Arm wants half of the data centre market. Will AI help it achieve this

Arm, dotąd kojarzony głównie z rynkiem mobilnym, coraz śmielej wkracza do centrów danych. Napędzany boomem na sztuczną inteligencję i rosnącym znaczeniem energooszczędności, brytyjski gracz celuje w zdobycie połowy globalnego rynku procesorów serwerowych jeszcze w tym roku.

Klaudia Ciesielska
Source: Freepik

Until a few years ago, Arm was mainly associated with energy-efficient chips for smartphones. Today, the British company is taking increasingly bold steps into territory dominated by Intel and AMD – in data centres. If its predictions come true, it could control up to half of the market for processors used in server infrastructure by the end of 2025. It’s a leap that hasn’t happened without a reason – and artificial intelligence plays a key role.

In the era of LLMs and increasing demand for computing power, energy efficiency issues are becoming increasingly important. AI servers are devouring electricity on an unprecedented scale, and Arm has a strong bargaining chip in this context – its architecture is known for its lower power consumption compared to x86 chips. This makes it attractive not only for hyperscalers, but also for companies building AI infrastructure from scratch.

Although Arm does not manufacture its own chips, its IP licensing business model works exceptionally well where scale matters. And data centres are just that case – customers using Arm’s architecture often implement more of the company’s proprietary solutions, resulting in higher licence and royalty revenues.

This can be seen in the examples of the major players. Amazon is designing its own Graviton chips, whose cores are based on Arm technology – these now account for more than half of the company’s added computing power in data centres over the past two years. Google and Microsoft are also designing their own Arm chips, although their designs are still at an earlier stage.

Ad imageAd image

Arm also benefits from the support of the software ecosystem. Just a decade ago, migrating to an alternative architecture meant rebuilding code and infrastructure. Today, more and more cloud-native applications and AI tools are being written with Arm in mind right from the start.

Despite the dominance of x86 chips to date, change is happening faster than you might have thought. The rise of AI has become a catalyst that is forcing a new approach to data centre design. In this environment, Arm is not so much chasing the competition as it is starting to set the direction, especially where not only watts and gigahertz, but also flexibility and scale are becoming crucial.

Will Arm really capture 50% of the data centre market by the end of the year? This is an ambitious target. But if current trends continue, Arm’s share of this transformation will be hard to ignore – especially for channel partners and integrators, who should already be watching this shift closely.

Udostępnij