JP/
EN
CONTACT
Awaken and Connect Your Data.

Open Data Spaces

INFORMATION
2026.04.07
IPA exhibits at HANNOVER MESSE 2026.
Coming Soon concept movie

The Dawn of Data Scarcity.@Progress in AI is hitting a wall.

The Dawn of Data @Scarcity.@Progress in AI is @hitting a wall.

The next breakthrough won't come from the web
—it lies in your operations: the "real data" your business generates every day.
Are you locking your data away in a vault, hiding behind "confidentiality"?
Isolated from the outside world, even the most valuable data eventually sinks into silence.
You think you are protecting it. In reality, you are letting it slowly decay.

The key is making your data “AI-ready”.

The key is making @your data “AI-ready”.

And building a foundation that connects only with those you trust.
Don’t dump it into a massive silo where context is lost.
Never surrender your control to others.

The answer is Open Data Spaces.@Contextualize. Then, unleash.

The answer is Open @Data Spaces.@Contextualize. Then, @unleash.

By enabling distributed and trusted data sharing,
Knowledge in organizations begins to create new value across boundaries.
Protect your own data. Then join forces as equals with selected partners.
Only then does data transform into a true asset.

Keep it isolated, or connect to create new value?
The key to evolving together in the global market.
Open Data Spaces.

Open Data Spaces (ODS) is an open and scalable foundation for distributed data, built on organizational and national diversity by design.

What is Open Data Spaces (ODS)?

What is Open @Data Spaces @(ODS)?

Open Data Spaces (ODS) provides a practical, implementable distributed data management architecture designed for the Agentic AI era. At its core are standardized protocols and an open-source technology stack.

ODS is an open and scalable foundation for distributed data management across enterprises and industries, built on organizational and national diversity by design.

feature 01
Vendor-Agnostic

No lock-in to any specific cloud, platform, or vendor.

feature 02
Institution-Agnostic

Not a closed paradigm optimized for any specific jurisdiction's regulations, but adaptable globally.

feature 03
Product-like,
Service-
Oriented
Design

Always asking the market one critical question: does this contribute to Make Money or Save Money?

Design Philosophy

Design @Philosophy

Projected growth of text data used for LLM training from 2022 to 2034 (in tokens). The amount increases from approximately 10¹¹–10¹² tokens in 2022 to around 10¹⁴ tokens by 2026, and reaches roughly 10¹⁴–10¹⁵ tokens by 2030–2034. In comparison, the total amount of publicly available human-generated text on the internet is estimated at around 10¹⁴–10¹⁵ tokens over the same period. Shaded areas indicate uncertainty ranges around the projections.
Projected growth of text data used for LLM training from 2022 to 2034 (in tokens). The amount increases from approximately 10¹¹–10¹² tokens in 2022 to around 10¹⁴ tokens by 2026, and reaches roughly 10¹⁴–10¹⁵ tokens by 2030–2034. In comparison, the total amount of publicly available human-generated text on the internet is estimated at around 10¹⁴–10¹⁵ tokens over the same period. Shaded areas indicate uncertainty ranges around the projections.
Growth of global annual data volume from 2010 to 2025 (in zettabytes, ZB), including data created, captured, copied, and consumed. The total volume increases from about 5 ZB in 2010 to around 16 ZB in 2015, 53 ZB in 2020, 86 ZB in 2022, 109 ZB in 2023, 138 ZB in 2024, and is projected to reach approximately 175 ZB by 2025. In 2025, enterprise data accounts for about 104 ZB (60%), while consumer data accounts for about 71 ZB (40%). An overall growth rate of approximately 27% per year is indicated.
Growth of global annual data volume from 2010 to 2025 (in zettabytes, ZB), including data created, captured, copied, and consumed. The total volume increases from about 5 ZB in 2010 to around 16 ZB in 2015, 53 ZB in 2020, 86 ZB in 2022, 109 ZB in 2023, 138 ZB in 2024, and is projected to reach approximately 175 ZB by 2025. In 2025, enterprise data accounts for about 104 ZB (60%), while consumer data accounts for about 71 ZB (40%). An overall growth rate of approximately 27% per year is indicated.
01

Data is Eating
the World

Real Data and Context

It is projected that high-quality data supporting AI training could be exhausted between 2026 and 2032. This brings into focus the vast reserves of real data held -- and largely untapped -- within enterprises. The key to transforming that dark data into valuable corporate capital is giving it domain context: the background, meaning, and operational circumstances in which it was created. Productizing data enriched with frontline context is how organizations drive enterprise value in the Agentic AI era.

Diagram illustrating the shift in data architecture from centralized to decentralized models. On the left, a centralized architecture shows data aggregated into a central big data platform (such as a data warehouse or data lake) and used for AI/ML, BI, and other purposes. On the right, a decentralized architecture shows multiple domains managing and sharing their own data as products (“Data as a Product”), connected in a distributed network. The diagram highlights the transition from aggregation to distribution in data management approaches. Diagram illustrating the shift in data architecture from centralized to decentralized models. On the left, a centralized architecture shows data aggregated into a central big data platform (such as a data warehouse or data lake) and used for AI/ML, BI, and other purposes. On the right, a decentralized architecture shows multiple domains managing and sharing their own data as products (“Data as a Product”), connected in a distributed network. The diagram highlights the transition from aggregation to distribution in data management approaches.
02

From
Aggregation
To Distribution

From Aggregation to Distribution

Conventional centralized data management ― concentrating all data in one place ― has reached its operational and cost limits. Going forward, a distributed approach is required: one where individual domains manage and provide their own data.　This shift accelerates data utilization within and across departments. But seamless data management across organizational and national boundaries requires a more advanced paradigm.

Diagram illustrating the evolution from Data Mesh to Open Dataspaces. On the left, the Data Mesh model shows data distributed across multiple departments (such as sales, production, and distribution) within an organization, with users accessing data across these domains (inter-department). On the right, the Open Dataspaces model shows data sharing across different organizations, such as logistics companies, wholesalers, and manufacturers (inter-organization). The diagram highlights the shift from internal, department-level data sharing to cross-organizational data collaboration as a new paradigm. Diagram illustrating the evolution from Data Mesh to Open Dataspaces. On the left, the Data Mesh model shows data distributed across multiple departments (such as sales, production, and distribution) within an organization, with users accessing data across these domains (inter-department). On the right, the Open Dataspaces model shows data sharing across different organizations, such as logistics companies, wholesalers, and manufacturers (inter-organization). The diagram highlights the shift from internal, department-level data sharing to cross-organizational data collaboration as a new paradigm.
03

From Data Mesh
To Open
Dataspaces

A New Paradigm

Open Dataspaces inherits the architectural paradigm and four foundational principles of data mesh ― accepting the paradigm shift from the traditional “Push and Ingest” model to a “Serving and Pull” model, and adopting a data management approach based on Domain-Driven Design (DDD). The most significant difference from data mesh is that Open Dataspaces is a paradigm designed for inter-organizational data management ― going beyond the inter-departmental scope.

Conceptual diagram of the Double Product Quanta Model (DPQM). At the center is the “Architectural Quanta (AQ),” with an “Ontology Product” layer above and a “Data Product” layer below, forming a two-layer structure. On the left, the upper layer corresponds to the Open World Assumption (OWA) and eventual consistency, while the lower layer corresponds to the Closed World Assumption (CWA) and strong consistency. The diagram illustrates a model that balances flexible data acceptance with strict and reliable data usage. Conceptual diagram of the Double Product Quanta Model (DPQM). At the center is the “Architectural Quanta (AQ),” with an “Ontology Product” layer above and a “Data Product” layer below, forming a two-layer structure. On the left, the upper layer corresponds to the Open World Assumption (OWA) and eventual consistency, while the lower layer corresponds to the Closed World Assumption (CWA) and strong consistency. The diagram illustrates a model that balances flexible data acceptance with strict and reliable data usage.
04

Double Product
Quanta
Model（DPQM）

Balancing Openness and Rigidness

Managing data securely and flexibly across organizational and national boundaries requires a mechanism that treats data and domain context as inseparable partners.
The architecture employs a dual-layered approach: embracing diverse data through flexible rules that accommodate unknown information (Open World Assumption), while ensuring reliability at the point of use through strict validation (Closed World Assumption). This structure ― balancing openness with rigidness ― enables secure, scalable distributed data management.

VIEW DETAILS opens in a new window

ODS in Action: Proven Commercial Success
It is not a theoretical concept but in active commercial use. ODS has already been scaling across multiple pioneering projects, demonstrating its reliability in real-world environments.

field 01

Battery & Automotive

2024~ (operating & scaling)

field 02

Unmanned Aircraft Systems & Traffic Management

2025~ (operating & scaling)

field 03

Chemical & Circularity Management

2026~ (implementing)

Why ODS is swiftly adopted and penetrated in the market?

―No, market chooses our architecture.

The Three Pillars of Open Dataspaces’ Distributed Architecture

Solving the three Governance Complexity problems that arise in cross-organizational distributed data management.

Issue1
Where to Get
  • Where is the data in the first place?
  • Does this data refer to the same thing as that data?
Solution1
Data Addressability and
Discoverability（DAD）
Issue2
What to Mean
  • What does this data mean?
  • Is the meaning of this data consistent with that data?
Solution2
Ontology and Semantic
Interoperability（OSI）
Issue3
Who and How to Use
  • Who is trying to access the data?
  • Who can access this data?
  • How must this data be used?
Solution3
Identity and Usage
Control（IUC）
Why Agentic AI Needs ODS

The evolution of Agentic AI depends on its ability to understand "Context."

Open Dataspaces enables Agentic AI to move beyond statistical guesswork and into genuine knowledge-based inference.

From Inter-Department To
Inter-Organization

Open Dataspaces a.k.a The Open Data Space

While data mesh enables cross-departmental data management within a single organization, Open Dataspaces extends this to inter-organizational data management across corporate and national boundaries. For example, wholesalers, logistics providers, and manufacturers can each provide data "only to the partners they choose, only the portions they choose, and only in the manner they choose."

Dynamic Ontology

Mutual Complementation of LLM and ontology Dynamic Ontology Mutual Complementation of LLM and ontology Dynamic Ontology

By decoupling data "semantics" from its "structure", Open Dataspaces allows meaning to evolve without breaking the underlying systems. Even when definitions differ across organizations, LLMs infer semantic connections, which are then validated and refined through logical ontological rules. This continuous cycle ensures that data sharing never stops—even with incomplete datasets. Through this process, "ambiguous guesses" are transformed into "verifiable knowledge," maturing the ecosystem during actual operation.

The Concept behind the Logo

Open Data Spaces Open Data Spaces

The distinctive structure of DPQM (Data Product Quanta Model) is at the heart of our Visual Identity (VI).

Accessing ODS Artifacts
Why Open Dataspaces: Design Philosophy and the Architectural Paradigm
A document explaining the design philosophy and architectural paradigm of Open Dataspaces — a new technical paradigm for distributed data management across organizations, enterprises, and national borders.
opens in a new window
Open Data Spaces Protocols (ODP)
A set of technical specifications that provide the functions for realizing distributed data management based on ODS-RAM and ensure the interoperability of open dataspaces.
opens in a new window
Open Data Spaces Software Development Kit
SDKs and documentation for developers.
opens in a new window
ODS-RAM
A reference architecture for distributed data management across enterprises, industries, and national boundaries —documented through technology paradigms, hierarchical structure models, and protocol relationships.
opens in a new window
ODS Middleware
A suite of open-source software implementing the ODS Protocols (ODP) as a reference implementation.
opens in a new window
ODS Guidebook
Guidebook for Users and Engineers.
opens in a new window
Why Open Dataspaces: Design Philosophy and the Architectural Paradigm
A document explaining the design philosophy and architectural paradigm of Open Dataspaces — a new technical paradigm for distributed data management across organizations, enterprises, and national borders.
opens in a new window
ODS-RAM
A reference architecture for distributed data management across enterprises, industries, and national boundaries —documented through technology paradigms, hierarchical structure models, and protocol relationships.
opens in a new window
Open Data Spaces Protocols (ODP)
A set of technical specifications that provide the functions for realizing distributed data management based on ODS-RAM and ensure the interoperability of open dataspaces.
opens in a new window
ODS Middleware
A suite of open-source software implementing the ODS Protocols (ODP) as a reference implementation.
opens in a new window
Open Data Spaces Software Development Kit
SDKs and documentation for developers.
opens in a new window
ODS Guidebook
Guidebook for Users and Engineers.
opens in a new window

FAQ

What are the official brand notation rules for Open Data Spaces (ODS)?
  • Be sure to use the full notation, “Open Data Spaces (ODS)” the first time you mention it in any document or material.
  • After the first use, you may use “ODS” for later references in the same document or section.
  • Generic or lowercase forms, such as “open dataspaces” or references without “(ODS)” are not recognized as part of the ODS brand.
  • This rule does not apply to file names, logos, or other cases with specific requirements.
* “Open Dataspaces” on this website refers to a new-generation distributed data management approach and its constituent concepts. Its design draws on the original dataspace papers in the U.S. (Franklin et al., 2005; Halevy et al., 2006) and data mesh (Dehghani, 2019; Dehghani, 2022).
How should Open Data Spaces (ODS) be concisely described?
  • Please use the following brand definition (tagline):
    Open Data Spaces (ODS) is an open and scalable foundation for distributed data, built on organizational and national diversity by design.
  • The ODS brand embodies both the technology concept and the foundation, including all associated artifacts.
What are the recommended expressions to use when referring to ODS artifacts?

Recommended expressions for each artifact are as follows:

Artifact Recommended Expression
Why Open Data Spaces: Design Philosophy and the Architectural Paradigm Aligned with <Artifact>
Open Data Spaces Reference Architecture Model (ODS-RAM)*,
Open Data Spaces Guidebooks (ODS Guidebook)		 Refer to <Artifact>
Open Data Spaces Protocols (ODP),
Open Data Spaces Middleware (ODS Middleware)		 Implements <Artifact>
* including older version: “Whitepaper: Ouranos Ecosystem Dataspaces Reference Architecture Model (ODS-RAM V1)”
Can expressions such as “based on”, “conforms to”, “compliant with”, or “certified by” be used when referring to ODS artifacts?
Currently, there is no established certification system or conformity assessment process. Therefore, the use of these terms is not officially recognized or guaranteed. Please refrain from using these terms when referring to ODS artifacts. Any future updates on this policy will be announced on the official website.