U.S. Regulated & Certified Silicon

A large and strategically significant portion of the global semiconductor industry serves markets where qualification regimes, not process nodes, create the real competitive moat. Medical devices, security and trust infrastructure (TPMs, HSMs, secure elements, hyperscaler custom security ASICs), automotive systems, and defense and aerospace electronics all depend on silicon that has been certified, qualified, and in some cases physically hardened to meet category-specific standards. Across these four categories, substitution cycles run 12 to 60 months per part, supplier concentration is the norm rather than the exception, and mature-node fab capacity is strategically more important than access to leading-edge nodes.

The structural thesis of this page is that qualification regimes function as the dominant moat in these markets. A TPM or secure element certified to Common Criteria EAL 5+ cannot be replaced by a functionally equivalent uncertified MCU even if the uncertified part is cheaper, faster, and smaller. An automotive MCU qualified to AEC-Q100 Grade 1 for a specific ECU cannot be substituted without an 18-24 month requalification cycle at the system integrator and OEM. A rad-hard FPGA qualified to QML-V and sourced from a Trusted Foundry cannot be replaced by a commercial-grade FPGA even at 100x the computational performance, because the qualification regime, not the performance, is what the program contracted for. These markets operate on qualification discipline, and that discipline compounds over decades into supplier concentrations that look superficially puzzling but are structurally stable.

The parallels across the four categories are deeper than they appear. All four have qualification cycles measured in years. All four concentrate on mature process nodes where reliability, yield, and process stability matter more than density. All four have duopoly or near-monopoly structure somewhere in their silicon stack. And all four are moving toward more explicit sovereignty discipline as geopolitical fragmentation accelerates. This page surfaces those parallels and routes to category deep-dives where the supplier landscape, certification details, and sub-category structure warrant depth.

Qualification Regime Comparison

The six rows below capture the qualification regimes that govern regulated and certified silicon across the four categories. Medical and defense each decompose into two rows because the qualification profiles differ substantially between subcategories. Automotive and security/trust each fit into a single row because the regime structures are more uniform within category.

Category Profiles

Each of the four categories has distinctive structure worth brief treatment here, with links through to deeper coverage. Defense has the deepest qualification ladder and the most explicit sovereignty discipline. Security and trust has the broadest chip-category span (TPM, HSM, SE, RoT IP, eSIM, automotive cyber, hyperscaler custom ASICs) and the sharpest supplier-concentration story anchored by Infineon and NXP. Medical bifurcates between implantable/wearable low-power analog and hospital industrial-grade MCU. Automotive is the most economically significant and is covered in depth elsewhere on SX.

Medical Silicon

Medical silicon bifurcates cleanly. Implantables (pacemakers, ICDs, neurostimulators, cochlear implants) and continuous-monitoring wearables run on ultra-low-power analog and mixed-signal silicon optimized for decade-plus battery life on primary cells. Hospital equipment (patient monitors, infusion pumps, ventilators, imaging systems) uses industrial-grade MCUs and analog with a medical qualification overlay (IEC 60601 for electrical safety, IEC 62304 for software lifecycle, ISO 13485 for manufacturing). The silicon profiles are fundamentally different even though the regulatory frames overlap.

Implantable silicon is often captive at the top three manufacturers (Medtronic, Boston Scientific, Abbott), with off-the-shelf ultra-low-power MCUs from Ambiq (subthreshold operation), TI, STMicro, and ADI filling secondary roles. Biocompatible packaging and primary-cell-compatible power budgets are as critical as the silicon itself. Hospital equipment silicon is mostly standard industrial-grade parts from Infineon, Renesas, NXP, Microchip, and STMicro, with the qualification discipline rather than the silicon itself creating the moat. Medical silicon lacks a dominant foundry concentration story comparable to automotive but does have meaningful qualification lock-in (IEC 60601 requalification on substitute parts runs 12-24 months). See: Medical Silicon deep-dive (forthcoming)

Security and Trust Silicon

Security and trust silicon spans a broader set of chip categories than the single "secure element" label suggests. The category includes TPMs (Trusted Platform Modules) certified to Common Criteria EAL 4+ for PC and server attestation, secure elements in payment cards and IoT devices certified to EMVCo and CC EAL 5-6+, HSMs (Hardware Security Modules) as FIPS 140-3 Level 3-4 certified appliances for banking and enterprise PKI, embedded Roots of Trust IP (Apple Secure Enclave, Arm TrustZone, Rambus CryptoManager) co-fabricated on host SoCs, eSIM and eUICC silicon for automotive telematics and connected devices, automotive cybersecurity ICs compliant with UN-R 155 and ISO/SAE 21434, and hyperscaler custom security ASICs (AWS Nitro, Google Titan, Microsoft Pluton) that enforce tenant isolation and secure boot in cloud infrastructure. Process nodes are typically mature (40-90nm for discrete security silicon) because physical security design characteristics, tamper meshes, active shielding, and side-channel countermeasures, are process-independent; security architecture matters more than density. Embedded Roots of Trust run at host-SoC nodes (Apple Secure Enclave on N3 alongside the A-series SoC it protects).

Supplier concentration is sharp at multiple tiers. Infineon dominates enterprise TPM (OPTIGA SLB9670 is near-standard in the PC and server TPM slot) and payment card secure elements (SLM76, SLE97). NXP dominates IoT secure elements (EdgeLock SE050) and holds significant payment and automotive cybersecurity IC share. STMicro, Samsung, Nuvoton, and Microchip occupy specialized positions. The HSM appliance market is dominated by Thales, Entrust, and Utimaco. Hyperscaler custom security silicon (Nitro, Titan, Pluton) is captive and not merchant — AWS, Google, and Microsoft design internally, with Microsoft Pluton licensed as IP for integration into AMD EPYC and Intel Xeon CPUs. The certification lock-in is structurally more severe than AEC-Q100: replacing a certified security IC requires re-running the full Common Criteria evaluation plus re-certifying the platform, an 18-36 month cycle at costs of $500K-$2M per device certification plus platform re-certification effort. See: Security Silicon

Automotive Silicon

Automotive silicon is the most economically significant of the four categories and is covered in depth across existing SX content. AEC-Q100 qualification (for ICs, grades 0-3 covering different temperature and reliability tiers), AEC-Q101 (discretes), AEC-Q200 (passives), and ISO 26262 functional safety (ASIL A through ASIL D) define the regulatory framework. Qualification cycles run 18-24 months for new AEC-Q100 Grade 1 parts, and OEM program qualification adds another layer. The result is supplier lock-in measured in vehicle program lifecycles (5-10 years typically), which is the structural basis for the $2 chip paradox described in the Semiconductor Bottleneck Atlas.

Dominant suppliers are Infineon (AURIX MCUs), Renesas (RH850), NXP (S32), STMicro (SPC5, Stellar), TI (TMS570, Hercules), and Bosch (captive silicon for Bosch-integrated ECUs). Power semiconductors (SiC and IGBT modules) come from Infineon, STMicro, Onsemi, Wolfspeed, and ROHM. ADAS compute SoCs at leading-edge nodes come from NVIDIA, Qualcomm, Mobileye, and increasingly Tesla captive silicon. The automotive silicon story ties directly into the electrification supply chain at EX. See: Mature Logic Fabs, Embedded MCU and MPUs, EX Semiconductor Dependencies

Defense and Aerospace Silicon

Defense and aerospace silicon has the deepest qualification ladder of any category on this page. The hierarchy runs from Mil-Temp (-55°C to +125°C operating range, otherwise similar to commercial silicon) through MIL-PRF-38535 Class Q and Class V (military and space-grade screening), through QML (Qualified Manufacturers List, where the fab itself is certified rather than just individual parts), through rad-tol (tolerance to moderate radiation environments like LEO satellites), through rad-hard (survival in the most severe environments including strategic satellites, ICBM guidance, and nuclear weapons effects). On top of the qualification ladder sits the Trusted Foundry program (DoD-administered, tracks supply chain integrity from design through packaging) and potentially ITAR and NSA Type-1 classification for the most sensitive applications.

Supplier concentration is sharp at the top of the qualification ladder. BAE Systems FAST Labs in Manassas, Virginia, is one of the few merchant rad-hard fabs in the United States, producing the RAD750 processor that flies on Mars rovers and the RAD5500 next-generation family. Microchip (through the Microsemi acquisition, Colorado Springs operation) is the other major rad-hard FPGA and microcontroller supplier. At the Trusted Foundry tier, the accredited fab list is short: Intel Ocotillo Arizona, GlobalFoundries Fab 8 Malta, SkyWater Bloomington Minnesota, TowerSemi Newport Beach, plus DMEA's own fab at McClellan California for the most sensitive DoD applications. ITAR restrictions, Berry Amendment-adjacent sovereignty rules, and DFARS trusted-foundry clauses make defense silicon the one market where U.S.-China decoupling is already complete and legislatively enforced. The Tesla AI7 radiation-tolerant chip for LEO commercial satellite applications represents a notable commercial-spillover into a historically defense-only silicon category. See: Rad-Hard and Rad-Tolerant Fabs, Defense and Aerospace Silicon deep-dive (forthcoming)

Cross-Category Structural Patterns

Looking across the four categories, several structural patterns emerge that characterize regulated and certified silicon as a coherent market segment rather than four unrelated niches.

Qualification cycles dominate substitution timelines. Every category has per-part qualification cycles measured in years, not months. AEC-Q100 Grade 1 requalification runs 18-24 months. IEC 60601 requalification on hospital equipment silicon runs 12-24 months. Common Criteria EAL certification on new secure MCU designs runs 12-24 months. MIL-PRF-38535 QML qualification runs 24-48 months. These cycles create supplier lock-in that cannot be designed around by price or performance alone. A superior uncertified part is not substitutable for an inferior certified part until the requalification completes. The 2021-2022 chip shortage exposed this dynamic for automotive specifically, but the same structural lock-in applies across all four categories.

Mature-node concentration is the norm, not the exception. None of these markets depend on leading-edge process nodes. Secure elements run at 90-180nm because physical security design characteristics (tamper detection, side-channel countermeasures) are more important than density. Automotive MCUs run at 28-180nm because reliability and process stability matter more than density for million-unit production volumes. Rad-hard silicon runs at 90-250nm because hardened transistor designs and SOI substrates, not density, determine radiation survivability. Medical implantables run at 90-180nm because power consumption and reliability matter more than density. This is a structurally important counter-narrative to the leading-edge focus of AI and consumer silicon. Mature-node fab capacity is a strategic asset for these markets specifically, and mature-node fab closures (as have occurred over the past two decades) directly erode the qualification-heavy silicon supply base.

Supplier concentration is the norm across every category. Every category has duopoly or near-monopoly structure somewhere in its silicon stack. Infineon-NXP dominate secure elements with roughly 70-80% combined global share. Medtronic-Boston Scientific-Abbott captive silicon dominates implantables. BAE-Microchip dominate merchant rad-hard. Infineon-Renesas-NXP-STMicro-TI dominate automotive-grade MCUs. The underlying reason is the same across all four: qualification infrastructure compounds over decades and creates barriers to entry that narrow the viable supplier set structurally. Once a supplier is qualified across enough customer programs, the qualification portfolio itself becomes the moat. New entrants cannot compete on price or performance until they have built an equivalent qualification portfolio, which takes multiple decades.

Sovereignty discipline is increasingly explicit across all four categories. Defense has always had explicit sovereignty through ITAR, Trusted Foundry, Berry Amendment, and DFARS clauses. Automotive is developing explicit sovereignty through CHIPS Act FEOC rules and EU Chips Act provisions. Security and trust silicon has de facto sovereignty through concentration: Infineon and NXP are European, making Western payment infrastructure and enterprise TPM supply structurally European-anchored; at the same time, hyperscaler custom security ASICs (AWS Nitro, Google Titan, Microsoft Pluton) concentrate strategic cloud trust infrastructure inside three U.S. companies' captive silicon programs. Medical is the least sovereignty-sensitive category but FDA and EU MDR certification create de facto regional concentration. The direction of travel across all four is toward more explicit sovereignty discipline, not less. This mirrors the broader geopolitical fragmentation affecting semiconductors more generally, but is more advanced in these regulated categories because the regulatory infrastructure already exists as a framework that sovereignty policy can attach to.

Commercial-to-defense spillover is becoming bidirectional. Historically, defense silicon R&D flowed into commercial applications (GPS originated as a military system, early networking and semiconductor manufacturing benefited from defense contracts). Increasingly, commercial-driven innovations are flowing back into defense. Tesla AI7's radiation-tolerant design for LEO commercial satellite applications is an early example. Automotive ADAS compute SoCs are finding applications in defense autonomous systems. Commercial secure element designs are being adapted for defense communications. The historical one-way flow from defense to commercial is becoming a two-way exchange, which has strategic implications for supplier positioning and qualification regime design in the coming decade.

Cross-Network Context

Regulated and certified silicon intersects the broader semiconductor supply chain covered in the Semiconductor Bottleneck Atlas and the U.S. sovereignty landscape covered in U.S. Semiconductor Sovereignty Constraints. Mature-node fab capacity is a binding constraint across multiple qualification-heavy categories simultaneously, a single mature-node fab closure affects automotive, medical, security and trust, and defense silicon supply simultaneously. Trusted Foundry capacity is its own sovereignty constraint with a short accredited list. Atlas anchor pages for Infineon and NXP security silicon concentration, BAE Manassas rad-hard, and Trusted Foundry fab capacity are candidates for future buildout.

The automotive silicon category ties directly into the broader electrification story covered at EX Electrification Bottleneck Atlas and EX Semiconductor Dependencies. Defense silicon intersects aerospace and autonomy edge cases relevant to next-generation commercial LEO satellite constellations and autonomous systems. Medical silicon is largely orthogonal to the Industrial Triad of gigafactory, fab, and datacenter. Security and trust silicon intersects DX directly through TPM and secure boot infrastructure in every server platform, hyperscaler custom security ASICs (AWS Nitro, Google Titan, Microsoft Pluton) enforcing tenant isolation in cloud datacenters, and HSM appliances providing PKI infrastructure for payment networks and enterprise key management. See the DX Datacenter Bottleneck Atlas (forthcoming) for datacenter-specific silicon dependencies including these security infrastructure categories.

Related Coverage

Semiconductor Bottleneck Atlas | Security Silicon | U.S. Semiconductor Sovereignty Constraints | Rad-Hard and Rad-Tolerant Fabs | Mature Logic Fabs | Embedded MCU and MPUs | Chip Types Hub | U.S. Semiconductor Reshoring | EX Semiconductor Dependencies | EX Electrification Bottleneck Atlas