Evaluating OEM In-House Chip Development: Feasibility Guide

Introduction

The automotive industry is witnessing a historic shift as leading automakers move chip development in-house. From custom SoCs for autonomous driving to proprietary power modules for electric drivetrains, this vertical integration challenges the traditional semiconductor supply chain. As a strategic investor, I see both promise and peril in this trend. The promise lies in differentiation and cost control; the peril is the enormous capital required to reach economic scale.

This guide provides a technical and financial framework to evaluate whether a given automaker can realistically justify self-developed silicon. We will examine the break-even volume for SoCs and power modules, and assess the long-term impact on Tier 1 suppliers who have historically controlled system integration.

Economic Thresholds for SoC Development

System-on-Chip (SoC) development for automotive applications demands massive upfront investment. A single leading-edge SoC for ADAS or cockpit computing can require over $500 million in design, verification, IP licensing, and mask costs. Variable costs include silicon wafers, packaging, and test, often ranging from $50 to $150 per unit depending on die size and process node.

To calculate the economic scale threshold, use the formula: Break-even volume = NRE cost / (external purchase price - internal variable cost). Suppose an external chip costs $200 per unit, and the internal variable cost is $100. The contribution margin is $100. With an NRE of $500 million, you need 5 million units over the chip's lifecycle. If the auto platform lifecycle is 5 years, that translates to 1 million vehicles per year with that specific SoC.

In practice, few automakers exceed 2 million annual sales for a single platform. Therefore, a custom SoC is only economically feasible for top-tier volume manufacturers with shared platforms across multiple brands. Even then, the chip must remain in production for years without major redesign, which creates technical lock-in risks.

Power Module Economics

Power modules for electric vehicles, such as silicon carbide (SiC) or insulated-gate bipolar transistor (IGBT) modules, have different economics. NRE costs are significantly lower than for advanced SoCs, often between $30 million and $80 million. However, the variable cost per module is critical, especially when commodity suppliers like existing Tier 1s offer aggressive pricing.

The scale threshold for in-house power modules is driven by annual vehicle production. If an automaker produces 300,000 EVs per year, each using two power modules, that is 600,000 modules. With an NRE of $50 million and a contribution margin of $20 per module, the break-even is 2.5 million modules, or about 4 years of production. This is achievable if the same module design is used across multiple models.

Yet, the true benefit comes from vertical integration of the semiconductor process. By controlling the design and potentially collaborating with foundries on process tweaks, automakers can improve efficiency by 2-5%. For a battery-electric powertrain, that translates into tangible range and cost advantages. But without sufficient volume, the NRE burden makes the in-house module more expensive than sourcing from specialty Tier 1s.

Vertical Integration Impact on Tier 1 Suppliers

The long-term impact on traditional Tier 1 suppliers is severe. Traditional Tier 1s earned profits by bundling microcontrollers, power devices, and software into black-box modules. When automakers design their own SoCs, they gain ownership of the architecture, and the Tier 1 is reduced to a contract manufacturer, losing design authority and margin.

We also see a shift in value capture. A custom SoC allows automakers to update software over the air and differentiate their user experience. This pushes Tier 1s toward lower-value roles: printed circuit board assembly, thermal management, and final packaging. Over time, the most capable Tier 1s will need to reinvent themselves as software or silicon foundry partners, or else face commoditization.

Another consequence is the fragmentation of the supply chain. In the past, Tier 1s insulated automakers from chip suppliers. Now, automakers are signing direct agreements with foundries and OSAT providers. This bypasses Tier 1s entirely for high-value components, leaving them with only bulky, low-margin mechanical assemblies. Strategic investors should watch for Tier 1s that lack a strong semiconductor design team or a proprietary power module portfolio.

Assessment Framework for Investors

To evaluate a specific automaker's vertical integration plan, I use a five-step framework. First, quantify the annual unit sales of the vehicle platforms that will use the in-house chip. If that number is below the break-even threshold calculated above, the program fails on pure economics.

Second, assess the automaker's engineering capability. Designing an SoC requires expertise in architecture, verification, safety certification (ISO 26262), and software toolchains. Many automakers will fail here and must rely on silicon partners, which dilutes the strategic benefit. Third, estimate the opportunity cost: what other projects could fund with that capital, and what is the expected return on investment?

Fourth, examine the competitive ecosystem. If multiple automakers develop similar SoCs, the differentiation advantage is temporary, and the costs remain sunk. Finally, consider the long-term supply chain security. In-house development reduces dependency on external vendors, but it also concentrates risk if the automaker's own design has flaws or production delays. A balanced approach, such as co-development with a specialized semiconductor firm, may be more effective than full in-house ownership.

Conclusion

Self-developed SoCs and power modules are not a universal strategy. They are only economically viable for automakers with very high production volumes, typically exceeding two million vehicles per year for custom SoCs and at least 500,000 electric vehicles per year for in-house power modules. Below these thresholds, the NRE burden destroys shareholder value.

For traditional Tier 1 suppliers, the trend is an existential threat. Those who fail to develop silicon-level capabilities will be reduced to low-margin assembly. However, the window for automakers to make this transition is closing, as next-generation semiconductor processes demand even higher NRE investments. My recommendation as an investor is to support select automakers that already have the scale and engineering talent to execute, while reducing exposure to Tier 1s without silicon expertise.