Custom AI chips now a priority for major tech firms
Major AI companies are increasingly designing their own custom chips to power advanced models like large language models. Building these specialized chips is tough—think a two-year timeline and teams of over 100 engineers. The surge in demand comes as companies chase breakthroughs in artificial intelligence and need more efficient, high-powered hardware to keep scaling up.
- Designing custom chips requires over 1,000 engineer-months
- Chip verification is now the key bottleneck
- Demand driven by pursuit of advanced AI models
- More companies than ever are building in-house silicon
Sources covering this
More in Chips
Micron warns RAM shortage to worsen through 2028
Micron says the current shortage of computer memory (RAM) will only get worse in 2027 and 2028.
Mod lets AMD GPUs run NVIDIA DLSS 4.5, still experimental
An open-source project called d4r has introduced early support for NVIDIA's latest DLSS upscaling tech on new AMD Radeon RX 9000 series…
Alleged leak details compromises in Apple's A20 chip for iPhone 18
A reported leak suggests that Apple's upcoming iPhone 18 will use a standard A20 chip lacking advanced packaging found in the A20 Pro.
AI drives memory prices higher into late 2026
Strong demand for AI servers is keeping memory chip makers focused on advanced server-grade DRAM, and that's pushing up prices for both…