AIMultiple ResearchAIMultiple ResearchAIMultiple Research
Accelerator
Updated on Apr 30, 2025

Top 20 AI Chip Makers: NVIDIA & Its Competitors in 2025

Based on our experience running AIMultiple’s cloud GPU benchmark with 10 different GPU models in 4 different scenarios, these are the top AI hardware companies for data center workloads. Follow the links to see our rationale behind each selection:

AI Chip VendorsBest For
1.
Revenue & volume leader. First choice for most buyers who can secure supply
2.
2nd player in terms of market valuation
3.
CPU market leader playing catch up in GPUs
4.
Hyperscaler with its Tranium chips
5.
Hyperscaler with its Ironwood and Trillium chips
Show More (5)
6.
Fast LLM inference
7.
High capacity hardware solutions for research labs & Fortune 500.
8.
Cloud-based AI training & inference
9.
Leading AI infrastructure in China
10.
Leading cloud AI solution in China
1.
NVIDIA logo
Revenue & volume leader. First choice for most buyers who can secure supply
2.
AMD logo
2nd player in terms of market valuation
3.
Intel logo
CPU market leader playing catch up in GPUs
4.
AWS logo
Hyperscaler with its Tranium chips
5.
GCP logo
Hyperscaler with its Ironwood and Trillium chips
6.
Groq logo
Fast LLM inference
7.
SambaNova Systems logo
High capacity hardware solutions for research labs & Fortune 500.
8.
Microsoft Azure logo
Cloud-based AI training & inference
9.
Huawei logo
Leading AI infrastructure in China
10.
Alibaba logo
Leading cloud AI solution in China

20+ AI chip makers by category

These chip makers focus on datacenter chips:

Last Updated at 04-30-2025
VendorCategorySelected AI chip*

NVIDIA

Leading producer

Blackwell Ultra

AMD

Leading producer

MI400

Intel

Leading producer

Gaudi 3

AWS

Public cloud & chip producer

Trainium3

Alphabet

Public cloud & chip producer

Ironwood

Alibaba

Public cloud & chip producer

ACCEL**

IBM

Public cloud & chip producer

NorthPole

Huawei

Public cloud & chip producer

Ascend 920

Groq

Public AI cloud & chip producer

LPU Inference Engine

SambaNova Systems

Public AI cloud & chip producer

SN40L

Microsoft Azure

Public AI cloud & chip producer

Maia 100

Untether AI

Public AI chip producer

speedAI240

Apple

Chip producer

M4

Meta

Chip producer

MTIA v2

Cerebras

AI startup

WFE-3

d-Matrix

AI startup

Corsair

Rebellions

AI startup

Rebel

Tenstorrent

AI startup

Wormhole

_etched

AI startup

Sohu

Extropic

AI startup

OpenAI

Upcoming producer

TBD

Graphcore

Other producers

Bow IPU

Myhtic

Other producers

M2000

*The selected models are based on latest announcements.

**ACCEL was developed by Chinese scientists in collaboration with Alibaba and China’s Semiconductor Manufacturing International Corporation (SMIC)​ 1

Sorting is by category. Vendors are ranked by estimated market share within top 3 categories (i.e. leading producer, public cloud, public AI cloud) because sales numbers or cloud usage can be estimated. Vendors in the last 3 categories (i.e. AI startup, upcoming producer, other producers) are sorted alphabetically.

5 mobile AI chip providers

Last Updated at 12-23-2024
VendorSelected Chips*Used in

Apple

A18 Pro, A18

iPhone 16 Pro, iPhone 16

Huawei

Kirin 9000S

Huawei Mate 60 series

MediaTek

Dimensity 9400, Dimensity 9300 Plus

Oppo Find X8, Vivo X200 series, Samsung Galaxy Tab S10 Plus, Tab S10 Ultra

Qualcomm

Snapdragon 8 Elite (Gen 4), Snapdragon 8 Gen 3

Samsung Galaxy S25 Ultra, Xiaomi 14, OnePlus 12, Samsung Galaxy S24 series

Samsung

Exynos 2400, Exynos 2400e

Exynos 2400, Exynos 2400e

*Most popular & recent chips are selected.

5 Edge AI Chips

The demand for low-latency processing has driven innovation in edge AI chips. These chips’ processors are designed to perform AI computations locally on devices rather than relying on cloud-based solutions:

Last Updated at 04-21-2025
ChipPerformance (TOPS)*Power ConsumptionApplications

NVIDIA Jetson Orin

275

10-60W

Robotics, Autonomous Systems

Google Edge TPU

4

2W

IoT, Embedded Systems

Intel Movidius Myriad X

4

5W

Drones, Cameras, AR Devices

Hailo-8

26

2.5-3W

Smart Cameras, Automotive

Qualcomm Cloud AI 100 Pro

400

Varies

Mobile AI, Autonomous Vehicles

*These are the maximum quoted values by the vendors. TOPS is tera operations per second.

Which are the leading AI chip producers?

1. NVIDIA

NVIDIA has been designing graphics processing units (GPUs) for the gaming sector since 1990s. NVIDIA is a fabless chip manufacturer that outsources most of its chip manufacturing to TSMC. Its main businesses include:

Desktop AI solutions

DGX Spark (formerly Project Digits) is a desktop AI supercomputer for AI engineers and data scientists that is:

  • expected to cost around $3k.
  • be about the same size as a Mac mini and powered by the NVIDIA GB10 Grace Blackwell Superchip with 128GB memory.
  • capable of handling LLM inference and fine-tuning for models with up to 200 billion parameters, leveraging NVLink-C2C for high-speed CPU+GPU communication.

Datacenter solutions

The company makes AI chips following its Ampere, Hopper and most recently Blackwell architectures. Thanks to the generative AI boom, NVIDIA had excellent results in the past years, reached a trillion in valuation and solidified its status as the leader of GPU and AI hardware markets. The following chart shows how NVIDIA’s revenue in this segment has grown over the years and how it has become the company’s primary source of income.

NVIDIA’s chipsets are designed to solve business problems in various industries. DGX™ A100 and H100 have been successful flagship AI chips of Nvidia, designed for AI training and inference in data centers.2 NVIDIA followed up on these with

  • H200, B300 and GB300 chips
  • HGX servers such as HGX H200 and HGX B300 that combine 8 of these chips
  • NVL series and GB200 SuperPod that combine even more chips into large clusters.3

Cloud GPUs

Thanks to the strength of its datacenter offering, NVIDIA almost has a monopoly on the cloud AI market with most cloud players offering only NVIDIA GPUs as cloud GPUs.

NVIDIA also launched its DGX Cloud offering providing cloud GPU infrastructure directly to enterprises bypassing cloud providers.

GPUs for graphics

Xbox uses a chipset codeveloped by NVIDIA and Microsoft. NVIDIA’s GPUs for retail users include GeForce series.

Recent developments

NVIDIA Dynamo

NVIDIA Dynamo, announced at GTC 2025, is a new open-source inference framework designed for high-throughput, low-latency deployment of generative AI models in distributed environments, boosting request serving by up to 30x on NVIDIA Blackwell as shown in the figure below. This framework, compatible with popular tools like PyTorch and TensorRT-LLM, utilizes innovations such as disaggregated inference stages and dynamic GPU scheduling to optimize performance and reduce costs. Available on GitHub for developers and included in NVIDIA NIM microservices for enterprise solutions, Dynamo facilitates scalable and cost-effective generative AI serving from single to multi-GPU systems.4

Figure 1. NVIDIA Dynamo significantly accelerates AI model performance. Specifically, it provides a 30x speedup for the DeepSeek-R1 671B model on the NVIDIA GB200 NVL72 platform. It also more than doubles the performance of the Llama 70B model when using NVIDIA Hopper GPUs.5

DeepSeek

Release of DeepSeek’s R1 showed that state of the art models could be trained with a relatively small number of GPUs. This led to a reduction in NVIDIA’s stock price. Though this is not investment advice, this can be positive for NVIDIA since the more utility computing power provides, the more widely it should be used (i.e. Jevons paradox6 ).

However, given that GPU systems’ performance is improved multiple times annually thanks to improvements in chip design and interconnect, buyers would be wise to not buy beyond their annual needs since this can lead to owning outdated systems.

Tariffs & export restrictions

NVIDIA is required to sell less powerful versions of its chips in China due to export restrictions and this incentivized the Chinese government and chip industry to develop competitive local chips.

While Chinese chips underperform latest NVIDIA chips, they are gaining ground and could pose a threat to NVIDIA’s market dominance in the future.

Inference Market Competition

While NVIDIA dominates the AI “training” market, competition is heating up in “inference” – the deployment of AI models for real-world tasks. Companies like AMD and numerous startups, including Untether AI and Groq, are developing chips that aim to provide more cost-effective inference solutions, particularly focusing on lower power consumption.

New “reasoning” AI techniques, which demand more computing power. NVIDIA believes that reasoning will favor its architecture in the long run and expects the inference market to eventually dwarf the training market in size, even if its market share is smaller.7

2. AMD

AMD is a fabless chip manufacturer with CPU, GPU and AI accelerator products.

AMD launched MI300 for AI training workloads in June 2023 and is competing with NVIDIA for market share. There are startups, research institutes, enterprises and tech giants that adopted AMD hardware in 2023 since Nvidia AI hardware has been difficult to procure due to rapidly increasing demand with the rise of generative AI triggered by the launch of ChatGPT.8 9 10 11

AMD will be releasing MI350 series to replace MI300 and compete with NVIDIA’s H200. AMD claims that MI325X, another recent chip, has market leading inference performance.12

AMD is also working with machine learning companies like Hugging Face to enable data scientists to use their hardware more efficiently.13

The software ecosystem is critical as hardware performance relies heavily on software optimization. For example, AMD and NVIDIA had a public disagreement over benchmarking H100 and MI300. The focus of the disagreement was the package and floating point to use in the benchmark. According to the latest benchmarks, it appears that MI300 is better or on par with H100 for inferencing on a 70B LLM.14 15

Software

While AMD hardware is catching up to NVIDIA, its software lags behind in terms of usability. While CUDA works out of the box for most tasks, AMD software requires significant configuration.16

Ecosystem

Like NVIDIA, AMD is selectively investing into users of its solutions to drive adoption of its hardware.17

3. Intel

Intel is the largest player in the CPU market and has a long history of semiconductor development. Unlike NVIDIA and AMD, Intel uses its own foundry to build its chips.

Gaudi3 is the latest AI accelerator processor from Intel. 18 However, Intel’s sales guidance for Gaudi3 was ~$500M for 2024 which is significantly lower compared to the billions that AMD is projecting to earn in 2024.

Intel is experiencing governance issues as shown by its CEO Pat Gelsinger’s departure in December 2024. A significant share of Intel’s board members lack experience in leading a semiconductor company in an operational manner.19 After the departure of its CEO, Intel’s strategy in the AI and foundry markets is not yet clear.

Which public cloud providers produce AI chips?

4. AWS

AWS produces Tranium chips for model training and Inferentia chips for inference. Though AWS is the market leader in public cloud, it started building its own chips after Google.

Hundreds of thousands of Tranium2 chips are used to form the Project Rainier cluster which powers LLM developer Anthropic’s models.

5. Google Cloud Platform

Google Cloud TPU is the purpose-built machine learning accelerator chip that powers Google products like Translate, Photos, Search, Assistant, and Gmail. It can be used via the Google Cloud as well. Google announced TPUs in 2016.20 Latest Trillium TPU is the 6th generation.21

Google has introduced Ironwood. This latest generation is specifically designed for complex “thinking models” like LLMs and MoEs, offering massive parallel processing (4,614 TFLOPs per chip) and scaling up to 42.5 Exaflops in 9,216-chip pods.22

Ironwood delivers significant advancements over Trillium, including 2x better power efficiency, 6x the High Bandwidth Memory capacity (192 GB/chip), 4.5x the HBM bandwidth (7.2 TBps/chip), and 1.5x the Inter-Chip Interconnect speed (1.2 Tbps). It also features an enhanced SparseCore for large embeddings. Google also produces the much smaller Edge TPU for different needs, designed for deployment on edge devices like smartphones and IoT hardware.

6. Alibaba

Alibaba produces chips like Hanguang 800 for inference. However, some North American, European and Australian organizations (e.g. those in the defense industry) may not prefer to use Alibaba Cloud for geopolitical reasons.

7. IBM

IBM announced its latest deep learning chip, artificial intelligence unit (AIU), in 2022.23 . IBM is considering using these chips to power its watson.x generative AI platform.24

AIU builds on “IBM Telum Processor” which powers AI processing capabilities of IBM Z mainframe servers. At launch, Telum processors’ highlighted use cases included fraud detection.25

IBM also demonstrated that merging compute and memory can lead to efficiencies. These were demonstrated in the NorthPole processor prototype.26

8. Huawei

Huawei’s HiSilicon Ascend 910C is part of the Ascend 910 family of chips introduced in 2019.

Due to sanctions, AI labs in China can not buy the newest highest performance chips from US firms like NVIDIA or AMD. Therefore, they are experimenting with Ascend 910C.

Huawei’s cloud is hosting DeepSeek models and a researcher at DeepSeek claims that it can reach 60% of NVIDIA H100 inference performance. 27 28

Which cloud AI providers produce their own chips?

These providers do not have public clouds with comprehensive capabilities like the hyperscalers. They provide limited cloud services typically focused on AI inference. We were able to sign up to these services without talking to sales teams:

8. Groq

Groq has been founded by former Google employees. The company represents LPUs, a new model for AI chip architecture, that aims to make it easier for companies to adopt their systems. The startup has already raised around $350 million and produced its first models such as GroqChip™ Processor, GroqCard™ Accelerator, etc.

The company is focused on LLM inference and released benchmarks for Llama-2 70B.29

Recently, Groq secured a significant $1.5 billion investment commitment from Saudi Arabia to expand the delivery of its advanced AI chips to the country. This investment will be used to expand Groq’s existing data center in Dammam, Saudi Arabia, built in partnership with Aramco Digital.30

In Q1 2024, the company shared that 70k developers signed up on its cloud platform and built 19k new applications.31

On March 1, 2022, Groq had acquired Maxeler, which has high performance computing (HPC) solutions for financial services.32

9. SambaNova Systems

SambaNova Systems was founded in 2017 with the goal of developing high-performance, high-precision hardware-software systems for high volume generative AI workloads. The company has developed the SN40L chip and raised more than $1.1 billion in funding.33 34

It is important to note that SambaNova Systems also leases its platform to businesses.35 AI platform as service approach of SambaNova Systems makes their systems easier to adopt and encourages hardware reuse for circular economy.

Which are the leading AI chip startups?

We would also like to introduce some startups in the AI chip industry whose names we may hear more often in the near future. Even though these companies were founded only recently, they have already raised millions of dollars.

10. Cerebras

Cerebras was founded in 2015 and is the only major chip maker focusing on wafer-scale chips. 36 Wafer scale chips have advantages in parallelism compared to GPUs thanks to their higher memory bandwidth. However, designing and manufacturing such chips is an emerging technology.

Cerebras chips’ include:

  • WSE-1 with 1.2 trillion transistors and 400k processing cores.
  • WSE-2 with 2.6 trillion transistors and 850k cores, announced in April 2021. It leveraged TSMC’s 7nm process
  • WSE-3 with 4 trillion transistors and 900k AI cores, announced in March 2024. It leverages TSMC’s 5nm process37

Celebra’s system works with pharmaceutical companies such as AstraZeneca and GlaxoSmithKline and research labs that rely on it for simulations. It also targets LLM makers since its chips can lower inference costs for frontier models.

Cerebras also offers its chips on its cloud to enterprises.

11. d-Matrix

d-Matrix follows a novel approach ditching the traditional von Neumann architecture in favor of in-memory compute. While this approach has the potential to resolve the bottleneck between memory and compute, it is a new and yet unproven approach.38

12. Rebellions

Korea based startup raised $124M in 2024 and is focused on LLM inference.39

Rebellions merged with another Korean semiconductor design firm, SAPEON and reached a unicorn valuation in 2024.40

13. Tenstorrent

Tenstorrent produces the Wormhole chip, desktop machines for researchers and servers (e.g. Tenstorrent Galaxy) powered by Wormhole chips. The company also provides the software stack for its solution.

Tenstorrent raised $700M at a valuation of more than $2.6 billion from investors including Jeff Bezos in December 2024.41

14. _etched

Their approach sacrifices flexibility for efficiency by burning the transformer architecture into their chips.

The team claims

  • To have built the world’s first transformer ASIC, Sohu.
  • That 8 Sohu chips can generate >500,000 tokens/second. This is an order of magnitude more than what 8 NVIDIA B200s can achieve.

Currently, these are based on team’s internal measurements. AIMultiple team has not yet come across any benchmarks or client references. We are curious about:

  • What happens when the model becomes outdated? Do users need to buy a new chip or can the old chip be reconfigured with the next model?
  • How did they run their benchmark? Which quantization and model were used?

We’ll be updating this as soon as the _etched team releases more details. It will be interesting to see whether burning model to chips will be sustainable given the release of new models every few months.

15. Extropic

Extropic raised a $14M round in late 2023 to use thermodynamics for computing. The company hasn’t released a chip yet.

What are upcoming AI hardware producers?

Though these are compelling AI hardware solutions, there are currently limited benchmarks on their effectiveness since they are newcomers to the market.

16. Apple

Apple’s project ACDC is reported to be focused on building chips for AI inference.42 Apple is already a major chip designer with its internally designed semiconductors used in iPhone, iPads and Macbooks.

17. Meta

Meta Training and Inference Accelerator (MTIA) is a family of processors for AI workloads such as training Meta’s LLaMa models.

The latest model is Next Gen MTIA which is based on TSMC 5nm technology and is claimed to have 3x improved performance vs. MTIA v1. MTIA will be hosted in racks containing up to 72 accelerators.43

MTIA is currently for Meta’s internal usage. However, in the future, if Meta launched a LLaMa based enterprise generative AI offering, these chips could power such an offering.

18. Microsoft Azure

At Hot Chips 2024, Microsoft unveiled Maia 100, their first custom AI accelerator designed to optimize large-scale AI workloads in Azure through hardware and software co-optimization. This vertically integrated system features a custom chip built on TSMC’s N5 process with advanced memory and interconnect technology, tailored for high throughput and diverse data formats. Maia 100 offers developers flexibility and portability through its SDK, allowing quick deployment of models written in PyTorch and Triton while achieving efficient data handling and workload performance.44

19. OpenAI

OpenAI is finalizing the design of its first AI chip with Broadcom and TSMC using TSMC’s 3-nanometer technology. OpenAI’s chip team’s leadership has experience with designing TPUs at Google and they aim to have their chip mass produced in 2026.45

What are other AI chip producers?

20. Graphcore

Graphcore is a British company founded in 2016. The company announced its flagship AI chip as IPU-POD256. Graphcore has already been funded with around $700 million.

Company has strategic partnerships between data storage corporations like DDN, Pure Storage and Vast Data. Graphcore’s AI chips serve research institutes like Oxford-Man Institute of Quantitative Finance, University of Bristol and Berkeley University of California.

The company’s long term viability was at risk as it was losing ~$200M per year.46 It got acquired by Softbank for $600m+ in October 2024.47

21. Mythic

Mythic was founded in 2012 and is focused on edge AI. Mythic follows an unconventional path, an analog compute architecture, that aims to deliver power-efficient edge AI computing.

It developed products such as M1076 AMP, MM1076 key card, etc., and has already raised about $165 million in funding.48

Mythic laid of most of its staff and restructured its business with its funding round in March 2023.49

Foundry partners and TSMC’s role

As the world’s leading pure-play foundry, TSMC manufactures semiconductors based on customer designs rather than creating its own chips, distinguishing it from companies like NVIDIA and AMD. While Samsung Foundry and Intel Foundry Services compete in this space, TSMC maintains a technological edge.

Its advanced process technologies, particularly its pioneering 5nm and 3nm nodes, provide the essential combination of performance and energy efficiency required for cutting-edge AI applications, as shown in its manufacturing partnerships with the AI chip designers listed below:

Last Updated at 03-14-2025
AI Chip MakerTSMC Manufactured

NVIDIA

GB200

AMD

MI325X

Intel

AI-related components

Amazon

Trainium

SambaNova Systems

SN40L

Cerebras

WSE-3

Alibaba

Hanguang 800

Meta

MTIA

Microsoft

Maia 100

Expansion plans

TSMC is seeking Nvidia, AMD, Broadcom, and Qualcomm to invest in a joint venture to run Intel’s foundry division, retaining operational control but less than 50% ownership. This initiative, backed by the Trump administration, comes after TSMC announced plans for a significant U.S. investment and aims to revive Intel and strengthen U.S. chip manufacturing. The deal faces challenges due to process differences but builds on TSMC’s strengths as a leading foundry.50 51

What are the AI chip makers in China?

Since the US sanctions prevented many Chinese companies from acquiring the most advanced AI chips from AMD and NVIDIA, Chinese buyers have increased their purchases from local producers.

Other than Huawei and Alibaba covered above, these are the leading AI chip producers in China:

  • Cambricon focuses on AI hardware and expects ~$150M in sales in its latest year of operations.52
  • Baidu is using Kunlun chips in its cloud and is designing the 3rd generation chip. Kunlun 2 was comparable to NVIDIA A100.
  • Biren, founded by NVIDIA alumni, produces BR106 & BR110 GPU chips.
  • Moore Threads produces MTT S2000 GPUs.

FAQ

What are other companies in the AI chip ecosystem?

Market map of semi-conductor value chain

Chips and the equipment that build them are the most complex machines ever built by humans. Though there are many companies in the semiconductor ecosystem, we focused on chip designers like NVIDIA in this article.
Most chip designer outsource chip manufacturing to foundries like TSMC. Foundries use lithography equipment produced by companies like ASML to manufacture these chips. The ecosystem is supported by providers like Arm and Synopsys that supply IP and design tools.

Why is AI hardware so important?

As seen above, increasing number of parameters, dataset size and compute led generative AI models to become more accurate. To build better deep learning models and power generative AI applications, organizations require increased computing power and memory bandwidth.
Powerful general-purpose chips (such as CPUs) cannot support highly parallelized deep learning models. Therefore, AI chips (e.g. GPUs) that enable parallel computing capabilities are increasingly in demand.
Hyperscalers are responding to this by designing their own chips which takes years. The rest need to follow one of these routes to build their own AI models: Rent capacity from cloud GPU providers or buy hardware from the top AI chip vendors listed in this article.
AI hardware is also called neural processing units (NPUs), AI accelerators or deep learning processors (DLPs).

Further reading

You can also check our sortable list of companies working on AI chips.

You might enjoy reading our articles on TinyML and accelerated computing.

If you have questions about how AI hardware can help your business, we can help:

Find the Right Vendors

References

Share This Article
MailLinkedinX
Cem has been the principal analyst at AIMultiple since 2017. AIMultiple informs hundreds of thousands of businesses (as per similarWeb) including 55% of Fortune 500 every month.

Cem's work has been cited by leading global publications including Business Insider, Forbes, Washington Post, global firms like Deloitte, HPE and NGOs like World Economic Forum and supranational organizations like European Commission. You can see more reputable companies and resources that referenced AIMultiple.

Throughout his career, Cem served as a tech consultant, tech buyer and tech entrepreneur. He advised enterprises on their technology decisions at McKinsey & Company and Altman Solon for more than a decade. He also published a McKinsey report on digitalization.

He led technology strategy and procurement of a telco while reporting to the CEO. He has also led commercial growth of deep tech company Hypatos that reached a 7 digit annual recurring revenue and a 9 digit valuation from 0 within 2 years. Cem's work in Hypatos was covered by leading technology publications like TechCrunch and Business Insider.

Cem regularly speaks at international technology conferences. He graduated from Bogazici University as a computer engineer and holds an MBA from Columbia Business School.

Next to Read

Comments

Your email address will not be published. All fields are required.

2 Comments
Dave
Aug 29, 2022 at 05:49

You forgot to include Tesla with their DOJO supercomputer. From the ground-up, the supercomputer was specifically designed for machine learning and image recognition – which means that every component was designed for it including, but not limited to, PCI board design, CPU, RAM, cooling, power, scalable hardware design and software. If I’m not mistaken, the AI is also the second most widely tested and used in the “wild”, just below that of Google due to Google using it in their Search.

Cem Dilmegani
Sep 06, 2022 at 13:52

Thank you for your feedback, Dave!
Here we are only covering companies that sell the chips that they produce. Therefore, companies like Tesla that build supercomputers for their own use or companies that embed chips in their products are out of our scope.

thayyil
Mar 19, 2022 at 11:48

surprised that brainchip (akida) missing in this report. any reasons?

Cem Dilmegani
Nov 18, 2022 at 07:36

All included companies here raised $100+M. Last time we collected the data, that wasn’t the case for akida. Why don’t you reach out to us at info@aimultiple.com and let’s discuss why it should be included.

Thank you!

Related research