Inference chips

Peak · emerged April 2017 · high December 2025 · +39% on the month before · 321 mentions on Trending in 14 days

Accelerators built to serve models cheaply and fast rather than to train them: TPUs, LPUs, wafer-scale and custom silicon. Named by Jouppi et al., Google: In-Datacenter Performance Analysis of a Tensor Processing Unit (Apr 16, 2017). Also called AI accelerators, AI chips, NPU, LPU, Custom silicon.

Index 0 to 100, monthly

Signals

Signal	Value	Reads as
Hacker News stories, last 12 months279 stories, +19% on the 12 months before	279 stories, +19% on the 12 months before	Adoption
arXiv papers, last 12 months88 papers, +91% on the 12 months before	88 papers, +91% on the 12 months before	Adoption
Wikipedia views, August 202611,788, -22% on a year before	11,788, -22% on a year before	Decline
GitHub repositories1,197	1,197	Adoption
Funding rounds naming it5, $2.1B in all	5, $2.1B in all	Adoption

Trending terms

Term	Last day	Days	PeakPeak score
Google TPUHardware2 days, last Sep 7 2 weeks ago	Sep 72 weeks ago	2	26
CerebrasCompany1 day, last Sep 3 3 weeks ago	Sep 33 weeks ago	1	14
Nvidia BlackwellHardware1 day, last Aug 25 4 weeks ago	Aug 254 weeks ago	1	14

Companies

Name	Mentions
Etched	4
Eliyan	2
SambaNova	2

GitHub projects

1,197 repositories

Name	Created	Stars
xLLM-AI/xllmA high-performance inference engine for LLM, VLM, DiT and REC models, optimized for diverse AI accelerators. It is hosted in OpenAtom Founda · Aug 12, 2025 1 year ago	Aug 12, 20251 year ago	1,579
kendryte/nncaseOpen deep learning compiler stack for Kendryte AI accelerators ✨ · Apr 4, 2018 8 years ago	Apr 4, 20188 years ago	898
InfiniTensor/InfiniTensorInfiniTensor is a high-performance inference engine tailored for GPUs and AI accelerators. Its design focuses on effective deployment and sw · Jul 27, 2022 4 years ago	Jul 27, 20224 years ago	473
facebookresearch/deepfloatAn exploration of log domain "alternative floating point" for hardware ML/AI accelerators. · Sep 27, 2018 8 years ago	Sep 27, 20188 years ago	401
bytedance/xpu-perf[HPCA 2026] AI Accelerator Benchmark focuses on evaluating AI Accelerators from a practical production perspective, including the ease of us · Apr 12, 2023 3 years ago	Apr 12, 20233 years ago	387
trevorpogue/algebraic-nnhwAlgebraic enhancements for GEMM & AI accelerators · Apr 25, 2023 3 years ago	Apr 25, 20233 years ago	296

Funding

5 rounds

Name	Date	Amount
SambaNovaSeries F · Jul 8, 2026 2 months ago	Jul 8, 20262 months ago	$1,000M
EtchedGrowth · Aug 18, 2026 5 weeks ago	Aug 18, 20265 weeks ago	$700M
EtchedSeries C · Jul 23, 2026 2 months ago	Jul 23, 20262 months ago	$300M
EliyanSeries C · Jul 29, 2026 8 weeks ago	Jul 29, 20268 weeks ago	$145M
nanoSkunkWorkXSeed · Sep 2, 2026 3 weeks ago	Sep 2, 20263 weeks ago	$2M

Movement	Stage
Local AIRunning language and image models on your own laptop, phone or server instead of a cloud API.	Peak
Mixture of expertsModels split into many expert blocks with only a few active per token, so they grow in size without growing in cost.	Peak

History

Date	What changed
Sep 24, 2026	Tracking started: emerged April 2017, peak, 5 other names recorded

Sources: Curve: Hacker News story titles, Wikipedia pageviews and arXiv papers by month, refreshed Sep 24, 2026. Related entities from the fru.dev sites' public APIs, matched by name. How stages work.