The Power Shift in Procurement Logic: From Buying Parts to Securing Capacity and Chip Definition Rights

Why automotive OEMs are moving beyond Tier 1 purchasing to secure foundry capacity, influence chip roadmaps, and control lifecycle risk.

Traditional Tier 1 suppliers like Bosch or Continental once served as the most reliable supply "firewalls" for automakers; however, during cycles of chip supply imbalance, this wall transformed into a barrier that isolated critical information. In the past, OEMs were accustomed to outsourcing chip shortage risks, assuming that signing a contract for a complete assembly meant the Tier 1 supplier should shoulder the burden of securing underlying components. This mindset proved incredibly naive when supplies of MCUs made on mature 40–90nm processes were cut off. The fundamental issue was that when capacity gaps emerged, Tier 1s could offer no tangible guarantees—they could only throw up their hands.

The traditional "tower-like" supply chain is being replaced by a "mesh-like" dynamic of strategic maneuvering. Previously, OEMs focused solely on component quotes without questioning the chip's origin; today, automotive procurement directors must appear in person at the offices of TSMC or SMIC. This "skip-level" engagement is not intended to replace the Tier 1s' integration work, but rather to eliminate information asymmetry. As intermediaries, Tier 1s derived their premium pricing from monopolizing underlying resources and maintaining a "black box" approach. Once an OEM gains visibility into a foundry's production schedules and yield data, the Tier 1's bargaining power regarding chip procurement instantly collapses.

Traditional Model

Risk Outsourcing

OEMs outsource supply risks to Tier 1s; information is opaque, underlying supply security is weak, and the OEM is left highly vulnerable during capacity crises.

Direct Engagement Model

Shared Risk

OEMs deal directly with foundries, signing Long-Term Agreements (LTAs) to lock in capacity; they gain insight into cost structures and control over supply, effectively relegating Tier 1s to the role of contract manufacturers.

Long-Term Agreements (LTAs) serve as the tangible mechanism driving this shift in power. An LTA is not merely a promise of delivery; in essence, it represents a "capacity ransom" paid by the OEM to the foundry. By making upfront deposits or committing to minimum purchase volumes directly with the foundry, automakers bypass the opaque procurement firewall maintained by Tier 1 suppliers. This approach fundamentally alters the nature of the transaction: the OEM ceases to be merely a customer purchasing components and instead becomes a "master lessor" to the wafer foundry, while the Tier 1 supplier is relegated to the role of a contract manufacturer responsible for soldering pre-selected chips onto PCBs. Gaining insight into the true cost structure is another key implication of "transparent procurement." While wafer manufacturing costs for mature process nodes are relatively transparent, the price of a chip within a controller assembly often skyrockets after passing through the layers of Tier 1 markups. By engaging directly with the foundry, OEMs can clearly break down wafer costs, packaging and testing fees, and Tier 1 management charges. This means that by assuming direct responsibility for capacity risks, OEMs also secure the right to audit the chip's gross margins.

The traditional "compare three quotes" approach is rapidly becoming obsolete for high-end autonomous driving chips. In the past, procurement teams simply compared similar MCUs from different brands to haggle over a 5% price difference. However, following Tesla’s move to bring its in-house FSD chips to the forefront, the logic of chip procurement has shifted from "buying off-the-shelf" to "defining specifications." If a procurement team cannot understand the distribution of NPU computing power, ISP processing capabilities, or SRAM bandwidth requirements within an SoC, they cannot even gain entry to strategic negotiation rooms with companies like NVIDIA or Qualcomm.

Procurement teams must possess the ability to influence chip roadmaps; otherwise, they risk being saddled with a supplier's obsolete production capacity. This "technical sourcing" approach requires procurement personnel to participate in specification definition 18 to 24 months before the chip enters the tape-out phase. To amortize the massive R&D costs associated with advanced process nodes, chipmakers tend to push high-premium, general-purpose chips. Procurement professionals with a semiconductor engineering background, however, can pinpoint redundant features that represent "wasteful costs"; by directly intervening in the chip's functional definition, they can eliminate unnecessary IP licensing fees at the source.

System-level substitution is bringing an end to the era of high-volume sales for low-end MCUs. As electronic-electrical architectures evolve from distributed systems to zonal controllers, the dozens of 8-bit or 16-bit MCUs previously scattered throughout the vehicle are being replaced by a single, high-performance SoC integrated within the zonal controller. This represents not merely a technological iteration, but a structural collapse of traditional procurement leverage. MCU purchasers, who once relied on economies of scale to secure bargaining power, find their order volumes plummeting due to architectural integration, while the value share of a single SoC within the Bill of Materials (BOM) skyrockets to a central position.

When chip specifications are defined "in reverse" by software algorithms, procurement teams clinging to a "component-centric mindset" become a heavy liability, hindering automakers' ability to respond to market changes.

In this context, procurement expertise is no longer demonstrated through the haggling over commercial terms, but through strategic bets on underlying technology roadmaps. Choosing between the ARM architecture and RISC-V determines the cost of software ecosystem migration for the next five years; deciding between Chiplet solutions and monolithic integration directly impacts medium-to-long-term yield risks and supply resilience. When chip specifications are defined "in reverse" by software algorithms, procurement teams clinging to a "component-centric mindset" become a heavy liability, hindering automakers' ability to respond to market changes.

A price reduction of a few cents secured at the time of contract signing can often morph into a financial black hole costing tens of millions of dollars five years later. The true cost of automotive-grade chips lies not in the unit price on the purchase order (PO), but in the "reckoning" costs associated with End-of-Life (EOL). The automotive industry operates on a lifecycle of 10 to 15 years, whereas semiconductor process cycles span only 3 to 5 years. This inherent temporal mismatch turns EOL management into a "silent killer"—the most destructive yet frequently overlooked threat in the supply chain.

10–15 years

Automotive lifecycle

3–5 years

Chip iteration cycle

>100x

EOL Change Costs

The real issue arises when procurement teams, aiming to meet annual price-reduction KPIs, select an emerging supplier that offers rock-bottom unit prices but lacks a mature End-of-Life (EOL) management system; in doing so, they are effectively underwriting the cost of future Engineering Change Notices (ECNs). Established automotive semiconductor giants like NXP and Renesas maintain strong bargaining power largely because they offer supply guarantees spanning up to 15 years. In contrast, OEMs face catastrophic consequences when "fabless" vendors—which prioritize an asset-light model and rapid product iteration—issue discontinuation notices due to a foundry shutting down a specific process node or internal product line adjustments.

The costs associated with re-tapeouts or PCB layout redesigns can easily wipe out all the procurement savings accumulated over the previous five years. If a core control chip must be replaced, the OEM not only incurs steep Non-Recurring Engineering (NRE) charges but also faces the need to restart lengthy validation processes for electromagnetic compatibility (EMC), thermal management, and functional safety. This forced redesign not only drains scarce engineering resources but can also lead to production line stoppages due to material shortages. Crucially, this risk is often obscured during the initial procurement phase by the illusion of a "low unit price."

Procurement decisions must shift from a "unit price mindset" to a "Total Cost of Ownership (TCO)" perspective. This means that during negotiations, factors such as supply longevity and "Last Time Buy" terms should carry more weight than the initial unit price. A mature procurement strategy requires suppliers to clearly define terms regarding wafer banks (die banks) within the contract. ...obligations to maintain inventory (or "bank" stock), or to provide free validation support for equivalent alternatives should an End-of-Life (EOL) event occur. The crucial point is that OEMs must realize they are not merely purchasing a piece of silicon, but entering into a decade-long supply agreement.

Relying solely on procurement contracts is no longer sufficient to ensure supply security amidst intense cyclical fluctuations. External dependence for advanced process nodes and capacity allocation for mature process nodes represent two entirely different battles. For advanced SoCs (5nm and below), automakers vie for TSMC’s extremely limited pilot-line slots and priority for initial tape-outs; conversely, in the mature process sector (40nm–90nm), the key to the game lies in who can provide long-term capital backing to offset the wafer foundries' razor-thin profit margins. This dual-dimensional pressure compels OEMs to move beyond simple buyer-seller relationships and enter the semiconductor industry's inner circle through equity ties.

The deep partnership between Volkswagen and Horizon Robotics offers a highly instructive case study. The establishment of their joint venture, CARIZON, appears on the surface to be a technical collaboration integrating hardware and software; however, the underlying logic is that Volkswagen has secured absolute priority for long-term supply by forging a community of shared interests. When a fabless chip designer and an OEM merge into a single entity via a joint venture, the previously opaque chip procurement process transforms into... Internal settlement. This implies that in extreme scenarios of tight wafer capacity, Horizon acts not merely as a supplier to Volkswagen but as an "extension" of its supply chain, capable of securing production slots from foundries with greater certainty.

The real issue lies not in the chip shortage itself, but in whether automakers are willing to share the risks associated with R&D and capacity expansion in the semiconductor industry. Foundries are extremely cautious about expanding production lines for mature process nodes, as the billions of dollars in capital expenditure (CAPEX) require years of high capacity utilization to recoup the investment. If an OEM merely places orders without providing a hedge against financial risk, foundries would prefer to allocate capacity to the consumer electronics sector, which offers higher gross margins and faster product iteration cycles. By participating in chipmakers' financing rounds—whether through joint ventures or equity investments—automakers are essentially paying an "insurance premium" for future production capacity.

This symbiotic relationship is reshaping the cost structure of chips. In traditional sourcing processes, R&D costs (NRE) and IP licensing fees are amortized into the unit price of each chip, leaving OEMs with little say in the matter. Under a joint venture model, however, R&D investment becomes a shared asset allocation for both parties. Crucially, capital involvement grants automakers a voice in determining underlying technology roadmaps. When Volkswagen decides to hard-code specific algorithms into the hardware, it is doing more than just purchasing chips; it is using the joint venture to control profit distribution across the entire chip lifecycle.