llms.txt
# SemiAnalysis
> SemiAnalysis is an independent research and analysis firm covering the semiconductor and AI supply chain, from capital equipment and foundries to accelerators, datacenters, and AI models.
Public articles are published on the newsletter at https://newsletter.semianalysis.com/. Institutional research and the industry models are subscription products; the pages below describe them.
## Company
- [About](https://semianalysis.com/about/): SemiAnalysis is an independent research and analysis firm covering the semiconductor and AI supply chain, from capital equipment and foundries to accelerators, datacenters, and AI models.
- [Dylan Patel](https://semianalysis.com/dylan-patel/): Dylan Patel is the Founder and CEO of SemiAnalysis, covering the semiconductor and AI supply chain end to end, from fab economics to frontier models.
- [SemiAnalysis Careers](https://semianalysis.com/semianalysis-careers/): Open roles at SemiAnalysis: research analysts, engineers, consultants, and operations across the semiconductor and AI supply chain.
- [SemiAnalysis Events](https://semianalysis.com/semianalysis-events/): Where SemiAnalysis meets the industry: upcoming conferences, keynotes, webinars, hackathons, and after-hours events across the AI, semiconductor, and datacenter world, plus the full archive.
- [FAQ](https://semianalysis.com/faq/)
- [Glossary](https://semianalysis.com/glossary/)
## Research
- [Newsletter archive](https://newsletter.semianalysis.com/archive): Public articles on semiconductors and AI. New research is published here first.
- [SemiAnalysis Core Research](https://semianalysis.com/core-research/): Weekly and ad hoc institutional research on semiconductors and AI: thematic drivers, inflection callouts, and fast-turn deep dives, with the implications for companies drawn out.
- [Models & Research](https://semianalysis.com/models-research/): Fourteen institutional models covering the AI buildout end to end: accelerators, networking, cloud economics, datacenters, power, and the silicon supply chain. Excel workbooks with dashboard access.
- [SemiAnalysis Data Products](https://semianalysis.com/semianalysis-data-products/): Data and dashboards on Semiconductor-related subjects.
## Industry models
- [Accelerator & HBM Model](https://semianalysis.com/accelerator-hbm-model/): SKU-level data on AI accelerator shipments, ASPs, specifications, and HBM supply across Nvidia, AMD, Google, AWS, and 60+ buyers. Forecasts to 2030, history from 2021.
- [Datacenter Industry Model](https://semianalysis.com/datacenter-industry-model/): Site-level data on AI datacenter capacity and power: more than 5,000 facilities tracked with satellite imagery and permits, capacity and capex by hyperscaler, and power supply-demand by region, 2017 to 2032.
- [AI Cloud TCO Model](https://semianalysis.com/ai-cloud-tco-model/): GPU rental economics and total cost of ownership for AI clouds: per-accelerator cost benchmarks, the all-in dollars-per-hour build-up, rental price scenarios, and cluster financials through a three-statement model.
- [AI Networking Model](https://semianalysis.com/ai-networking-model/): Device-level data on AI cluster networking: switch, transceiver, and cable units, spend, pricing, and vendor share across every major hyperscaler, 2023 to 2026.
- [Wafer Fab Equipment Model](https://semianalysis.com/wafer-fab-model/): Bottoms-up wafer fab equipment forecasts: capacity and layer-by-layer lithography flows drive tool demand across lithography, deposition, etch, wafer track, and metrology, with a tool-level litho database and segment sales through 2028.
- [Foundry Industry Model](https://semianalysis.com/foundry-industry-model/): The whole foundry chain in one bundle: wafer shipments and ASPs by node, capacity, utilization, margins, and capex, plus customer wafer-demand models, the EDA Model, the TSMC MOPS Capital Budget Model, and live Foundry and EDA dashboards.
- [SemiAnalysis Tokenomics Model](https://semianalysis.com/tokenomics-model/): The first demand-side model of AI: bottoms-up token throughput by hardware system, model architecture, and workload, the addressable market of the token economy, ROIC by deployer, and AI revenue and hardware demand forecasts.
- [Memory Model](https://semianalysis.com/memory-model/): Bottoms-up DRAM and NAND analysis from fab floor to spot market: wafer capacity and utilization by fab, bit supply and demand, contract pricing, and HBM by generation and supplier, with quarterly data from 2022, through 2030 for DRAM and 2027 for NAND.
- [SemiAnalysis ChipBook](https://semianalysis.com/chipbook/): The monthly AI and semiconductor industry tracker: 40 to 50 pages of charts and equity references from 250+ actively tracked datasets, spanning core value chain trackers, timely picks, and distillations from the SemiAnalysis models.
## Tools and benchmarks
- [InferenceX](https://inferencex.semianalysis.com/): Open source. The open standard for inference benchmarks.
- [ClusterMAX](https://clustermax.ai/): GPU cloud ratings by SemiAnalysis.
- [SemiAnalysis Fab Map](https://semianalysis.com/semianalysis-fab-map/)
- [SemiAnalysis Die Yield Calculator](https://semianalysis.com/die-yield-calculator/)
## Subscribe
- [Subscribe to SemiAnalysis](https://semianalysis.com/subscribe-to-semianalysis/)
## Latest articles
- [ASML Is Building a Hybrid Bonding Tool!](https://newsletter.semianalysis.com/p/asml-is-building-a-hybrid-bonding): Published 2026-10-15
- [Beijing Will Not Pace the Frontier: China’s Speed-First AI Safety Regime](https://newsletter.semianalysis.com/p/beijing-will-not-pace-the-frontier): Published 2026-10-08
- [Anthropic Subscriptions Offer 5x+ More Value Than OpenAI](https://newsletter.semianalysis.com/p/anthropic-subscriptions-offer-5x): Published 2026-10-05
- [How GLM5.3 Sparse Attention Affects HBM Memory Usage](https://newsletter.semianalysis.com/p/sparse-savings-persistent-demand-inside-glm53): Published 2026-09-28
- [Intel Panther Lake Teardown](https://newsletter.semianalysis.com/p/intel-panther-lake-teardown): Published 2026-09-26
- [The Chinese AI Infrastructure Boom: Introducing the SemiAnalysis China Datacenter Model](https://newsletter.semianalysis.com/p/the-chinese-ai-infrastructure-boom): Published 2026-09-25
- [ClusterMAX 3.0: The Industry Standard GPU Cloud Rating System Returns](https://newsletter.semianalysis.com/p/clustermax-30-the-industry-standard): Published 2026-09-23
- [Computation and Data Movement for Inference](https://newsletter.semianalysis.com/p/computation-and-data-movement-for): Published 2026-09-21
- [Engrams Embedding Entendre: Codesign for Efficient DRAM/SSD Offloading](https://newsletter.semianalysis.com/p/engrams-embedding-entendre-codesign): Published 2026-09-18
- [Everyone Says Datacenter Moratoriums Are Killing the US Buildout. We disagree](https://newsletter.semianalysis.com/p/everyone-says-datacenter-moratoriums): Published 2026-09-15
