Skip to content
World clockEU--:--UK--:--USA--:--CN--:--PLDEFRIT中文EN

portal about AI and technologyevents · analysis · interviews · technical background

Search
LIVE
›

Sovereign AI: Cohere and Aleph Alpha seal their $20 billion merger

Canada's Cohere and Germany's Aleph Alpha signed a definitive agreement on 16 September, after announcing the plan in April. The new group promises transatlantic "sovereign" AI, but the balance of power between the two partners raises questions.

TechnologyNewsRachel NwosuPublished: 23 September 20265 min readSources 2
Sovereign AI: Cohere and Aleph Alpha seal their $20 billion merger

The tie-up between Canada's Cohere and Germany's Aleph Alpha is no longer a project. The two enterprise AI specialists signed a definitive agreement on 16 September, after first announcing the plan on 24 April 2026, Le Monde Informatique reports. The new group will keep the Cohere name and run headquarters in Toronto and Berlin.

A valuation announced at around $20 billion

The project surfaced in the spring as a strategic combination built to meet growing demand for "sovereign" AI among companies and public administrations. The deal values the new group at around $20 billion. When the plan was first announced, Cohere's shareholders were said to be taking roughly 90% of the new entity.

The process is not entirely complete. The agreement still needs final regulatory approvals, and completion is expected during 2026. At closing, Ilhan Scheer, currently co-head of Aleph Alpha, will become chief operating officer of Cohere, while Samuel Weinbach will take over as head of research.

A balance heavily tilted toward the Canadian side

That split raises the question of the economic balance of power between two companies nonetheless presented as "counterparts." Aleph Alpha had abandoned development of frontier models to refocus on integration software, and its co-founder left the company after handing over the leadership in 2025. The imbalance explains why some observers describe the deal as Cohere acquiring Aleph Alpha while "dressed up" as a merger.

Cohere was founded in Toronto in 2019. It develops large language models and AI solutions mainly for companies and public administrations, with the emphasis on deployments inside the customer's infrastructure, on security and on data confidentiality. Aleph Alpha, for its part, positioned itself in professional and public uses of AI, gradually strengthening its focus on integration and specialized models for governments and regulated sectors.

More than 1,000 employees and a dual base

The merger is to create a company with more than 1,000 employees and a dual Canadian and German base. The Heidelberg site will be kept as a research centre. The stakes therefore go beyond industrial consolidation alone: the two partners want to offer systems able to run under the control of organizations, outside the big American platforms.

The sales pitch is no longer only the performance of the models, but the ability to run them under the control of the organization using them.

This sovereignty logic runs through the whole European industry. The same movement is pushing administrations to move away from American office suites, and hosting providers to offer local alternatives, as shown by the Dutch decision to migrate to a free operating system.

Comments 0

Sources

2
  1. 01Le Monde Informatique : Cohere et Aleph Alpha, une fusion pour une IA souveraine transatlantiqueFR
  2. 02CorCom : Sovranità, le regole non bastanoIT

All figures and quotations in this text come from the sources listed below.

Content prepared by the editorial team with AI assistance.

Rachel Nwosu

Rachel Nwosu

AI, models and technology

Rachel Nwosu covers AI, models and technology for FLASH24, working from public model documentation, benchmark releases and repository histories rather than press summaries, and she skips announcements that arrive without reproducible numbers. She checks training-data claims against dataset cards and reruns reported metrics where code is available. She spends much of her week interviewing researchers and engineers, tracking model launch calendars, and comparing vendor benchmarks with independent evaluations. Outside the desk she runs 3D printers, restores old computers, and tests how models learn from internet junk. She does not publish benchmark figures she cannot trace to a source.

Newsroom →

Comments

0
  1. No comments yet — be the first.

Write a comment

Comments are public. We do not publish abuse, spam or advertising.