TL;DR
Supplier Push: LG Innotek is reportedly pursuing Tesla-linked FC-BGA substrate work, but no award has surfaced. Packaging Role: FC-BGA substrates handle AI chip power and signal routing, and one 2024 outlook valued the segment at 16.4 USD billion by 2030. Competitive Stakes: Samsung brings a production record dating to October 2022, while LG is building broader automotive and substrate capacity around the same lane.
LG Innotek appears to be considering seeking to enter Tesla’s AI semiconductor supply chain in 2026, targeting ABF-based FC-BGA substrate orders in a direct supplier challenge to Samsung Electro-Mechanics. No award has surfaced, leaving the move at the pursuit stage rather than the contract stage.
FC-BGA substrates are the high-density package substrates that sit between an advanced chip and the printed circuit board, routing power and signals from the silicon die through microscopic wiring layers to the broader system. ABF, or Ajinomoto Build-up Film, is commonly used in these substrates because it supports the fine-line routing, high I/O counts, and electrical performance needed by AI accelerators, server processors, and autonomous-driving compute chips.
A 2024 outlook put the FC-BGA segment at 16.4 USD billion by 2030. Tesla’s 2024 AI chip supply strain already showed how compute allocation can tighten around AI programs and shift pressure onto the packaging stack beneath those chips.
Tesla’s Optimus program adds robotics context, and autonomous-driving hardware keeps the same substrate family tied to vehicle compute. Both lanes sit beside the supplier contest, but neither one names a winning vendor.
Why Tesla’s packaging demand matters
FCBGA package substrates connect dense chips to a main board, and package layers route power and signals for semiconductors used in AI, server and automotive systems. Buyers that need dense AI compute often treat the part as a performance bottleneck, not a commodity board component.
FCBGA High-Density Package Substrates architecture (Source: Samsung)
Large-body, high-multilayer designs can require ~110x110mm, 26L support for high-performance computing. High yields, clean thermal behavior and stable signal integrity become separate qualification hurdles, which means supplier readiness in this lane depends on process control as much as factory scale.
How Samsung and LG approach the same lane
Samsung Electro-Mechanics enters any Tesla-linked race with a longer substrate record. Its KPCA Show 2025 and October 2022 record covers AI, server and automotive package-substrate work, while Advanced substrate competition adds broader context around materials strategy and packaging investment.
October 2022 is a practical divider between the two contenders. Samsung says Korean mass production of AI and server FCBGA substrates began then, while the 2025 showcase extended that pitch into glass-core development and newer AI packaging work. Samsung’s timeline gives it a dated production foothold instead of a fresh expansion promise.
Glass-core work also matters because it places Samsung’s public pitch on future package design as well as current manufacturing. Readers weighing the rivalry get a clearer contrast between one supplier highlighting production history and another stressing expansion capacity.
LG Innotek, by contrast, has been widening its business scope from camera modules toward AI chip substrates, ADAS and vehicle lighting. LG framed the same diversification drive as a way to reduce reliance on Apple and move deeper into higher-value components.
Vehicle lighting and ADAS also place LG’s push inside automotive electronics, which moves the substrate story closer to Tesla’s operating mix than legacy camera-module work alone. That does not confirm an order, but it does show why Tesla would fit LG’s broader direction.
LG’s Dream Factory can automate all ten FC-BGA steps. An earlier LG target for meaningful FC-BGA sales fits that expansion path.
At the plant level, LG can generate over 100GB of production data daily for defect prediction, inspection and equipment monitoring. Factory automation and data volume strengthen the case for manufacturing ambition, even though neither detail names Tesla as a customer.
The AI5-AI4 mismatch and earlier Tesla signals
In the live claim, the AI4 chip detail in LG’s supplier push keeps the reported order target on AI4 chips rather than AI5. Samsung, LG Innotek and Tesla have not publicly tied the current substrate claim to an AI5 order, which keeps the higher-numbered chip in the background lane for now.
In January 2026, Tesla’s AI5 roadmap pushed the chip line into a broader compute discussion, and a January 2026 TeslaNorth account quoted Musk saying, “Solving AI5 was existential to Tesla,” as Dojo 3 plans returned to the picture. January 2026 belongs here as background, not as proof of the live supplier pursuit.
An eventual AI4 supplier designation from Tesla, LG Innotek or Samsung would move the story from supplier pursuit to confirmed program work. Until then, the strongest supported reading is a reported challenge to Samsung in a packaging lane that could touch Tesla’s AI and robotics hardware at the same time.