HTEC’s latest State of AI in the Semiconductor Industry 2025-2026 report reveals a growing gap between AI ambition and execution in the semiconductor sector.
Almost all surveyed semiconductor organisations are pursuing AI, but fewer than half have embedded it across multiple functions. Most deployments remain concentrated in specific use cases or pilot programmes rather than being rolled out at enterprise scale.
Leadership misalignment at the execution level, along with integration challenges across EDA toolchains, manufacturing execution systems, skills gaps, and IP-sensitive environments, is constraining enterprise-wide AI scale. In practice, organisations are applying AI to accelerate R&D, improve yield, enable digital twins, and differentiate through software and architecture – but converting these targeted wins into sustained, enterprise-level performance improvements remains the key challenge.
The report also highlights strong interest in deploying AI at the Edge to enable precision and control across distributed manufacturing. Notably, among the six industries analysed in the report – healthcare, automotive, telecommunications, financial services and insurance, retail, and semiconductors – the semiconductor sector is the only one where building in-house ranks as the leading strategy. This reflects the sector’s emphasis on preserving architectural control and IP ownership, while platforms and partners are used to scale delivery and fill skills gaps.
However, semiconductor leaders are not overwhelmingly confident in applying AI strategically, with 68.8% reporting moderate or low AI literacy levels, indicating that execution maturity still lags ambition. But, leaders do estimate that falling behind on AI and Edge initiatives would set their organisations back by nearly 1.77 years on average, yet only 27.4% believe they can adopt and scale AI rapidly.
In light of this report, I spoke with Craig Melrose, Global Managing Partner, Mobility and Advanced Technologies at HTEC to dig a little deeper.
Why do you think so many organisations remain stuck in pilot phases despite heavy investment?
All new approaches start with pilot challenges. It is a change management and risk management issue – organisations want to test new approaches in a controlled environment, typically in low-risk or non-critical areas.
As a result, pilots often generate little visible impact, which makes it difficult to assess whether the technology or the general approach failed, or whether it was simply applied in the wrong context. In many cases, it is the latter. Over time, organisations move toward more meaningful use cases where the impact is clearly defined.
This process can be inefficient. Organisations that can accelerate this learning curve, by selecting more appropriate pilot environments from the outset and applying insights carefully, will be better positioned to scale faster and stay ahead of the competition.
What are the top one or two barriers that are consistently holding companies back from scaling AI across workflows?
AI is unique in that, for possibly the first time, the new approach and solution often need to be placed ‘in the middle’ of existing workflows. This needs organisations to rebuild or significantly adapt processes around it. Historically, new technologies could be bolted onto legacy systems and operating models without fundamentally changing how work gets done, but AI does not typically work in that way.
This creates a key barrier. Organisations must rethink and redesign their workflows, instead of just simply layering AI on top of them. Combined with the organisational effort needed to integrate AI into core processes, this shift can slow adoption and make scaling more challenging.
The report mentions fragmented deployments – what does ‘good’ look like when AI is truly embedded across an organisation?
‘Good’ looks like AI being placed at the centre of core workflows, rather than layered on the edges. In practice, this means embedding AI directly into how key processes operate and redesigning the workflows around it.
AI should be on the organisation’s ‘Top 3’ business challenges – the areas that have the greatest potential to drive meaningful impact and value. When AI is applied to lower-priority or less critical use cases, it tends not to deliver sufficient value to justify a bigger change or spend. As a result, organisations often treat these as incremental add-ons, which reinforces fragmented deployments.
This creates something of a catch-22: smaller use cases don’t push organisations to rethink their workflows, while higher-impact use cases require that deeper redesign to succeed. Organisations that align AI with their most important priorities and are willing to embed it into the core of their operations are the ones that achieve true, organisation-wide integration.
Are there any quick wins companies can implement to start connecting these siloed AI initiatives?
Companies are often at very different starting points with different needs. The key is to implement AI in high-impact, high-priority areas and build around it accordingly, rather than trying to force quick wins in isolation.
AI may not be a technology that lends itself to traditional ‘quick wins,’ which are typically initiatives that can be carved out or sandboxed within existing workflows. Those efforts don’t necessarily drive meaningful integration across the organisation. While they can deliver incremental improvements, often single-digit gains, or even double-digit impact in narrow areas, they are typically limited in scale.
Instead, the focus should be on identifying and pursuing transformational opportunities, use cases that sit at the core of the business and require AI to be embedded into how processes operate. In practical terms, ‘transformational’ means initiatives that can deliver double-digit impact across a significant portion of business spend or revenue, often translating into tens to hundreds of millions in value.
It is these kinds of initiatives that help connect siloed efforts and ultimately deliver lasting, organisation-wide value.
Why is Edge AI becoming such a strategic priority, specifically for the semiconductor industry right now?
Edge AI is becoming a strategic priority because it enables AI to run at the point of use without relying on constant data transfer to the Cloud. For the semiconductor industry, this is important given its role in powering devices and systems that increasingly require real-time, on-device intelligence.
Manufacturers with large, centralised operations can deploy Edge AI to support real-time decision-making across production lines and workforces, improving efficiency, responsiveness, and automation. The rise of physical AI use cases increases the need for low-latency, on-device processing that Edge AI provides.”
How does the rise of Edge AI change the way chips need to be designed and manufactured?
The rise of Edge AI is shifting chip design and manufacturing requirements in several key ways. Chips now need to prioritise lower energy consumption, as many Edge devices are mobile or operate in environments where access to power is limited. They also need to be physically smaller and lightweight to fit within constrained form factors.
Edge AI chips must deliver extremely fast, real-time processing, since many applications depend on immediate responses rather than cloud-based latency. As a result, semiconductor design is increasingly focused on optimising for power efficiency, compactness, and high-speed Edge inference.
Over the next 12-24 months, what will separate the AI leaders from the laggards in this industry?
Willingness to reinvent the organisation will be the key differentiator. Leaders will put AI at the centre of their operating model, using it to transform critical, high-impact workflows. Think of it as a ‘blank sheet of paper’ mindset, being open to redesigning processes instead of defaulting to ‘we’ve always done it this way’.
Just as important, the winners will clearly and consistently communicate what the future looks like, why it matters, and what it means for customers and employees. Success depends on building AI as a new core, rather than simply layering it onto existing ways of working.
Are we approaching a tipping point where companies that fail to scale AI risk falling permanently behind?
Yes. This has historically been true with each major technological shift, but it is even more pronounced with AI because of the level of organisational change required to adopt it at a transformational level.
If companies are unable to adapt, others will move faster, embed AI more effectively, and build a significant lead. Over time, that gap can widen to the point where it becomes increasingly difficult to close, making early and effective scaling of AI a critical competitive factor across every sector and industry.
If you could give semiconductor leaders three immediate actions to improve their AI strategy, what would they be?
First, think about business-critical areas, workflows, or challenges, and focus on radically improving those rather than incremental use cases.
Second, plan on putting AI in the middle of those workflows, and consider what else needs to change beyond simply solving the problem, so the approach is truly built around AI at the core.
Third, over-communicate, over-involve, and over-share results across the organisation to support strong change management and ensure alignment as the transformation takes shape.
