July 28, 2025 The Future of AI is Forged in Silicon, Not Just Code

Three key hardware trends — manufacturing capacity, custom chips, and talent gaps — are reshaping AI’s future beyond software innovation.

By Ivan Houlihan, SVP for West Coast US, IDA Ireland

Much of the talk about the future of AI has focused on how end users will interact with generative AI systems, highlighting major implications for the workforce, education, and the environment. These are critical conversations, but some less-discussed AI hardware trends will make just as big of an impact – and it all starts with silicon.

Semiconductors – “chips,” made with silicon – are the tiny backbones of the entire AI supply chain. In fact, AI tools are effectively useless without the hardware they run on. Long known for powering our everyday devices, the industry has seen a huge spike in demand as businesses across sectors weave generative AI into their operations. Well-known chipmakers like NVIDIA, Intel and AMD have benefited from this spike, but AI chip demand is far outpacing global manufacturing capacity. At the same time, geopolitical risk and trade tensions are making the semiconductor supply chain more vulnerable to shortages.

Global governments recognize this challenge and are taking action. In recent years, both the U.S. and the EU have enacted their own CHIPS Acts designed to boost domestic semiconductor industries. They aim to fuel the development of new advanced manufacturing facilities called “fabs” (short for “fabrication”), support startups innovating new chip designs, and encourage private investment in the industry.

These market and regulatory dynamics are quickly changing the chip industry landscape, and are sure to affect our everyday devices. Here are three hidden trends shaping the future of AI hardware:

Removing manufacturing roadblocks

The race to build new semiconductor manufacturing capacity is intensifying globally. Unlike software development, chip manufacturing requires massive physical infrastructure investments.  These advanced fabrication facilities can cost upwards of $20 billion and take years to construct. Why? Complex requirements for land, power and water use, plus permitting rules and lack of local construction expertise can slow down fab builds.

To speed up the process and attract more investment, countries like Ireland, France and Germany are pooling funds, streamlining regulatory requirements and setting aside parcels of land that meet the needs of these complex projects. In Ireland in particular – which is the region I spend my time focused on – the government has focused on reducing barriers to entry both for semiconductor manufacturers based in Ireland and those headquartered elsewhere. Ireland is planning to spend more than €100M to support these goals.

The EU CHIPS Act also encourages additional investment in open EU foundries, which reserve some of their production capacity to manufacture chips designed by companies without their own dedicated fabrication facilities, spreading opportunity to newer innovators in the market.

In the U.S., lawmakers have prioritized tax credits for companies investing in these facilities in states like Arizona, setting a 2026 deadline to encourage them to break ground on these projects as soon as possible. This infrastructure-first approach recognizes that AI’s future depends not just on chip design innovation, but on having the manufacturing capacity to produce billions of chips at scale.

Custom chip design demand

We’re witnessing a fundamental shift as industries beyond traditional tech companies embrace custom semiconductor designs to unlock new opportunities enabled by AI. For example:

• Both established chip companies like Qualcomm and NVIDIA and automotive manufacturers with internal chip design teams like General Motors are building specialized processors for autonomous vehicles.
• Medical device companies like Boston Scientific are working with chipmakers to develop custom chips for smart healthcare applications. This trend is accelerating because off-the-shelf chips often can’t deliver the performance these specialized applications demand.
• To address the need for smaller, more powerful chips, companies like Synopsis, Siemens, and Cadence are developing the next generation of Electronic Design Automation (EDA) tools – the software that enables chip design. These new tools will help companies design more energy-efficient chips, directly addressing the massive energy consumption challenges facing data centers and AI systems.
• Startups like Mbryonics are developing quantum photonics chips that will enable the next generation of AI by speeding up the ability to transfer large amounts of data without excessive heat.

Semiconductor talent gaps

Perhaps the most critical bottleneck isn’t manufacturing capacity or design tools – it’s talent. The semiconductor industry requires a unique blend of skills spanning materials science, electrical engineering, software development, and manufacturing processes. Traditional university programs often can’t keep pace with the rapidly evolving requirements of AI chip development, leaving critical gaps as the technology takes off.

Research institutions like Ireland’s Tyndall National Institute are stepping up to bridge this gap, working directly with industry and academia to transform cutting-edge semiconductor research into commercial products. The focus isn’t just on training more engineers, but on creating collaborative ecosystems where academia, established companies, and startups can share knowledge and accelerate innovation.

This talent development is particularly crucial for emerging areas like photonics and edge AI, where the expertise pool is still relatively small but demand is exploding. In the EU, member states have established 27 competence centers backed by the EU Chips Act to further support technical skills development in the semiconductor industry. Ireland’s center, dubbed I-C3, is focused on providing startups and SMBs with access to design tools, expertise, and funding that would otherwise create prohibitive barriers to entry.

Silicon shapes AI’s destiny

These three trends – manufacturing capacity expansion, custom chip proliferation, and talent development – are converging to reshape not just the semiconductor industry, but the entire AI ecosystem. As companies across sectors recognize that AI’s potential is fundamentally limited by the chips that power it, we’re entering an era where hardware innovation will be just as critical as software breakthroughs. The countries and companies that successfully navigate this transition – by building the right infrastructure, fostering cross-industry collaboration, and developing specialized talent – will ultimately determine how quickly AI transforms from today’s experimental applications into tomorrow’s indispensable tools.

For business leaders, the message is clear: the future of AI isn’t just about algorithms and data – it’s about the silicon that makes it all possible. Understanding how global supply chains, talent upskilling, and industry-specific demand will shape AI’s future is critical for every business.

About the Author:
Ivan Houlihan is Senior Vice President and Head of the West Coast of the United States for IDA Ireland, the national investment development agency for Ireland. IDA Ireland partners with companies worldwide to provide financial assistance, on-the-ground support and advice to help companies establish and transform their operations in Ireland. Based in California, Houlihan works closely with existing and potential clients in technology, financial services, life sciences and engineering throughout the Western US and Mexico.

Read more from the author:
How Companies Can Create an Ethical AI Framework | TechStrong.ai, May 10, 2024

Subscribe to Industry Today