CompanyGraph
IQE plc
IQE · United Kingdom
Grows the ultra-thin semiconductor layers inside 5G chips, lasers, and power electronics at factories in Cardiff, Pennsylvania, Taiwan, and Singapore.
IQE grows the ultra-thin compound semiconductor layers — gallium arsenide, gallium nitride, indium phosphide — that sit inside the RF chips and photonic devices used in phones, base stations, and power electronics, running reactors across Cardiff, Pennsylvania, Taiwan, and Singapore. Each growth recipe is built atom by atom inside a vacuum chamber to a customer's exact device specification, so if a Skyworks or Qorvo wanted to move that recipe to a different foundry, it would have to restart 12 to 18 months of wafer testing before its engineers could trust the output again. Because IQE's sites are each qualified to grow multiple material systems rather than just one, a customer sourcing both GaAs RF wafers and GaN power wafers can qualify a single supplier rather than two, which is something a single-material specialist cannot offer no matter how much capital it raises. That same multi-material configuration is also the point of fragility: if US export controls sever the ability to ship jointly developed wafers to customers served through the Taiwan and Singapore facilities, those reactors shift from being the network's reach into Asia to being stranded capacity that nobody else in the qualified chain can easily absorb.
How does this company make money?
The company charges per wafer, with the price varying based on how many layers the wafer contains, how large the substrate is, and how many wafers the customer has committed to buying. Customers do not only pay for finished wafers that pass quality checks — they also pay a portion of the cost for development runs made while a new product recipe is being proven out. That means revenue begins before a customer's product is even ready to manufacture.
What makes this company hard to replace?
Qualifying a new epitaxial supplier requires 12 to 18 months of wafer testing before a customer's chip design team can trust the output. The specific layer structures built jointly with this company are tied to its approved recipes and cannot be handed to a competing foundry and re-run without starting that entire clock over. On top of that, customers sign multi-year supply agreements with minimum volume commitments, which lock them into reserved reactor capacity — walking away means either breaking a contract or paying for wafers they are not taking.
What limits this company?
Each reactor can run only one growth cycle at a time, and each cycle takes between 4 and 12 hours. When the company switches a reactor between different materials — say, from gallium arsenide to gallium nitride — it must run cleaning cycles to prevent contamination before the next material can go in. Those cleaning cycles eat up reactor hours that a competitor focused on only one material never loses, which caps how much output can be squeezed from each machine.
What does this company depend on?
The company cannot operate without trimethylgallium, trimethylindium, and other metalorganic precursor gases from specialty chemical suppliers. The MOCVD reactors themselves come from only two equipment makers, Aixtron and Veeco. Ultra-pure silicon, sapphire, and other substrate wafers must be sourced externally. Export licenses from US authorities are required to ship wafers to Asian customers. And the cleanroom operators who actually run the growth processes must be trained specifically in epitaxial techniques — that knowledge cannot be hired off the street.
Who depends on this company?
RF power amplifier makers like Skyworks and Qorvo rely on this company's gallium arsenide wafers to build 5G base station components; losing that supply would stall their production. Infrared sensor companies depend on indium phosphide wafers for thermal imaging devices. VCSEL producers — the companies making the tiny lasers inside data center fiber-optic systems — need gallium arsenide epitaxial wafers, and a supply cut would disrupt optical interconnects across data centers.
How does this company scale?
Once a growth recipe has been developed and proven in one reactor, it can be copied to additional reactors, which means production volume can be expanded by adding more machines. What does not scale easily is the human expertise behind those recipes: when the company opens a new facility or adds a new material system, it needs operators who already understand epitaxial growth, and that knowledge builds slowly over time. New chambers add capacity; building the team knowledge to run a new site at full quality is the persistent bottleneck.
What external forces can significantly affect this company?
US export control rules on compound semiconductor technology are the most direct outside pressure — restrictions aimed at limiting transfers to China can cut off customers the Asian facilities were built to serve. At the same time, the global push toward electric vehicles is driving demand for gallium nitride power electronics, which strains the supply of metalorganic precursor gases those devices require. And 5G network rollouts in emerging markets create sudden surges in demand for wafers that the company's fixed reactor capacity cannot always absorb quickly.
Where is this company structurally vulnerable?
The company's Asian facilities in Taiwan and Singapore serve customers who rely on jointly developed wafer recipes shipped from those sites. If the US government revoked or further tightened export licenses for shipping compound semiconductor wafers to those customers, the Taiwan and Singapore legs of the network would be cut off. The reactors there would still exist, but they could not legally serve the customers they were qualified to supply — turning a global advantage into expensive idle capacity.