Archived

Ampere Computing was acquired by SoftBank — this page is preserved as an archive. Visit fobi.com for current frontier-tech coverage.

Ampere Computing

Acquired by SoftBank · 2025 · $6.5B

Ampere designs Arm-based server CPUs optimized for cloud-native workloads, competing with x86 in data centers. Founded by ex-Intel president Renée James; raised $800M+ including from Oracle. AmpereOne with up to 192 cores shipping. Acquired by SoftBank for $6.5B, closed November 2025.

Ampere's Altra processors power Oracle Cloud and Microsoft Azure, proving Arm's viability for hyperscale AI infrastructure.

amperecomputing.com AI AcceleratorsUSAacquired

At a glance

Total raised

$40M

last round acquired

Headcount

—

no public estimate on record

Headquarters

Santa Clara

USA

Founded

2017

9 years active

Status

Acquired

by SoftBank

Latest valuation

$6.5B

Acquired · 2025

For subscribers

Subscribe to see the full profile

Including Funding rounds, Bull / Bear thesis, Stock + earnings, Roster changes, Patents, News, and Open roles.

Founding

50% off

/month

91 of 100 spots left

Full

$10

/month

Available once all 100 Founding Member spots are claimed.

Get full access

Already subscribed? Sign in →

Funding history

Total raised

$40M

Rounds tracked

Latest stage

Acquired

Latest valuation

$6.5B

Date	Stage	Amount	Valuation	Lead investors
Mar 19, 2025	Acquired	$6.5B	$6.5B	SoftBank
Sep 27, 2019	Growth	$40M	—	Oracle

How they make money

Hardware (one-time)Pattern

Ampere runs a fabless semiconductor hardware model: it designs server CPUs and sells them to cloud providers and data-center operators as one-time chip purchases. Its AmpereOne processors, built on the Arm architecture and scaling up to 192 cores, are bought by hyperscale customers such as Oracle Cloud and Microsoft Azure, which integrate the chips into their server fleets and resell compute capacity to their own end users. Revenue is recognized at the point of shipment or delivery to the customer, typical for hardware sales. The economic structure follows the fabless-semiconductor pattern: Ampere handles architecture and design in-house but outsources fabrication to third-party foundries, keeping capital expenditure off its balance sheet. Gross margin depends on the gap between the selling price per chip and the combined cost of design amortization, wafer production, packaging, and testing. What scales is the design — once a chip is taped out, incremental units carry relatively low marginal cost until the next design cycle begins. Research and development is the primary recurring expense, and customer concentration risk is high because hyperscale buyers negotiate aggressively on volume pricing.

Generated by deepseek-v4-flash · June 2, 2026Analytical summary · not investment advice

Founder backgrounds

Renée James

Founder, Chairman & CEO

Holds a B.S. in international business and an M.B.A., both from the University of Oregon. Joined Intel in 1988 as a product manager for 386-family motherboards and early in her tenure served as chief of staff to CEO Andy Grove. Rose through Intel's software, services, and manufacturing organizations, becoming Executive VP and General Manager of the Software and Services Group before being named President of Intel in May 2013 — a role she held until leaving in 2016. She founded Ampere Computing in October 2017, backed by The Carlyle Group, building Arm-based server processors optimized for cloud-native workloads.

AI bull / bear

Bull case

Ampere is the strongest Arm-based server CPU contender for the AI era, and the $6.5B SoftBank acquisition validates that thesis at a premium. With AmpereOne shipping at 192 cores and Altra already deployed across Oracle Cloud and Microsoft Azure, the company has proven hyperscale readiness for cloud-native and AI inference workloads. Founder Renée James, a former Intel president, brings deep x86 insider knowledge that Ampere weaponizes against incumbents. Arm's architectural efficiency gives Ampere a power-per-watt advantage that directly translates to lower total cost of ownership for hyperscalers running massive inference clusters. SoftBank's backing provides patient capital and a strategic bridge to Arm Holdings itself, potentially unlocking tighter chiplet integration or ISA-level optimizations that pure x86 players cannot match. As AI inference shifts from training dominance to deployment at scale, Ampere's high-core-count, power-efficient design is precisely positioned to capture workloads that GPUs handle inefficiently.

Bear case

Ampere enters the SoftBank orbit already behind. The $6.5B acquisition price, while significant, reflects a company that failed to IPO and needed a buyer despite $800M+ in funding and marquee customers like Oracle and Microsoft. Ampere's core thesis — that Arm efficiency would displace x86 in the cloud — has not materialized at scale. AMD and Intel now ship their own high-core-count, power-efficient server chips (Bergamo, Sierra Forest) that neutralize the architectural advantage without requiring workload migration. Meanwhile, NVIDIA dominates AI infrastructure with CUDA lock-in, and custom silicon from AWS (Graviton), Google (Axion), and Microsoft (Cobalt) threatens to squeeze merchant silicon vendors entirely. Ampere's reliance on TSMC for manufacturing leaves it exposed to capacity constraints and pricing pressure. SoftBank may attempt integration with Arm, but that risks alienating the very hyperscalers that Ampere needs as customers — no cloud provider wants to buy CPUs from a competitor's sibling.

Generated by deepseek-v4-flash · June 2, 2026Refreshes quarterly · see disclosures

In the news

Ampere Computing · last 30 days

No articles ingested yet for Ampere Computing. Once the hourly news pipeline is live, every article the classifier tags as mentioning this company appears here with its one-line AI summary and sentiment.

Watch

Leadership

Renée James

Founder, Chairman & CEO

—

Products

AmpereOne / AmpereOne M
Ampere's flagship line of cloud-native server CPUs built on custom Arm-compatible cores, scaling to 192 single-threaded cores per socket. Designed for predictable performance, high core density, and energy efficiency in scale-out cloud and AI-inference workloads, AmpereOne avoids the noisy-neighbor effects of SMT. The AmpereOne M variant adds wider memory channels. It is Ampere's primary product driving its data-center CPU strategy.
Altra / Altra Max
Ampere's first-generation Arm Neoverse-based server processors, offering 32–128 cores aimed at cloud-native and scale-out workloads. Deployed by Oracle Cloud Infrastructure, Microsoft Azure, Google Cloud, and others, Altra established Ampere's credibility for hyperscale Arm computing with strong performance-per-watt. It remains widely available in public clouds and seeded the market ahead of the custom-core AmpereOne generation.
Ampere AI / AIO (AI Inference)
A software and platform effort that runs AI inference efficiently on Ampere CPUs—using the Ampere Optimized AI (AIO) inference engine and framework integrations—so customers can serve models on general-purpose Arm cores without dedicated GPUs for many workloads. Combined with partnerships (including work tied to SoftBank's AI ambitions), it positions Ampere CPUs as a power-efficient option for cost-sensitive, CPU-based AI inference.

Most recent patents

32 patents on file, but none with both an extractable figure and an abstract on Google Patents yet.