Geopolitics in the Fab Lane

Mark Kennedy

October 10, 2025

What I Saw, Heard, and Didn’t Hear at a Semiconductor Conference

A glowing computer microchip with the letters AI in the center, surrounded by circuitry, representing artificial intelligence technology in a digital, blue-toned theme.

This essay bridges my Mind the Gap and Innovative Imperative series — exploring where policy and technology intersect. My time at SEMICON West underscored how semiconductor innovation and strategic policy must align if democracies are to lead in the next industrial era.

 

I just spent three full days at one of the world’s largest semiconductor gatherings — SEMICON West — part pilgrimage, part endurance sport, given the number of venues and steps involved in shuttling among them. The experience was equal parts exhilarating and humbling.

First impressions? I saw a lot of dense slides only an engineer could love, filled with acronyms only an EE (electrical engineer) could understand. What I did not see was hardly anyone else in a suit and tie.

As expected, I heard plenty about AI, ML, and physics — but what struck me most was what I didn’t hear, or at least what I didn’t hear enough of. These observations are not criticisms — the conference was world-class — but reflections from a policy lens on where the strategic conversation may next need to catch up with the technical one.

What I Didn’t Hear

  1. China — The Dog That Didn’t Bark

For all the talk of geopolitics shaping the semiconductor landscape, I rarely heard the word China mentioned — and saw little participation from it. Asia was there in force — Taiwan, Japan, Korea, and Malaysia — but the largest player in the global semiconductor story was less visible at this particular event — understandable, given the current policy environment and the regional balance of participants.

When geopolitics surfaced, it was typically framed as a constraint requiring supply-chain adjustments, not as a strategic driver of long-term realignment. A panel of government affairs executives noted that while export controls and investment restrictions impose real costs on Western firms, their effectiveness in achieving strategic goals remains uncertain. Their collective plea was for a more consistent and predictable policy framework that allows industry to plan around — not simply react to — geoeconomic friction.

A final-day address by Keith Krach, from Purdue’s Krach Institute for Tech Diplomacy, powerfully reminded attendees that “trust is the currency of influence.” He argued that techno-aggression has created global demand for trustworthy alternatives, that alliances like the Clean Network prove shared values can scale markets, and that exposure is weakness, but trust is power.

Strategic competition implication:
The limited discussion of China highlights how the technical and strategic conversations still run largely in parallel. Engineers understandably focus on performance and yield, while policymakers wrestle with ensuring technology empowers rather than controls citizens. Democracies must align their innovation ecosystems around shared principles, ensuring the open, rules-based order continues to shape the standards that govern the digital world.

  1. Tech Stacks — The Missing Architecture

For all the talk of AI models, packaging, and process nodes, I rarely heard mention of tech stacks — the integrated layers of hardware, software, and standards that define digital ecosystems. Yet these stacks increasingly determine who leads and who follows in the global technology race.

Strategic competition implication:
In strategic competition, dominance is shifting from individual components to control of the whole stack. Democracies must ensure their chips, cloud systems, telecommunications, and software standards work seamlessly together and remain accessible to trusted partners. The U.S. government’s recent emphasis on promoting the American AI Tech Stack underscores that leadership depends not just on fabs and funding, but on aligning the full architecture of innovation — from silicon to standards.

  1. National Resilience — The Unspoken Layer

I heard much about company-level resilience — diversifying suppliers, producing locally for local markets — but less about national resilience. Understandable, perhaps, for a global industry association — with strong chapters from Silicon Valley to Shanghai — whose members depend on selling worldwide.

A key vulnerability is that only 3 percent of chip packaging occurs in the U.S., versus 75 percent in Asia. While advanced packaging dominated discussion, legacy packaging — the quiet vulnerability — barely rated mention. A walk through the exhibit halls drove home how globally intertwined the ecosystem remains. Achieving true national resilience will require securing not just fabs, but the entire supply chain — from rare earths to specialty gases to packaging.

Strategic competition implication:
Company resilience buffers disruptions; national resilience deters them. Without secure domestic and allied capacity for critical stages like packaging, materials, and substrates, democracies remain one shock away from crisis. Strengthening resilience should mean shared redundancy across trusted partners — not isolationism, but an intentional network of reliable interdependence.

  1. Energy — The Missing Input

For all the talk of AI’s insatiable appetite, energy barely registered as a central theme. A few presenters alluded to its importance and side sessions explored efficiency and sustainability, but the sheer energy demand of “AI factories” — which literally turn energy into intelligence — seemed left for energy-focused conferences to address.

Strategic competition implication:
In the age of physical AI, energy-to-intelligence conversion is the new measure of industrial power. Nations that can reliably generate and deliver low-cost, clean energy will lead in AI compute and manufacturing. Integrating energy security into technology strategy is essential to maintaining leadership in both innovation and resilience.

Ten Things I Saw (and Heard) with Strategic Competition Implications

  1. Compute Is the New Oil

In 2006, five of the world’s ten largest companies by market capitalization were oil firms, with only Microsoft representing the digital sector. By 2025, nine of the ten are digital. The doubling of the chip industry in this decade to a $1 trillion market — 45% for AI/HPC, 25% for smartphones, 15% for autos and industrial, 10% for IoT — confirms it: the world’s capacity to turn energy into intelligence now defines strategic power.

Strategic competition implication:
The new industrial base is digital. Democracies must secure access to AI compute density and work together to shape a standards-based “tech club” — an open, trusted network built on interoperability and transparency. The aim is not exclusion, but confidence: keeping the club broad enough that global sales and scale continue to fund the R&D that keeps innovation ahead. Shared standards and open markets are what sustain both competitiveness and democratic leadership.

  1. Local for Local

“Just in time” has yielded to “just in case.” Firms are diversifying suppliers, building buffers, and developing local-for-local ecosystems.

Strategic competition implication:
Resilience has become the new efficiency. Democracies that master cost-effective redundancy will gain long-term strategic depth, while those that chase short-term efficiency risk critical exposure in the next crisis. Yet alternative hubs are more costly than Asia. Efficiency gains must be relentlessly pursued, and consistent policy is needed to sustain resilience over the long term.

  1. From Chips to Chiplets — The New Front Line

Packaging is no longer an afterthought — it’s the battlefield. More than half of advanced chip value now resides in the package, where 3D stacking and chiplets connect multiple dies into complex systems.

Strategic competition implication:
The value chain’s center of gravity has shifted from fabs to packaging. This creates both vulnerability and opportunity. For the U.S. and its allies, investing in advanced-packaging sovereignty is essential to converting design leadership into system-level advantage — the next step in technological deterrence.

  1. Physical AI: Intelligence with a Body

AI is leaving the data center and entering the physical world — powering robotic warehouses, AI-driven vehicles, and humanoid robots. The next trillion-dollar frontier will not just process data, but move matter.

Strategic competition implication:
Leadership in physical AI will shape industrial productivity, defense logistics, and societal resilience. Those who lead in AI-enabled robotics will define the next revolution in manufacturing and mobility.

  1. The Talent Paradox — and the Discovery Gap

Despite all the talk of automation, the most persistent concern was human — the shortage of technicians, engineers, and operators. Digital twins and simulators can ease pressure, but not replace people. Many speakers noted the need to strengthen education pipelines, yet I sensed concern that federal research funding may not be keeping pace with the scale of private and foreign investment.

Strategic competition implication:
Talent and discovery remain the twin engines of democratic strength. Democracies that invest in STEM education, basic research, vocational excellence, and cross-border talent flows will outcompete authoritarian systems constrained by demographic decline and limited academic freedom. Restoring balance between federal research funding and industrial policy is vital to sustaining innovation.

  1. Collaboration Across the Chain

Firms are moving from internal optimization to cross-supply-chain collaboration — sharing real-time data, collaborating through digital twins, and coordinating across borders.

Strategic competition implication:
Collaboration is the new comparative advantage — but only if it’s trusted. Democracies must lead in setting interoperable data and IP standards that balance openness with security, ensuring collaboration strengthens rather than exposes them.

  1. Agents and Multi-Agent Learning

The use of agentic AI was frequently discussed, including multi-agent reinforcement learning.

Strategic competition implication:
This mirrors the geopolitical challenge itself: distributed allies must coordinate without ceding control. The ability to manage multi-agent collaboration among nations — sharing intelligence, aligning supply chains, coordinating R&D — will define allied competitiveness.

  1. Digital Twins & Smart Fabs — The Cyber-Physical Race

Factories are becoming intelligent systems — governed by predictive AI models and connected through digital twins.

Strategic competition implication:
Whoever sets the interoperability standards for these smart fabs will govern the semiconductor ecosystem. Democracies must act now to ensure digital manufacturing norms reflect transparency, safety, and trusted data flows — not authoritarian control.

  1. Materials — The Quiet Foundation

Behind the high-tech glamour lie the humble building blocks — rare earths, gases, and substrates — often sourced from geopolitical competitors or fragile regions.

Strategic competition implication:
Resource security is strategic security. Efforts to explore alternative materials and recycling pathways should be accelerated. Democracies must link development finance, energy policy, and trade strategy to diversify material supply chains. Building trusted raw-material corridors is as critical as securing fabs.

  1. The Geopolitics of Trust

Krach’s reminder lingers: trust is the currency of influence. In an industry built on interdependence, nations that combine technological excellence with moral credibility will lead. Yet trust also operates at the micro level: vendors sharing data, coordinating through digital twins, and integrating systems for efficiency all depend on mutual confidence. The same transparency that enables collaboration also elevates geopolitical exposure — creating new points of vulnerability where trust can be exploited.

Strategic competition implication:
In the long run, trust compounds faster than subsidies — but it must be fortified. Democracies must make values an asset by embedding transparency, reciprocity, and reliability into every alliance and supply chain, while protecting data and standards from authoritarian exploitation.

Final Reflections

As a policy mind in a room full of technologists, I was reminded that semiconductors are no longer just an industry — they’re an index of power. The conversations I heard (and didn’t hear) confirmed that geopolitics has entered the fab cleanroom — not always acknowledged, but impossible to ignore.

At day’s end, while the engineers speak in acronyms and policymakers in frameworks, we’re ultimately focused on the same challenge: how free nations can keep turning energy into intelligence — and trust into influence — in the age of physical AI.


In line with my belief that responsibly embracing AI is essential to both personal and national success, this piece was developed with the support of AI tools, though all arguments and conclusions are my own.

Author

Mark Kennedy

WISC Director, DRI Senior Fellow