Best Computing Power Stocks: Barclays Likes These Key AI Infrastructure Targets

Investing.com - Barclays analysts have identified a group of leading companies that are expected to benefit from the construction of artificial intelligence infrastructure. Predictions for the tech sector indicate that annual AI infrastructure spending by Western hyperscale cloud providers and AI labs could surpass $1 trillion before peaking in 2028.

The investment bank has compiled a list of over 400 publicly traded and private companies that play a critical role in 19 subcategories of digital and power infrastructure and are expected to gain additional growth potential from sovereign AI programs and the Chinese market.

Nvidia

As a bellwether in the AI space and the company with the highest market value globally, Nvidia maintains its global leadership in AI infrastructure through its integrated Blackwell and Rubin GPU architectures.

Nvidia’s CEO recently announced that the company has over $1 trillion in order visibility for the Blackwell and Rubin platforms during the period from 2025 to 2027.

Microsoft

This tech giant has deployed proprietary Maia AI accelerators and Cobalt CPUs to optimize Azure’s AI infrastructure.

Microsoft announced the launch of Nvidia’s Nemotron model on Microsoft Foundry and the deployment of a new Vera Rubin NVL72 supercomputer in its data centers.

Alphabet

Google’s parent company designs proprietary Tensor Processing Units (TPUs) for internal AI training and external Google Cloud services.

Recently, Alphabet’s Google launched new features for its Gemini AI assistant, including chat import tools and a new Gemini 3.1 Flash Live audio model aimed at developers.

Meta

This social media company is developing an internal Meta Training and Inference Accelerator, known as MTIA.

Reports indicate that Meta Platforms is in discussions with the Adani Group to explore a potential partnership for building data centers in India.

Amazon

This e-commerce and cloud computing leader has developed dedicated AWS Trainium and Inferentia chips to optimize cost-effectiveness for cloud customers.

Reports suggest that Amazon Web Services (AWS) is developing AI to automate functions in its sales department, while JPMorgan notes an increase in demand for AWS services.

AMD

This semiconductor manufacturer provides high-performance Instinct GPUs and EPYC CPUs for data center training and inference workloads.

Advanced Micro Devices (AMD) announced a collaboration with Celestica to develop the Helios rack-level AI platform for data center infrastructure and signed a multi-year licensing agreement with Adeia.

Broadcom

The company provides custom AI accelerators and XPU design services for the hyperscale data center commercial market.

Broadcom and Carahsoft secured a five-year, $970 million contract to provide cloud software to the U.S. Defense Information Systems Agency, and the company has also begun shipping its Tomahawk 6 switch chips in volume.

Alibaba

This Chinese tech company designs proprietary GPUs for internal and external cloud services to meet AI computing demands.

Alibaba recently launched its next-generation AI chip, the Xuantie C950, which is built on the open-source RISC-V architecture and is reported to perform over three times better than its predecessor.

Arm

This chip design company provides infrastructure for almost all hyperscale custom silicon chips, including Cobalt, Graviton, and Axion, as well as Nvidia’s Grace-Blackwell clusters.

Following the “Arm Everywhere” event, Arm received upgrades and target price increases from several analysts after launching a new AGI CPU chip aimed at agent-based AI.

TSMC

This foundry giant manufactures the world’s most advanced AI chips and provides critical CoWoS advanced packaging capacity.

TSMC reported a 29.9% increase in revenue year-over-year for the first two months of 2026.

Other computing targets that Barclays emphasizes as crucial to digital and power infrastructure construction include Intel, Marvell, Qualcomm, and Tencent.

The bank predicts that the market consensus expectation for capital expenditures by hyperscale cloud providers could have over $300 billion of upside potential, as recursive self-improvement reduces AI training demands, and spending growth is expected to slow in the coming years.

This article was translated with the assistance of artificial intelligence. For more information, please see our terms of use.