Every claim in this report traces back to one of 50 evidence records below. Each was captured passively during recon, hashed at capture for chain-of-custody, and graded per the Admiralty Scale (NATO STANAG 2511). Click any ev_xxx chip elsewhere in the report to jump straight to its source record.
ev_001 B-2 A neural processing unit (NPU), also known as an AI accelerator or deep learning processor, is a class of specialized hardware accelerator or computer system designed to accelerate artificial intelligence and machine learning applications, including artificial neural networks and computer vision.
ev_002 B-2 GPUs are also increasingly being used for artificial intelligence (AI) processing due to linear algebra acceleration, which is also used extensively in graphics processing.
ev_003 B-2 Tensor Processing Unit (TPU) is a neural processing unit (NPU) application-specific integrated circuit (ASIC) developed by Google for neural network machine learning... Google began using TPUs internally in 2015, and in 2018 made them available for third-party use.
ev_004 B-2 Nvidia Corporation is an American multinational technology company headquartered in Santa Clara, California. The company develops graphics processing units (GPUs)... Founded in 1993 by Jensen Huang, Chris Malachowsky, and Curtis Priem.
ev_005 B-2 Advanced Micro Devices, Inc. (AMD) is an American multinational semiconductor company headquartered in Santa Clara, California. It develops central processing units (CPUs), graphics processing units (GPUs), field-programmable gate arrays (FPGAs)...
ev_006 B-2 Intel Corporation is an American multinational technology company headquartered in Santa Clara, California... Intel was the world's third-largest semiconductor chip manufacturer by revenue in 2024.
ev_007 B-2 Taiwan Semiconductor Manufacturing Company Limited... holds a commanding majority in the semiconductor foundry market, accounting for approximately 70% of the global market share. As the leading dedicated contract chipmaker, it serves as the main supplier for Nvidia, Apple, Broadcom, and Qualcomm.
ev_008 B-2 Broadcom Inc. is an American multinational designer, developer, manufacturer, and global supplier of a wide range of semiconductor and infrastructure software products... As of 2026—amid the AI boom—Broadcom is one of the largest companies globally.
ev_009 B-2 Cerebras Systems Inc., headquartered in Sunnyvale, California, develops semiconductors, supercomputers, and related software to power artificial intelligence deep-learning applications such as inference engines. Products include its wafer scale engine (WSE)-3 semiconductors, its CS-3 supercomputers, and its 'AI inference cloud' and 'AI training cloud' APIs.
ev_010 B-2 Groq, Inc. is an American artificial intelligence (AI) company that builds an AI accelerator application-specific integrated circuit (ASIC). The architecture was originally introduced as a Tensor Streaming Processor (TSP) but was later rebranded as a Language Processing Unit (LPU) following the widespread adoption of large language models after the breakthrough of ChatGPT.
ev_011 B-2 Huawei Technologies Co., Ltd. is a Chinese multinational corporation and technology company headquartered in Longgang, Shenzhen, Guangdong... founded in Shenzhen in 1987 by Ren Zhengfei.
ev_012 B-2 Annapurna Labs is Amazon's semiconductor division building Amazon's Nitro, Graviton, and Trainium chips product lines. Established in 2011 by Hrvoje Bilic and Nafea Bshara, and acquired by Amazon in January 2015... top 5 customers of TSMC.
ev_013 B-2 AlexNet is a convolutional neural network architecture developed for image classification tasks, notably achieving prominence through its performance in the ImageNet Large Scale Visual Recognition Challenge (ILSVRC)... regarded as the first widely recognized application of deep convolutional networks in large-scale visual recognition.
ev_014 B-2 Hopper is a graphics processing unit (GPU) microarchitecture developed by Nvidia. It is designed for datacenters and is used alongside the Lovelace microarchitecture.
ev_015 B-2 Blackwell is a graphics processing unit (GPU) microarchitecture developed by Nvidia as the successor to the Hopper and Ada Lovelace microarchitectures.
ev_016 B-2 CUDA is a proprietary parallel computing platform and application programming interface (API) developed by the American technology corporation Nvidia... CUDA was created by Nvidia starting in 2004 and was officially released in 2007.
ev_017 A-1 NVIDIA CORPORATION; LEI 549300S4KLFTLO7GSQ80; status ACTIVE; jurisdiction US-DE; HQ 2788 San Tomas Expressway, Santa Clara, CA 95051; creation date 1998-02-24; conformity FULLY_CORROBORATED.
ev_018 A-1 ADVANCED MICRO DEVICES, INC.; LEI R2I72C950HOYXII45366; status ACTIVE; jurisdiction US-DE; creation date 1969-05-01; FULLY_CORROBORATED.
ev_019 A-1 INTEL CORPORATION; LEI KNX4USFCNGPY45LOCE31; status ACTIVE; jurisdiction US-DE; creation date 1989-03-01.
ev_020 A-1 BROADCOM INC.; LEI 549300WV6GIDOZJTV909; HQ 1320 Ridder Park Drive, San Jose CA 95131; status ACTIVE; previous legal name 'Broadcom Limited'; creation 2018-01-17.
ev_021 A-1 ALPHABET INC.; LEI 5493006MHB84DD0ZWV18; HQ 1600 Amphitheatre Parkway, Mountain View, CA 94043; creation date 2015-07-23; FULLY_CORROBORATED.
ev_022 A-1 AMAZON.COM, INC.; LEI ZXTILKJKG63JELOEG630; HQ 410 Terry Ave North, Seattle WA 98109; creation 1996-05-28; FULLY_CORROBORATED.
ev_023 A-1 MICROSOFT CORPORATION; LEI INR2EJN1ERAN0W5ZP974; HQ One Microsoft Way, Redmond WA 98052-8300; creation 1993-09-22; FULLY_CORROBORATED.
ev_024 A-1 META PLATFORMS, INC.; LEI BQ4BKCS1HXDV9HN80Z93; HQ 1 Meta Way, Menlo Park CA 94025; creation 2004-07-29; FULLY_CORROBORATED.
ev_025 B-1 NVIDIA (NASDAQ: NVDA) today reported revenue for the fourth quarter ended January 26, 2025, of $39.3 billion, up 12% from the previous quarter and up 78% from a year ago.
ev_026 C-3 The data center GPU market is projected to reach USD 228.04 billion by 2030 from USD 119.97 billion in 2025, at a CAGR of 13.7% from 2025 to 2030.
ev_027 C-3 Revenues have shifted significantly since 2021, with Nvidia as the market share leader at 86% as of late 2025.
ev_028 B-2 Nvidia data center revenue Q3 FY2026: 51.2bn USD.
ev_029 C-4 The global AI Inference Market is projected to grow from USD 106.15 billion in 2025 to USD 254.98 billion by 2030, representing a compound annual growth rate of 19.2%.
ev_030 A-1 In November 2023, after BIS controlled the A800 and H800 chips, Nvidia announced H20, L20, and L2 chips for China. The H20 is a modified H100.
ev_031 B-2 Notably, the new controls cover NVIDIA's H20 chips and their equivalents, which were developed specifically for the Chinese AI market after prior export controls.
ev_032 A-1 BIS will now review export license applications for the Nvidia H200, AMD MI325X, and similar chips on a case-by-case basis provided certain conditions are met.
ev_033 A-2 Huawei Technologies plans to begin mass shipments of its advanced 910C artificial intelligence chip to Chinese customers as early as next month.
ev_034 B-2 Huawei appears to be the clear No. 1 choice for replacing Nvidia. In 2025, Ascend production is expected to be 400,000 units with plans for more.
ev_035 B-2 Ascend 910C, a dual-chiplet accelerator built using stacked HBM2E memory and a DaVinci NPU architecture tailored for AI workloads.
ev_036 B-1 AMD reported record revenue of $25.8 billion, we delivered more than $5 billion of AMD Instinct accelerator revenue.
ev_037 B-2 Google, Amazon, Microsoft, and Meta collectively plan to spend $725 billion on capex in 2026, up 77% from last year's record $410 billion.
ev_038 B-2 Google expects to spend $91 to $93 billion in capital expenditures for 2025, up from the $85 billion it previously forecast.
ev_039 B-2 Every major hyperscaler now designs its own AI ASIC, with Broadcom as the dominant co-design partner across Google TPU, Meta MTIA, ByteDance and OpenAI's Titan program.
ev_040 B-2 Nvidia is king in AI chips, but custom ASICs are gaining ground with Google, Amazon, Meta, Microsoft and OpenAI now making their own.
ev_041 A-2 Artificial intelligence startup Groq Inc. has raised $640 million in new funding, underscoring investor enthusiasm for innovation in chips for AI systems.
ev_042 B-3 Groq, which also sells data center computing power, previously raised $640 million at a $2.8 billion valuation in August 2024, making this more than a doubling of valuation in roughly a year.
ev_043 C-3 At the IPO price, Cerebras has a fully diluted valuation of $56.4 billion, according to CNBC. Cerebras filed for an IPO in 2024 but withdrew.
ev_044 A-1 Group 42 Holding Ltd (together with its affiliates, 'G42') accounted for 83% and 87%, respectively, of our total revenue for the year ended December 31, 2023 and the six months ended June 30, 2024.
ev_045 A-2 TSMC's advanced chip packaging technology, CoWoS, will be constrained through next year. CoWoS is considered essential for producing cutting-edge AI accelerators.
ev_046 A-2 We will also transition the CoWoS-S capacity to CoWoS-L... TSMC can make them but packaging has remained a bottleneck due to capacity constraints.
ev_047 C-3 2025e: 65,000 ~ 75,000 WPM; 2026e: 90,000 ~ 110,000 WPM. The rapid growth of TSMC advanced packaging — CoWoS (Chip-on-Wafer-on-Substrate).
ev_048 C-4 NVIDIA commands approximately 80-90% of the AI accelerator market by revenue as of 2025, generating over $100 billion annually from data center GPUs. Despite growing competition, NVIDIA's CUDA ecosystem, full-stack platform, and priority TSMC CoWoS allocation maintain dominance.
ev_049 C-3 The most immediate risk is the pending export of over 1.3 million NVIDIA H20 chips, worth more than $16 billion, to Chinese tech giants.
ev_050 B-2 As best we can figure, Marvell did $272 million in custom AI XPU revenue in Q4, up by a factor of 10.7X compared to a year ago, and electro-optics added to that.
No evidence matches the current filters.