No Single Successor to Silicon: The Heterogeneous Future of Chip Materials
There is no anointed replacement for {{silicon}} as a chip material. Instead, several materials are advancing for different niches: {{silicon carbide}} and {{gallium nitride}} are already winning in power electronics, while {{graphene}} and {{transition metal dichalcogenides}} are research bets for logic and continued miniaturization. The likely outcome is heterogeneous integration rather than wholesale replacement.
Discussions of "replacing silicon" usually assume a single successor material will displace it across all chips. In reality, no anointed successor exists, and the research landscape is split across several materials, each targeting a different limitation of silicon. First, a terminology distinction worth keeping straight: chips are made from silicon (the element, Si), not *silicone*, the rubbery polymer used in caulk and bakeware. **Power electronics (already commercialized):** The so-called "third-generation" wide-bandgap semiconductors are being adopted rapidly. Silicon Carbide (SiC): The Wide-Bandgap Semiconductor for Power Electronics is used in electric-vehicle inverters; the Tesla Model 3 (2018) was the first EV to use SiC MOSFETs in its main traction inverter, sourced from STMicroelectronics at 650 V per device. Gallium Nitride (GaN): The Wide-Bandgap Semiconductor Behind Blue LEDs and Fast Chargers dominates fast chargers and RF power amplifiers. Both handle higher voltages, higher temperatures, and faster switching than silicon, which improves efficiency and shrinks the supporting components. **Logic and miniaturization (still research):** Graphene: The Single-Atom-Thick Carbon Sheet With No Bandgap offers extremely high charge-carrier mobility, attractive for ultra-fast transistors and flexible electronics, but its lack of a bandgap makes it hard to switch transistors fully off. Transition Metal Dichalcogenides (TMDs): Atomically Thin 2D Semiconductors such as molybdenum disulfide (MoS2) and tungsten diselenide (WSe2) do have a usable bandgap and are candidates for scaling past silicon's size limits. However, 2026 research from TU Wien (published in *Science* by Mahdi Pourfath and Tibor Grasser) found that pairing 2D materials with the insulating oxide layers a transistor needs leaves a roughly 0.14 nm van der Waals gap that weakens gate control, a real obstacle the team proposes addressing with structurally bonded "zipper materials." Gallium arsenide (GaAs), long used in RF and optoelectronics, is also being explored for flexible electronics. The honest takeaway: GaN and SiC are winning in power, while graphene and TMDs remain bets for logic with no near-term path to replacing silicon in CPUs. The most plausible future is heterogeneous integration, where silicon does what it remains good at and specialty materials are bonded on top for the jobs it cannot do.