AAPL

TSLA

AMZN

META

AMD

NVDA

PEP

COST

ADBE

GOOG

AMGN

HON

INTC

INTU

NFLX

ADP

SBUX

MRNA

Markets

ARM

The Next Phase of the AI Chip Supercycle Is Here. 1 Growth Stock Buy Before It Is Too Late (Hint: It Isn't Intel or Broadcom)

May 12, 2026 — 09:35 am EDT

Written by

Harsh Chauhan for

The Motley Fool->

Key Points

- Arm Holdings has the potential to become one of the most important chip companies in AI inference.

- The company's energy-efficient processor designs are already in strong demand, and its move into making its own chips could unlock a new growth opportunity.

- Arm's ambitious earnings-per-share guidance for fiscal 2031 suggests this stock is poised for more upside.

- 10 stocks we like better than Arm Holdings ›

Artificial intelligence (AI) has been the biggest growth driver for the stock market in recent years, as the proliferation of this technology has positively impacted multiple industries and applications.

The semiconductor industry has been one of the biggest beneficiaries of the AI boom. Chips have been deployed in massive numbers in data centers over the past few years to train AI models. But now, a significant portion of chips are used for AI inference applications. That's not surprising, as inference is the process of putting a trained AI model to work in the real world by feeding it new data so it can make decisions in real-time, generate responses, and provide recommendations.

Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »

In fact, inference is going to be bigger than the training phase. Here's why.

Image source: The Motley Fool.

AI inference will command more computing power than training, and that's great news for this company

While training is the first phase of teaching AI models with the help of large datasets, inference is the second phase, where the model is put to work. Not surprisingly, demand for AI inference is increasing rapidly as enterprises and consumers seek to unlock productivity gains by integrating AI into their daily operations.

McKinsey estimates that inference will overtake training and become the dominant workload in AI data centers by 2030, accounting for more than half of all AI computing power. What's more, inference workloads are poised to command 30% to 40% of total data center compute demand by the end of the decade.

There are several semiconductor stocks, such as Intel and Broadcom, that you can consider buying to capitalize on the growing AI inference demand. Both companies are developing custom AI processors designed for AI inference. Additionally, Intel's server central processing units (CPUs) are being lapped up by hyperscalers to run inference workloads.

However, British technology company Arm Holdings (NASDAQ: ARM) could be one of the biggest beneficiaries of the shift toward AI inference. Arm has traditionally designed and licensed chip architecture to semiconductor companies and original equipment manufacturers (OEMs).

The company charges customers a licensing fee and receives a royalty on the sale of each chip designed using its intellectual property (IP). Arm has a solid customer base, with the likes of Apple, Nvidia, Amazon, and others utilizing its IP to design and manufacture chips for smartphones, personal computers (PCs), and data centers.

Arm's designs are in solid demand due to their energy efficiency. This explains why flagship smartphones and custom AI processors that prioritize high energy efficiency and performance-per-watt in large-scale applications such as data centers are designed using its IP. So, it is easy to see why the likes of Google, Nvidia, Amazon, Microsoft, and Marvell Technology are designing their AI data center CPUs using Arm technology.

Counterpoint Research predicts that Arm is on track to dominate the server CPU and custom AI processor markets over the next three years, cornering 90% of this space. Importantly, Arm is now going beyond just licensing its architecture and getting royalties. The company announced in March that it is now developing its own chip -- the Arm AGI CPU -- to power agentic AI and inference workloads.

The good news for Arm investors is that its foray into manufacturing its own silicon seems to be paying off. As pointed out by Arm CEO Rene Haas on the company's latest earnings call:

Customer response to the Arm AGI CPU has been very strong. We now have more than $2 billion of customer demand across fiscal 2027 and fiscal 2028. This is more than double what we stated at launch.

That's quite impressive, considering that the AGI CPU was announced on March 24 this year