semiconductor
A semiconductor is a material—mostly silicon—whose electrical conductivity can be controlled; chips used in every computer and artificial intelligence are made from it.
In plain terms
Copper conducts electricity well, while glass practically does not at all. A semiconductor stands between the two: its conductivity depends on what materials are mixed into it and what voltage, heat, or light it receives. This allows precise adjustment of where and when it conducts.
This property forms the basis of the transistor, a tiny switch. It is like a faucet that you do not open or close by hand, but with another, weak stream of water. On a chip of just a few square centimeters, tens of billions of such switches work today, and their on-off switching is computation itself.
In the news, the word "semiconductor" now mostly refers not to the material, but to the finished chip and the entire industry. For AI, this is the foundation: training language models and the operation of every agent requires a huge quantity of specialized chips. Therefore, chip manufacturing capacity directly determines how much AI is available and at what cost.
How it works
- The starting material is very pure silicon, produced from sand (quartz). From this, a single large crystal, called an ingot, is grown, then cut into thin, round slices, called wafers.
- Foreign atoms are intentionally introduced into the silicon (this is doping), creating areas where electricity flows more easily or with more difficulty.
- Transistors are formed from these areas. A small control voltage determines whether current flows at a given point, and the switch's two states represent 0 and 1.
- Patterns are "drawn" onto the wafer with light (this is lithography), layer by layer, in hundreds of steps, in specially built dust-free rooms. Extreme ultraviolet (EUV) light is used for the finest patterns.
- The finished wafer is cut into individual chips, which are packaged and tested, then installed in phones, cars, or data center servers.
An everyday example
When you ask an AI assistant something on your phone, your question usually goes to a remote data center. There, graphics processing units (GPUs) or other AI accelerators, built from semiconductors, calculate the answer, and such data centers house thousands of these chips. Semiconductor chips also operate in your phone, your router and your car's braking system.
Who is affected
A common misconception
Limits and risks
- The most advanced manufacturing is concentrated in a few companies and regions: high-end processors are made mainly in Taiwan, and most memory chips in South Korea. The EUV machines required for the finest patterns are produced by a single Dutch company, ASML. Consequently, an earthquake or political crisis could disrupt global supply.
- Building a new factory takes years and costs tens of billions of dollars, so supply cannot quickly adapt to a sudden surge in demand.
- Transistor sizes are approaching atomic limits, making further shrinking increasingly expensive and slow.
- Manufacturing requires vast amounts of water, energy, and chemicals, and chips consume significant electricity during operation in data centers.
Similar, but not the same
- chip (integrated circuit) – The semiconductor is both the raw material and the broader industry name, while the chip is the finished circuit manufactured from it.
- grafikus processzor (GPU) – A GPU is a type of semiconductor chip. Designed for many parallel calculations, it is used for AI models.
- superconductor – A superconductor conducts electricity without resistance below a certain, typically very low, temperature. The essence of a semiconductor, however, is precisely controllable conduction.
Related concepts
hardver, computing infrastructure, elektronika
tranzisztor, chip, grafikus processzor, AI accelerator, memory chip
data center, computing capacity, Moore's Law, chip manufacturing, export restriction
A brief history
The transistor was developed in 1947 by John Bardeen, Walter Brattain, and William Shockley at the American Bell Labs. The first integrated circuit was created in 1958 by Jack Kilby at Texas Instruments, while the silicon-based version serving as the basis for modern chips was developed in 1959 by Robert Noyce. Gordon Moore described his observation in 1965 that the number of transistors that can be packed onto a chip doubles regularly. Originally, he calculated annual doubling, shifting to biennial doubling from 1975.
Where things stand
The semiconductor is the basic technology of the modern economy, and at the end of 2026 the industry's growth is driven mainly by demand for AI chips. Manufacturing has become a strategic issue. The United States, the EU, and China are all trying to secure their own supply with state support, while the United States and China also apply export restrictions.
In the news
- AI chip demand lifts third-quarter growth at European chemical and industrial gas firmsNov 5, 2026 · Ipar
- Semiconductor equipment makers secure supply contracts through 2028 on AI demandNov 4, 2026 · Ipar
- Nvidia orders cloud clients to sell at least 51% of their services to non-Chinese firmsNov 3, 2026 · Agent Age
- Asia's factory growth slows in major chip-exporting economies, cracks show in AI boomNov 2, 2026 · Ipar
- Korean PM: Korean plants in Hungary and Poland account for much of Europe's battery cell outputNov 1, 2026 · Ipar
- Korea and Japan become AI investment hubs: Altman in Seoul, 10 trillion won fund for JapanOct 30, 2026 · Agent Age
- Munich event on 10–13 November could shape direction of Europe's semiconductor industryOct 30, 2026 · Ipar
- AI data centers' chip and memory demand squeezes carmakers and the PC marketOct 29, 2026 · Ipar
Also known as
semiconductor · semiconductors · semiconductor industry · chip · csip · mikrochip · semiconductor · semiconductors
Sources
This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.