Arm’s share price wobble ends as reports of new AI division triggers surge

Arm has published its results for January to March 2024 but, despite better-than-expected revenue and profits, the Nasdaq verdict was one of disappointment and the Cambridge chip designer’s share price dipped 7.65 per cent.

Why? Expectations for the company were sky-high following its IPO last September. Its shares have more than doubled since its launch: the stock is now pricier than any AI-related stock in the Nasdaq 100. Was a backlash inevitable? Perhaps - the markets are always in thrall to knee-jerk reactions - but Arm’s actual performance remains hugely impressive.

Revenue for the quarter was $928million (against $880.4m expected). This represents a year-on-year hike of 47 per cent.

Gross profit, at $887m (vs $840m expected), was driven by two factors. Firstly, quarterly royalty revenue was $51m, up 37 per cent year-on-year, thanks to the success of its high-royalty Armv9-based chips in a reinvigorated semiconductor industry. Second, license revenue was $414m, up 60 per cent year-over-year, as Arm’s technology powers the fast-moving AI sector.

An Arm spokesperson said: “We believe these fundamental trends will continue and we expect next quarter to also deliver strong year-over-year growth in revenue and profits.”

The share price stood at $107 before the fiscal announcement, dropped to a low of $95.9 at the close of May 8, and recovered to $117 by Monday, stabilising at $116.5 on Wednesday (15 May). The hike was almost certainly the result of a report by Nikkei Asia on Monday (13 May) announcing Arm’s plan to develop an AI chips division - Nikkei Asia reported the first products are due in 2025.

All eyes now on Arm’s potential for AI data centres and edge computing.

Arm CEO Rene Haas said: “We finished our financial year achieving over $3bn in revenue for the first time, and with strong tailwinds heading into FYE25 as AI is driving increased demand for Arm-based technology across all end markets. From cloud to edge, all AI software models, from GPT to Llama, rely and run on the Arm compute platform.

“As these models become larger and smarter, their requirements for more compute with greater power efficiency can only be realised through Arm.”