38 companies · 24 sourced backers · 261 milestones on record
Silicon Brains covers chips that train and run AI models, alongside the memory, connections and manufacturing systems they need. Companies are trying to make AI computing more energy-efficient and practical in data centres and devices, while improving data movement and reliably producing complex chips.
Where the 38 companies stand today, from their profiles and 10-year milestones. Tap a phase to see who is there.
Publicly listed: 30 companies (owned by public shareholders).
ACM's aim is to be a leading global supplier of wafer-fabrication equipment across cleaning, plating and other process steps, competing with much larger incu…
Everyone · publicly listed
To make reliability testing a standard part of producing advanced chips at scale, detecting devices likely to fail before they reach customers.
Some users · publicly listed
To make precise sensing and efficient power control foundational components of electric vehicles, automated machines and industrial equipment through its sem…
Everyone · publicly listed
To bring capable visual intelligence into vehicles and other devices using specialised chips.
Everyone · publicly listed
To become a credible second major supplier of data-centre AI accelerators alongside Nvidia, giving cloud providers an alternative for training and running la…
Everyone · publicly listed
To provide the sensing, control and edge-compute layer that allows machines and robots to interpret and respond to the physical world.
Some users · publicly listed
To provide the materials-engineering systems required for each new generation of smaller, denser and more energy-efficient semiconductors.
Everyone · publicly listed
To provide the connectivity chips that let AI computing systems move data effectively between their components.
Everyone · publicly listed
To enable advanced chip production through precise ion implantation, which changes the electrical properties of semiconductor materials.
Everyone · publicly listed
To be the leading partner for hyperscalers building their own custom AI accelerators and the networking that ties large AI clusters together — an alternative…
Everyone · publicly listed
To make wafer-scale processors a practical computing platform for AI.
Everyone · publicly listed · backed by Altimeter, Benchmark
To make audio and signal processing increasingly capable within compact electronic devices.
Everyone · publicly listed
To move data efficiently between the components of AI data centres.
Everyone · publicly listed
To make running AI models more efficient through chips that perform computation close to stored data.
Some users · backed by Microsoft, Nvidia
To provide the discrete and analog semiconductor components that let electronic products control power and signals.
Some users · publicly listed
To enable reliable production at increasingly small semiconductor nodes by controlling material purity and contamination throughout the fabrication process.
Everyone · publicly listed
To make specialised chips a practical, efficient way to run transformer-based AI models.
Lab · backed by Primary Venture Partners
To make testing and measurement keep pace with advanced semiconductor production.
Everyone · publicly listed
To make AI inference fast and practical using specialised language-processing chips.
Everyone · backed by BlackRock, Cisco Investments
To combine leading-edge semiconductor manufacturing with processor and accelerator design as a large-scale alternative in advanced computing.
Everyone · publicly listed
To detect and measure the defects that limit advanced-chip production.
Everyone · publicly listed
To provide the etching and deposition processes needed to manufacture more advanced chips.
Some users · publicly listed
To make low-power programmable chips useful building blocks for intelligent devices.
Some users · publicly listed
To use light to move data within AI computing infrastructure and relieve electrical communication bottlenecks.
Trials · backed by Fidelity, Viking Global
To provide custom chips and connectivity for demanding AI data centres.
Everyone · publicly listed
To provide the high-bandwidth memory needed by advanced AI processors.
Everyone · publicly listed
To make power conversion more efficient through gallium-nitride and silicon-carbide chips.
Everyone · publicly listed
To be the core computing platform of the AI era — supplying the chips, networking and software that data centres worldwide use to build and run artificial in…
Everyone · publicly listed
To make advanced semiconductor production more predictable through inspection and measurement.
Everyone · publicly listed
To make AI inference more energy-efficient through specialised computing hardware.
Some users · backed by Gradient Ventures, Valor Equity
To bring capable AI computing into phones and other connected devices.
Everyone · publicly listed
To improve how computing systems connect to and use memory through interface chips and design technology.
Some users · publicly listed
To make RISC-V-based processors a practical platform for data-centre and AI computing.
Trials · backed by Matrix Capital, Walden Catalyst
To provide a practical AI computing platform built around reconfigurable dataflow hardware.
Some users · backed by BlackRock, GV
To provide precise, dependable timing components for advanced electronic systems.
Some users · publicly listed
To supply the radio-frequency and analog components that make wireless connections work.
Everyone · publicly listed
To make RISC-V-based AI processors a practical computing platform.
Some users · backed by Fidelity, Hyundai Motor
To provide the analog and embedded-processing chips that connect electronic systems to the physical world.
Some users · publicly listed
Keep following breakthroughs like these. Every Monday, get up to 5 major breakthroughs and launches explained in plain English.
Free. No account needed. One email a week. Unsubscribe anytime.