It is no secret that as AI systems continue to grow in complexity and scale, their reliance on advanced semiconductor technologies intensifies. For electrical design engineers and procurement professionals alike, the intersection of AI demands and supply chain constraints presents a critical area of focus in 2025. With hardware playing such a foundational role in enabling AI innovation, whether that be through GPU’s, TPUs, or increasingly specialised chips, the availability of and efficiency of semiconductors will directly influence how quickly and cost-effectively AI applications can be developed and deployed.
At the same time, the semiconductor supply chain remains in a delicate state of recovery following years of disruption. Lead times, obsolescence risks, and geopolitical factors all continue to impact the procurement landscape. Against this backdrop, Procurement Pro spoke with Rob Picken, SVP of Digital Transformation at Sourceability, to explore how the semiconductor industry is evolving to meet the demands of AI – and what strategies are emerging to stabilise supply chains and support innovation.
Performance, efficiency, and deployment
The need for AI models to process vast data sets quickly and efficiently is influencing chip innovation. Rob Picken, SVP of Digital Transformation at Sourceability, explained:
“These complex applications demand efficient processing of vast data sets and therefore are best served by high-performance, low-power chips. Innovations in chip design, such as more energy efficiency, faster processing speed and increased memory bandwidth, improve the performance of AI applications. These factors reduce operational costs, ultimately accelerating the deployment of AI solutions across various industries.”
In 2025, this trend is likely to intensify as AI continues to scale. Designers and developers are pushing for silicon that supports deeper learning algorithms, reduced latency, and greater parallelism, all while minimising power consumption – critical for both cost and sustainability.
Supply chain pressures and shift inventory strategies
Recent years have exposed how vulnerable global semiconductor supply chains can be. For AI applications, which often require the most cutting-edge of components, even minor delays can stall progress.
According to Picken: “Over the past year or so, semiconductor shortages have hindered the progress and scalability of AI initiatives by causing delays in hardware production. This initially resulted in slower development cycles for AI technologies and constrained the deployment of advanced applications.”
While global capacity is recovering, demand is set to outpace supply once again as AI adoption surges. One adaptation gaining traction is the shift from just-in-time to buffered inventory models.
Picken noted: “A solution that is expected to gain popularity is shifting from just-in-time inventory to buffered inventory strategies, which can mitigate supply chain vulnerabilities and ensure adequate stock levels to meet fluctuating demand and unforeseen disruptions the AI boom brings.”
Tools such as lead time analysis and availability snapshots are being used more widely to provide transparency and enhance planning, particularly for companies navigating the evolving AI component landscape.
Sourceability’s supply chain strategy
For it’s part, Sourceability, has developed strategies that support customers through periods of constrained supply and unpredictable market shifts.
Picken said: “Sourceability empowers customers with cutting-edge strategies to conquer semiconductor shortages, leveraging robust market intelligence to foresee and tackle challenges like product obsolescence. By harnessing comprehensive data, procurement professionals and designers can accurately forecast market shifts, ensuring they stay ahead of the curve.
“Key tactics such as lead time analysis and availability snapshots enable designers to synchronise product development with market demands, minimising design risks and boosting product stability. Additionally, Sourceability has emphasised the importance of shifting from just-in-time inventory to buffered inventory strategies, significantly reducing supply chain vulnerabilities,” he added.
This data-driven and proactive approach positions Sourceability and its partners to maintain continuity, even as AI’s demands reshape the ecosystem.
Policy and geopolitical implications
The global nature of the semiconductor industry makes it vulnerable to policy shifts and geopolitical tensions. Interventions at the policy level could play a stabilising role in 2025 and beyond.
As Picken outlined: “Policy interventions can play a crucial role in stabilizing the semiconductor industry by encouraging local manufacturing investments, supporting geographic supplier diversification, and addressing labour shortages through expanded visa programs. Policies promoting the maintenance of legacy equipment and driving sustainability can ensure a more resilient and environmentally friendly supply chain, mitigating risks associated with geopolitical tensions and market closures.”
Such interventions are especially relevant as countries seek to onshore chip production and reduce dependency on foreign manufacturing hubs.
Impact of recent US tariffs
In April 2025, the US administration imposed a 10% tariff on all imports, with higher rates for certain countries, citing large trade deficits as a national emergency. Notably, semiconductors were exempt from these tariffs, recognising their critical role in various industries. However, related products, such as AI accelerator boards and servers, were not exempt, leading to increased costs for AI infrastructure.
These tariffs have introduced additional complexities into the semiconductor supply chain. While the exemption for semiconductors provides some relief, the increased costs for related components could impact the overall affordability and accessibility of AI technologies. Companies may need to reassess their supply chain strategies, potentially seeking alternative suppliers or considering domestic manufacturing options to mitigate the impact of these tariffs.
Innovations on the horizon
To keep pace with Ai’s requirements, the semiconductor industry is evolving rapidly. Innovations such as high-bandwidth memory (HBM), advanced packaging techniques, and system-on-chip (SoC) architectures are expected to play a central role.
Picken explained: “Looking ahead, the semiconductor industry is poised to meet the growing demands of AI through innovations like high-bandwidth memory, advanced packaging, and system-on-chip designs. These advancements will enhance processing capabilities and efficiency, which are crucial for AI applications.”
However, this forward momentum does not address every segment. Some high-reliability sectors, including aerospace and defence, continue to rely on legacy components—many of which face abrupt discontinuation as original component manufacturers wind down ageing production lines.
Picken warned: “Many high-reliability industries like aerospace and defence (A&D) still rely on legacy chips. When original component manufacturers (OCMs) find it financially unfeasible to keep production lines operational, they often shut them down with no plans to restart. This abrupt end to production leads to instant obsolescence for many products, leaving industries scrambling for solutions.”
Conclusion
In 2025, AI’s growth is deeply intertwined with the semiconductor supply chain. As demand for high-performance chips rises, companies and governments alike will need to adapt to new inventory models, foster innovation, and implement forward-looking policies. For organisations such as Sourceability, success lies in combining data-driven foresight with flexible procurement models – ensuring the semiconductor supply chain remains resilient in the face of AI’s relentless evolution.
