The global semiconductor race has escalated into a high-stakes contest, fuelled by massive public and private investment. At the heart of this strategic dilemma for every chip company is a fundamental choice: to build its own factories or to specialise in design alone, and while political rhetoric often champions domestic production as the only path to security, business reality is more nuanced.
Global demand for microchips continues to outstrip supply, underscoring their critical role as a foundational technology. This importance is reflected in government actions worldwide. In the USA, the White House has threatened
tariffs of up to 300% on semiconductor imports, while offering exemptions for foreign firms that establish domestic manufacturing. Concurrently, China
is pushing for the consolidation of its semiconductor industry into a few ‘national champions’ capable of competing with major US and European firms. These moves highlight a stark reality — semiconductors are the essential components that process data now widely considered the lifeblood of the modern economy.
A complex supply chainSemiconductors, or chips, are constructed by depositing materials onto silicon wafers. They are fundamental to a vast range of technologies, from consumer electronics like smartphones and appliances to the core systems of supercomputers, energy grids, and advanced weapons. While often invisible to consumers, chips are a primary concern for national security and defence officials. The stability and control of the semiconductor chain are now directly tied to critical issues of military readiness, economic security, and geopolitical power.
Consequently, policy efforts are often focused on one critical segment — expanding domestic chip fabrication. This motive is clear in the strategic moves by the USA and China, as well as in the
European Chips Act, which establishes criteria for certifying and subsidising advanced production facilities within the European Union (EU). However, this political emphasis on onshoring manufacturing often overlooks the intricate global web — including specialised materials, equipment, and intellectual property — required to support it.
The semiconductor supply chain is a complex ecosystem involving design, fabrication, assembly, testing, and the procurement of specialised materials and machinery. A company’s ability to produce chips depends not only on the physical location of its factories, but on its overall position within this global network, which determines both where manufacturing occurs and which business model —fabless or integrated — a firm must adopt to compete.
The strategic divide between fabless and integrated device manufacturer (IDM, or fab) models is central to this issue. A company's choice to outsource manufacturing or operate its own fabrication plants is a core decision dictated by its resources and product goals that represents a fundamental trade-off between capital expenditure and control. IDMs, such as those producing specialised analogue or power chips, require immense capital but retain full command over their proprietary processes. In contrast, fabless firms avoid the burden of constructing and maintaining billion-dollar fabs. This model prioritises agility and design innovation, allowing chip designers to leverage the advanced production capabilities of dedicated foundries instead.
The impact of this choice — whether it hinders development or enables competitive advantage — depends heavily on a company’s specific strategy, resources, and geographic position. While political rhetoric often suggests a one-size-fits-all solution, the business reality is far more nuanced. A model that offers one firm a strategic edge may impose significant constraints on another.
Heavy investment for production For certain sector leaders, in-house manufacturing is a clear competitive advantage, as demonstrated by Taiwan Semiconductor Manufacturing Company (TSMC). By serving as the primary foundry for major global clients like Apple and Nvidia, TSMC has become one of the world’s most profitable and strategically vital corporations. Its financial success attracts attention from investors and policymakers alike, a point underscored by CEO C.C. Wei’s
remark to Fortune: “If someone had told me 10 years ago that I would meet two presidents in one week to discuss hundreds of billions of dollars in investments to expand our global footprint, I would have thought they were joking.” This influence comes at a cost — the capital required to maintain technological leadership is enormous. TSMC reported operating
expenses of $53.49 billion for the 12 months ending June 30, 2025 — an increase of 18.42% year-on-year — reflecting the financial burden of scaling advanced chip manufacturing.
This financial reality underscores why many firms refuse manufacturing; the model demands extreme capital intensity. Constructing and equipping a ‘state of the art’ semiconductor fabrication plant (fab), coupled with ongoing R&D and the need to upgrade equipment every few years to remain competitive, generates immense fixed costs. Profitability depends on maintaining near-full production capacity to justify the investment. This high-risk, high-reward strategy is feasible for only a handful of global players. These include contract manufacturers like TSMC, which produces designs for others without branded chips of its own, and integrated giants like Samsung, whose diverse revenue streams and
position as both a competitor and supplier to companies like Apple help ensure consistent demand for its chips, while also granting advantages such as full production control and deeper hardware-software integration.
The integrated device manufacturer (IDM) model also appeals to firms producing specialized chips that do not require leading-edge process nodes. American semiconductor company Texas Instruments (TI) exemplifies this approach. While not manufacturing the world’s most advanced semiconductors, TI’s components are widely used in everything from consumer electronics to AI hardware. As Mohammad Yunus, TI’s senior VP of technology and manufacturing,
noted: “If you have anything that plugs into the wall, or has a battery in it, or has a cord in it, you probably carry more than one TI chip in it.”
Bolstered by US policy support such as potential tariffs, TI has
committed to a $60 billion manufacturing expansion. However, analysts view this investment as risky. The company has yet to fully recover market share lost during the 2020 chip shortage, and it continues to face stiff competition from Asian manufacturers who benefit from significantly lower production costs — a disadvantage that directly impacts TI’s pricing and competitiveness.
The risks inherent in TI's move are substantial indeed. Constructing a new fabrication plant amid heightened geopolitical tensions and economic volatility represents a significant gamble. This is underscored by the industry shift exemplified by Intel, a company that has historically been the definitive IDM. This company’s recent pivot to outsourcing its own most advanced production signals a profound change, suggesting that even established players are re-evaluating the long-term viability of the capital-intensive fab model.
Intel, long the global leader in semiconductor sales, has faced multiple consecutive quarters of declining revenue and was overtaken by Nvidia in 2023. In response, the company announced major investments to expand its manufacturing capacity in 2024. This strategy aimed to both produce its own processors and serve as a foundry for other major fabless firms — including Nvidia, Apple, and Qualcomm. As then-CEO Pat Gelsinger
stated: “Job number one was to accelerate our efforts to close the technology gap that was created by over a decade of underinvestment.”
The problem, however, was not so much in underinvestment as in long-standing strategic challenges. The company has struggled to balance capital investment in manufacturing with sufficient spending on research and development. This imbalance caused it to miss pivotal industry shifts, such as the mobile revolution following the 2007 iPhone launch, leading to the loss of major clients. More recently, Intel has been largely absent from the AI boom, as tech giants like Meta, Microsoft, and Google secured vast supplies of AI accelerators from its competitors.
These setbacks eventually forced Intel to implement severe cost-cutting measures in 2025. The development of its next-generation 14A processing node now reportedly
depends on securing firm commitments from external customers. At the same time, the company is actively evaluating a shift toward a fabless business model—a move that would further reduce expenses but mark a fundamental departure from its integrated manufacturing history.
Intel’s strategic deliberations reflect a broader industry shift, framing a critical choice for chip firms — prioritise capital-intensive production control or asset-light innovation. The ascent of Intel’s competitor, Nvidia, offers a compelling case study in the advantages of the latter. Nvidia is now so powerful that it followed the US government, which will take a 10% stake in the company, by announcing in September 2025 that it would acquire a stake in Intel for
$5 billion (just over 4% of the capital). A first step towards a future takeover?
Fabless — a more agile approach By adopting a fabless model, Nvidia concentrated its resources on design and architecture — not manufacturing — freeing
capital for aggressive R&D investment. This focus enabled the firm to develop industry-leading GPUs that capture booming demand in AI and accelerated computing, while relying on manufacturing partners like TSMC for advanced production. The approach has proven decisive: Nvidia’s market capitalisation now
exceeds Intel’s, demonstrating that strategic agility and specialised innovation can outweigh the benefits of in-house fabrication.
Apart from the market giants, the fabless model offers strategic advantages to smaller firms as well. This approach represents the inverse of the strategy employed by a company like Texas Instruments, which prioritises control over specialised manufacturing. For many smaller players, however, maintaining internal production is less critical than the ability to innovate rapidly. By outsourcing manufacturing, these companies can concentrate financial and human capital on research and design — activities that directly drive their competitive differentiation and market responsiveness. This resource allocation allows them to compete effectively despite not owning fabrication facilities.
A prime example of this dynamic is SiPearl, a French semiconductor startup focused on developing high-performance, energy-efficient semiconductors. The company recently
completed the design phase of its Rhea1 chip — a key component destined for Jupiter, Europe’s first exascale supercomputer — and has moved the product to manufacturing partner TSMC. This approach aligns with SiPearl’s strategic goal of establishing a homegrown European microprocessor industry, currently dominated by US and Asian firms.
Yet the path forward remains challenging. Europe’s semiconductor ecosystem is still in
development, requiring investment across the entire value chain — from design to fabrication. Emerging companies like SiPearl contend with structural hurdles, including a shortage of specialised engineering talent and regulatory frameworks like the European Chips Act, which critics argue primarily benefits established players with existing fabrication infrastructure rather than supporting new entrants.
From a strategic standpoint, the choice to adopt a fabless model is less a matter of management preference and more a consequence of a company’s geographic and economic circumstances,
according to SiPearl CEO Philippe Notton: “In a perfect world, TSMC would have a fab in Europe, or a 100% European fab would be our partner to produce chips. Still, this perspective today feels far away as far as I can see.”
All the factors considered, this approach remains pragmatically justified. The fabless structure enables innovation by freeing SiPearl from the financial burden of building and maintaining advanced fabrication facilities. By partnering with renowned foreign contractors while waiting for future development of domestic fabs, European firms like the French designer can mitigate production risk and focus resources on proving the value and uniqueness of their designs, even amid a constrained European semiconductor ecosystem.
Although it is too early to judge, SiPearl shows some potential to follow the trajectory of fabless industry leaders such as Nvidia, AMD, and Qualcomm. The company’s entry into production represents a major step forward for Europe’s semiconductor sector and aligns with the global shift forcing firms to choose between investing in innovation or controlling manufacturing. As Intel’s recent challenges illustrate, pursuing both goals simultaneously presents considerable risks, even for well-established players. In contrast, effective and clearly targeted collaboration within the global chip ecosystem can provide a more scalable route to market leadership and technological impact.