The industry's long-standing reliance on inventory as a buffer is diminishing. Suppliers increasingly prioritize committed demand and long-term visibility over speculative forecasts. Engineering, operations and procurement teams should work together to create demand signals that are grounded in realistic production plans.
Pay attention to memory markets. DRAM demand has become one of the clearest indicators of AI-related activity across the semiconductor industry. As AI infrastructure investment continues, memory booking patterns can serve as an early warning signal for constraints that may later affect adjacent component categories. Teams that monitor these trends can often identify risks before they appear elsewhere in the supply chain.
Treat Edge AI challenges as design opportunities. Survey respondents identified data quality, integration, sustainability and ongoing maintenance as major barriers to AI adoption. The products that succeed will not necessarily be the ones with the largest models; they will be the ones that solve these challenges most effectively.
Ultimately, successful Edge AI products require collaboration between engineering, procurement and operations from the beginning. Designing for manufacturability now means designing simultaneously for performance, supply resilience, lifecycle management and future scalability. Those considerations are rapidly becoming as important as processor performance or model accuracy.
Opportunity belongs to the prepared
The next phase of AI growth will look very different from the first.
While hyperscale infrastructure investments created the initial surge in demand, the long-term opportunity lies in the systems, devices and equipment that bring AI into real-world applications. As that transition accelerates, the competitive advantage will shift from access to AI technology toward the ability to deploy it efficiently, repeatedly and at scale. Engage early, design for flexibility, build credible demand plans and account for lifecycle realities from the outset. The organizations that do so will move faster from concept to production while reducing risk along the way.
Edge AI represents one of the most significant growth opportunities facing the electronics industry. The engineers who understand this new wave- and the organizations that apply the same rigor to supply-chain strategy as they do to engineering innovation- will be the ones that shape what comes next.
