Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsSoftBank completed its $6.5 billion all-cash acquisition of Ampere Computing on November 25, 2025. Ampere is a Santa Clara-based semiconductor-design company that makes Arm-based server processors; it is now a wholly owned SoftBank subsidiary. The deal gives SoftBank direct control of a CPU designer, but does not by itself establish that the company has launched a new AI accelerator or displaced GPU platforms.
What SoftBank bought—and when
SoftBank agreed to acquire 100% of Ampere Computing Holdings LLC for $6.5 billion in cash. Its board approved the transaction on March 17, 2025, and the agreement was signed on March 19 U.S. time. SoftBank announced the terms on March 20 in Japan. The acquisition closed on November 25, 2025, U.S. time, through SoftBank subsidiary Silver Bands 6 (US) Corp. Ampere is now an indirect, wholly owned subsidiary of SoftBank Group. SoftBank’s transaction announcement and its closing announcement document the agreement and completion.
Ampere retained its name and Santa Clara headquarters under the announced transaction. It is a chip designer, not a semiconductor foundry that fabricates chips. SoftBank’s filing dates Ampere’s founding to September 27, 2017; Ampere’s deal announcement described it as founded in Silicon Valley in 2018, so those public materials differ on the founding date.
What Ampere makes
Ampere designs processors using Arm technology, with a focus on cloud-native computing, servers and data centers. Its CPUs are intended for general-purpose processing and workloads where performance per watt matters. In AI infrastructure, that can include CPU-side inference, orchestration and smaller or distributed workloads, as well as working alongside GPUs in hybrid systems.
That distinction matters: Ampere is not simply another name for a GPU maker. Some AI training and high-throughput workloads rely heavily on specialized accelerators. Arm supplies processor architecture and related intellectual property; Ampere uses Arm technology to design complete server CPUs. SoftBank described Ampere as having about 1,000 semiconductor engineers when it announced the deal. Ampere’s announcement and SoftBank’s filing set out the companies’ stated roles and rationale.
Why SoftBank wanted the company
The strategic logic is to add processor-design expertise to a portfolio that already includes control of Arm. SoftBank said Ampere’s chip-design and tape-out capabilities could complement Arm’s design capabilities. In principle, that gives SoftBank influence across more of the path from processor architecture to a finished server CPU. It does not mean Arm and Ampere have been merged or that SoftBank has announced a single combined chip operation.
SoftBank framed Ampere as part of a wider effort to build AI infrastructure and advanced-computing capacity, alongside initiatives and investments involving Arm, Cristal intelligence and Stargate. The broader interpretation is that SoftBank is assembling assets across an AI stack—from compute technology to infrastructure and software—rather than making an isolated semiconductor bet. That is an analytical reading of the portfolio, not a disclosed operating plan or proof that Ampere processors will be deployed in Stargate.
Rank #2
- Computer Hardware Technology design. Computer processor design, great for IT computer technicians, software engineers, or any engineer that deals with microprocessors. This funny computer scientist shows a CPU or circuit board.
- CPU Electronic Chip Circuit Board Gift. Ideal for computer science students, software developers, administrators and all who like to work with computers.
- Lightweight, Classic fit, Double-needle sleeve and bottom hem
Power efficiency is another part of the rationale. Data centers need power and cooling as they scale, and CPUs can be appropriate for selected inference or general cloud tasks that do not require a GPU for every operation. Whether that makes a particular AI service cheaper depends on the whole system: software, utilization, memory, networking and the mix of CPUs and accelerators. SoftBank executives have also identified execution and power availability as risks in accelerated computing. Their earnings-call comments do not support treating the acquisition as a guaranteed route to lower AI costs.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
The price against Ampere’s financial record
SoftBank’s filing reported sharply lower revenue and substantial losses in the fiscal years shown. Figures below are in U.S. dollars; losses are shown in parentheses.
| Fiscal year | Revenue | Operating profit/(loss) | Net profit/(loss) |
|---|---|---|---|
| 2022 | $151.822 million | $(518.290) million | $(513.815) million |
| 2023 | $46.704 million | $(714.689) million | $(830.848) million |
| 2024 | $16.460 million | $(510.623) million | $(580.767) million |
For 2024, SoftBank also reported net assets of negative $1.513 billion. The $6.5 billion purchase price was therefore not a straightforward bet on an established stream of profits. It implies a strategic value assigned to technology, engineering talent, intellectual property and the chance to build a stronger position in future computing markets.
Dividing the purchase price by 2024 reported revenue gives roughly 395 times revenue. That arithmetic is not a conventional valuation multiple for judging the deal: the revenue base was unusually small, Ampere was loss-making, and the strategic case depends on future products and adoption rather than that year’s sales. The purchase price should also be distinguished from financing: SoftBank’s FY2025 financial report disclosed a bridge-loan commitment of up to $6.5 billion, arranged by Mizuho Bank, Sumitomo Mitsui Banking Corporation and JPMorgan Chase Bank, Tokyo Branch, to fund the acquisition, repay Ampere borrowings and cover related expenses. The commitment is not evidence that the full amount was drawn or a statement of the final financing cost. SoftBank’s FY2025 financial report describes the facility.
How Ampere was already connected to SoftBank
Before the acquisition, Ampere was not wholly separate from SoftBank’s orbit. SoftBank’s transaction filing listed voting interests of 59.65% for Carlyle Partners VI Denver Holdings, 32.27% for Oracle Project Denver Holdings and 8.08% for Arm Technology Investment. Arm’s stake matters because SoftBank is Arm’s majority shareholder and Ampere was an Arm licensee. The transaction converted that indirect connection into direct ownership of Ampere; SoftBank did not acquire Arm as part of this deal. The ownership figures are from SoftBank’s filing.
What has happened since the acquisition closed
A concrete post-closing development came in February 2026, when SoftBank Corp. and Ampere announced a joint validation of CPU-based AI inference. Their evaluation covered small language models, mixture-of-experts models and AI-agent workloads, using CPU-only and CPU/GPU hybrid nodes, multi-node deployments, and dynamic model placement and orchestration. They also tested an Ampere-optimized version of llama.cpp.
SoftBank Corp. said the evaluation increased concurrent inference capacity while reducing power use relative to typical GPU-based configurations, and reduced model-loading times. Those are company-reported results, not independent benchmark findings. They show work on CPU inference, but do not establish that Ampere has displaced major GPU platforms or that SoftBank has launched a proprietary AI accelerator. The companies’ February 2026 announcement describes the evaluation.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.The main risks to the strategy
Commercial traction
The disclosed financials show that Ampere’s revenue fell steeply through 2024 while losses remained large. Sustained customer adoption, product shipments and a return to meaningful revenue growth are therefore central to whether the strategic purchase pays off. Technical capability alone does not establish commercial success.
Arm’s customer relationships
Arm’s licensing business serves companies that compete with one another. Common ownership of Arm and a chip designer could prompt questions about whether Ampere receives preferential access to roadmaps or other advantages, and whether Arm will remain neutral in customers’ eyes. That is a potential governance and ecosystem risk, not evidence that Arm has favored Ampere.
Free tools Windows power users keep installed
One-click scans. No signup required.
Software, manufacturing and execution
A competitive processor platform requires more than chip design. It also needs software and compiler support, customer qualification, manufacturing capacity, reference systems, memory and networking integration, and developer adoption. SoftBank’s executives have acknowledged execution challenges and power constraints; the acquisition alone does not resolve either.
Limits of CPU substitution
CPUs can be suitable for selected inference, orchestration and general-purpose workloads, but they are not universal substitutes for GPUs or other accelerators. A CPU/GPU mix may be more useful than an all-or-nothing comparison, and the right configuration depends on each workload’s performance and cost requirements.
What remains unproven
- Whether SoftBank or Arm will introduce a jointly branded system or a new chip platform.
- Whether Ampere will remain operationally independent or how its organization will change.
- Which cloud providers and systems companies will deploy Ampere processors at scale.
- Whether the business can restore meaningful revenue growth and improve profitability.
- How SoftBank will report Ampere-specific financial performance.
- Whether Ampere’s role will remain centered on CPUs or expand into custom AI accelerators.
SoftBank has not publicly established a binding Stargate procurement agreement for Ampere processors or a commercial SoftBank-designed AI chip. Its purchase gives it a strategically relevant Arm-based CPU designer and a platform for further AI-inference work; the evidence so far is not proof of a complete, competitive AI-chip business.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
