The semiconductor market got a bit of a surprise this week as holding company Softbank announced its intent to acquire Ampere Computing for $6.5 billion. If approved, the deal is expected to close later in 2025.
The announcement, which follows Softbank's previous foray into AI chip startups with its acquisition of Graphcore, raises several questions about the strategic direction for Softbank and the broader implications for the semiconductor industry.
Ampere, founded in 2017 by Renée James, aims to develop commercially successful Arm-based server CPUs to rival Intel's Xeon and AMD's EPYC products. Since launching the Ampere Altra CPU in 2020 with Oracle Cloud Infrastructure being its first customer, the company has expanded its customer base to include tech giants such as Microsoft Azure and Google Cloud.
Despite initial successes in bringing Arm-based designs to a sizable audience, the competitive landscape has shifted. As cloud service providers (CSPs) have begun developing their own custom silicon, Ampere faces pressure in the market. Softbank's acquisition appears not only to bolster Ampere's position but indicates a more profound industry trend towards customized, in-house chip designs for performance efficiency.
Softbank’s acquisition may also be seen as a strategic move to assemble a powerful custom silicon portfolio. Currently, Ampere boasts about 1,000 design staff, while Graphcore employs around 500. By combining these two companies, Softbank aims to create a robust team capable of competing for significant market share in CPUs and accelerators.
This acquisition further raises speculation about whether it coincides with Softbank's involvement in the Stargate Project, a collaboration with Oracle. Industry analysts question if this acquisition is part of a preparatory move to tailor specific silicon solutions for emerging hyperscale environments. However, some argue that Oracle may be the more fitting acquirer for Ampere, given their existing relationship and shared interests.
With the increase in in-house silicon design by CSPs, a pressing question for Ampere is whether it can maintain relevance in a cloud market that increasingly favors customized solutions.
Despite this challenge, experts believe there is still potential for Ampere within the hyperscaler market. Although some major companies have turned towards in-house solutions, many remain interested in bespoke compute platforms but may lack the resources to develop these in-house. Thus, Softbank's strategic consolidation of Ampere could satisfy this void.
To succeed, Softbank must address an even bigger challenge: integrating different teams with varying corporate cultures while cultivating innovative designs that resonate with market demands. While acquiring technology is feasible, ensuring cohesive teamwork across various teams remains a significant hurdle.
The landscape of semiconductor development continues to evolve alongside cloud technology and artificial intelligence, which have drastically altered the dynamics of design, production, and deployment of CPUs. Softbank's venture into custom silicon with Ampere is significant, as new workloads necessitate tailor-made chips to achieve optimal performance.
Ultimately, as Softbank embarks on this acquisition journey, they manifest aspirations to establish themselves at the forefront of custom silicon innovation. Whether through the Stargate Project or other avenues, the blend of Ampere and Graphcore marks the inception of what could be an impactful change in high-performance computing.
