10 top AI hardware and chip-making companies in 2025

10 top companies in the AI hardware market

The following AI hardware and chip-making companies are listed in alphabetical order.

Alphabet

Alphabet, Google's parent company, offers various products for mobile devices, data storage and cloud infrastructure.

TPU v6e is the newest Trillium chip, released in October 2024. This chip is similar to TPU v5e, with a pod size of 256 chips and four ports per chip. However, the TPU v6e chip has a peak compute performance per chip that is 4.7 times higher than that of TPU v5e. Ironwood TPU v7 will have 9,216 chips per pod and is expected to be released at the end of 2025.

Alphabet has focused on producing powerful AI chips to meet the demand for large-scale projects. In December 2024, Alphabet released a new quantum computing chip, Willow. With 105 qubits and the ability to scale up, the Willow chip reduces error in quantum computing faster and more accurately than its predecessors.

AMD

AMD released its latest CPU microarchitecture chip design, Zen 5, and its next generation of Ryzen processors, Ryzen 300 and 200 Series, at the beginning of January 2025.

AMD's Instinct MI300 Series chip, MI325X, was released in 2024. This upgrade from MI300X has a larger bandwidth of 6 TBps. The MI350 series, including the MI355X chip, was released in June 2025. The MI355X chip is 4 times faster than the MI300X. These AI GPU accelerators are meant to rival Nvidia's Blackwell B100 and B200.

Apple

Apple Neural Engine, specialized cores based on Apple chips, has furthered the company's AI hardware design and performance. Neural Engine led to the M1 chip for MacBooks. Compared to the generation before, MacBooks with an M1 chip are 3.5 times faster in general performance and five times faster in graphics performance.

After the success of the M1 chip, Apple announced further generations. As of 2024, Apple has released the M4 chip. The M4 chip has a Neural Engine that is three times faster than the M1 chip and 1.5 times faster than the M2. The M5 chip is expected to be released in the fall of 2025.

Apple and Broadcom are developing an AI-specific server chip, Baltra. This chip is expected to be released in 2026 but will only be used internally by the companies.

AWS

AWS has switched its focus from cloud infrastructure to chips. Its Elastic Compute Cloud (EC2) Trn1 instances are purpose-built for deep learning and large-scale generative models. They use AWS Trainium AI accelerator chips to function.

AWS Inferentia is a machine learning chip that generates high-performance inference predictions at a low cost. Trainium accelerators train models, while Inferentia accelerators deploy models. The trn1.2xlarge instance was the first iteration. It had one Trainium accelerator, 32 GB of instance memory and 12.5 Gbps network bandwidth.

AWS now has the EC2 Trn2 instance and UltraServer. A Trn2 instance has 16 Trainium2 chips, 1.5 TB of accelerator memory and 46 TBps bandwidth. The Trn2 UltraServer has 64 Trainium2 chips over four Trn2 instances, 6 TB of shared accelerator memory and 185 TBps bandwidth. Trainium3 chips are expected to be released in late 2025. Trainium3-powered UltraServers will have four times better performance than Trainium2 UltraServers.

In 2024, AWS released Graviton4, a 96-core ARM-based processor ideal for a range of cloud workloads, such as databases, web servers and high-performance computing. The fourth generation of AWS's Graviton processor, which powers EC2 R8g instances, provides up to 30% better performance and has three times more vCPUs and memory than the Graviton3.

Cerebras Systems

Cerebras is making a name for itself with the release of its third-generation wafer-scale engine, WSE-3. WSE-3 is deemed the fastest processor on Earth with 900,000 AI cores on one unit. Every core has access to 21 petabytes per second of memory bandwidth.

Compared to Nvidia's H100 chip, WSE-3 has 7,000 times larger bandwidth, 880 times more on-chip memory and 52 times more cores. This WSE-3 chip is also 57 times larger in area, so more space is necessary to house the chip in a server.

IBM

Telum was IBM's first specialized AI chip, and Telum II is expected to be released in late 2025. IBM has also set out to design a powerful successor to rival its competitors.

In 2022, IBM created the Artificial Intelligence Unit. The AI chip is purpose-built and runs better than the average general-purpose CPU. Based on a similar architecture, IBM will release the Spyre Accelerator in 2025. Spyre will have 32 AI accelerator cores and contain 25.6 billion transistors over 14 miles of wire.

IBM is working on the NorthPole AI chip, which does not have a release date. NorthPole differs from IBM's TrueNorth chip. The NorthPole architecture is structured to improve energy use, decrease the amount of space the chip takes up and provide lower latency. The NorthPole chip is set to mark a new era of energy-efficient chips.

Intel

Intel has made a name for itself in the CPU market with its AI products.

Xeon 6 processors launched in 2024 and have been shipped to data centers. These processors offer up to 288 cores per socket, enabling faster processing time and enhancing the ability to perform multiple tasks at once.

Intel has released the Gaudi 3 GPU chip, which competes with Nvidia's H100 GPU chip. The Gaudi 3 chip trains models 1.5 times faster, outputs 1.5 times faster results, and uses less power than Nvidia's H100 chip.

Intel has canceled the launch of the Falcon Shores AI GPU chip. The Jaguar Shores chip, the successor to the Gaudi 3 chips, is still set to launch in 2026. However, the company is focusing away from standalone AI accelerators.

In late 2024, Intel released the Core Ultra AI Series 2 processors. The release included multiple processors under the Core Ultra 200 series, including 200H, 200HX, 200S and 200V. Each series focuses on specific features, such as enhanced security, AI capabilities, performance and energy efficiency. The Core Ultra 200 processor series is designed for desktop and mobile platforms, creating AI PCs.

Nvidia

Nvidia became a strong competitor in the AI hardware market when its valuation surpassed $1 trillion in early 2023. The company's current work includes its B200 chip, Blackwell GPU microarchitecture and Vera Rubin. Nvidia also offers AI-powered hardware for the gaming sector.

The Blackwell GPU microarchitecture is replacing the Grace Hopper platform. Blackwell is 2.5 times faster and 25 times more energy-efficient than its predecessors. The Blackwell microarchitecture is designed to increase efficiency with scientific computing, quantum computing, AI and data analytics. The B300 chip series, or Blackwell Ultra, is expected to be released in the second half of 2025.

Vera Rubin is Nvidia's next-generation GPU superchip architecture, expected to be released in late 2026. It combines the Vera CPU with the Rubin GPU, the successor to Blackwell.

Qualcomm

Although Qualcomm is relatively new in the AI hardware market compared to its counterparts, its experience in the telecom and mobile sectors makes it a promising competitor.

Qualcomm's Cloud AI 100 chip beat Nvidia H100 in a series of tests. One test was to see the number of data center server queries each chip could carry out per watt. Qualcomm's Cloud AI 100 chip totaled 227 server queries per watt, while Nvidia H100 hit 108. The Cloud AI 100 chip also managed to net 3.8 queries per watt compared to Nvidia H100's 2.4 queries during object detection.

In 2024, Qualcomm released Snapdragon 8s Gen 3, a mobile chip that supports 30 AI models and has generative AI features, like image generation and voice assistants. Later in the year, the company released the newest version, Snapdragon 8 Elite, which improved AI performance by 45%. The Snapdragon 8 Elite Gen 2 is expected to be unveiled in September 2025.

Tenstorrent

Tenstorrent builds computers for AI and is led by the same man who designed AMD's Zen chip architecture, Jim Keller. Tenstorrent has multiple hardware products, including its Wormhole processors and Galaxy servers, which create the Galaxy Wormhole Server.

Tenstorrent released the Blackhole series, an AI accelerator, in April 2025. It has 16 RISC-V cores and 32 GB of GDDR6 memory per chip. The p100a chip has 120 Tensor Cores and 28 GB of GDDR6. The p150a has 140 Tensor Cores and 32 GB of GDDR6. Both chips operate at up to 300 Watts.

Wormhole n150 and n300 are Tenstorrent's scalable GPUs. N300 nearly doubles every spec of n150. These chips are for network AI and are put into Galaxy modules and servers. Each server holds up to 32 Wormhole processors, 2,560 cores and 384 GB of GDDR6 memory.

Editor's note: This article was updated in July 2025 to reflect the AI chips and processors each company has to offer.

Devin Partida is editor in chief of ReHack.com and a freelance writer. She has knowledge of niches such as biztech, medtech, fintech, IoT and cybersecurity.

Kelly Richardson is site editor for Informa TechTarget's SearchDataCenter site.