Can data collaboration save AI?

Written by:

Fouad Talaouit-Mockli

Charlotte Ledoux

Reading duration:8 min


When we were consultants, we were regularly asked by our clients to advise them on the best strategy to evolve and remain competitive.

There are two possible strategies. The first is obviously to develop internal innovation capacities, whether in the form of investment in R&D, new product launches, service improvements, etc. This means that companies must invent and improve new products or services themselves in order to reap the benefits.

The second strategy consists of developing partnerships and cooperation strategies to seek complementary resources and skills. This approach is less natural for the majority of companies, except for those operating in sectors where technological intensity is high (automotive, aerospace, electronics, hardware, software, etc.) because they require a wide range of skills.

Today, the shortening of the product life cycle and competition on globalized markets are pushing companies to cooperate to accelerate their time-to-market and remain competitive. Cooperation allows savings to be made by pooling efforts on projects that are too large or too risky for a single company to manage. And this strategy, which for a long time was reserved for large groups, is becoming more and more widespread (1).

There are two different cooperation strategies:

  • the vertical cooperation strategy, with customers and suppliers,
  • the horizontal cooperation strategy, with competitors. The latter strategy is called coopetition.

Coopetition, a history of compromise

The notion of coopetition was popularised by Nalebuff and Brandenburger in 1996 in their book “La Co-opétition, une révolution dans la manière de jouer concurrence et coopération” (2).

Coopetition is a collaboration or cooperation of circumstance or opportunity between different economic actors who are otherwise competitors.

A company thus shares certain resources with some of its competitors, with the aim of developing a new complex product, conquering a new market, or creating longer-term synergies. (3)

However, coopetition on data subjects raises strategic questions. It goes against the protectionist strategies of companies regarding their data assets. Indeed, companies expose themselves to the risk of transferring knowledge to a competitor by making key assets available to it: its data. This competitor could then imitate some of the company's strategic resources and jeopardise its competitive advantage.

Even more than in vertical cooperation, coopetition requires teams with individuals who are open to the cooperation mentality. This is not always obvious as the emotional impact is not negligible, we tend to think in exclusive terms. Competing and cooperating requires careful consideration of the risks and benefits, but above all it requires building trust both internally and with competitors.

In coopetition strategies, the challenge is to find the best compromise between sharing resources and protecting assets.

New technologies obviously have a role to play in guaranteeing the confidentiality of exchanges as well as the confidentiality of the data handled. But is this enough to create trust?

Once the objectives and perimeter of the cooperation have been set, where and how to work?

In the field, companies already share a certain amount of information with their partners: exchange of files by e-mail or FTP, sometimes without going as far as the evaluation of the risks according to the associated stakes. These peer-to-peer exchanges are not always traced and can be a source of failure or a vector of attack (as was the case for Doctolib last year, which was the victim of a theft of data on 6,000 appointments made on certain connected third-party software (4)).

The benefit/risk ratio of a collaboration

Collaboration allows several parties to share access to certain resources to implement partnerships, alliances, marketing efforts and strategic initiatives. A true ecosystem is created between several organizations.

But before taking this step, the potential benefits of collaboration must be assessed (5). It may be a question of developing an innovation with your partner that will enable you to reach a new market together. Or reducing the time-to-market of your churn use case thanks to direct access to quality data (information on your end customers from your distributors for example). You can also claim to have more efficient R&D by pooling your resources with those of a university or laboratory. The idea is above all to estimate these gains, and not only quantitatively (it is very difficult to predict the impact on your market share, for example, but collaboration can be key to staying on the market).

As for the risks, we can for example name some unavoidable dimensions:

  • Market: how will your market share be impacted if you decide not to cooperate? What are the alternatives for your partner or competitor?
  • Governance of the collaboration: which organizational model? Which governance bodies? What audit / monitoring of the work carried out?
  • Legal: what legislation is in force? What contract governs the agreement between partners, including the sharing of intellectual property and the sharing of profits?
  • Workspace: what collaboration space? What associated security? What levels of access and roles? What reversibility? What assets will be shared?

To start evaluating the benefit/risk ratio, a very simple tool can be used: the SWOT (Strengths/Weaknesses/Opportunities/Threats). This already gives a first 360° view of the potential collaboration. Let's take the case of Ford and Volkswagen who have jointly invested in Argo AI, a Ford start-up dedicated to autonomous cars. As Ford is well placed in the US and Volkswagen in Europe, this collaboration is unlikely to threaten their existing market shares (6). Rather, it gives them the opportunity to attack a new market together: here the ratio was deemed sufficiently positive.

How to structure the agreement/alliance

When drafting an agreement, it is also necessary to consider the balance of power inherent in the collaboration: depending on the type of actor, their size and available resources. Indeed, a technological dependence between the actors risks unbalancing the relationship. But if the agreement is structured correctly, a successful collaboration is possible: one can think for example of the surprising 2019 partnership between Samsung and Apple, which allows Samsung users to buy and rent TV shows and movies through Apple (8).

A successful collaboration is based on solid foundations, the following points should be considered very carefully:

  • The recommendations of the Competition Authority, to ensure compliance;
  • The scope of the agreement, to formalize the scope and control its application;
  • The distribution of costs and benefits, to ensure that everyone benefits;
  • The mode of collaboration, as a means of balancing forces.

Check what the Competition Authority recommends

There is a guide that gives the principles and limits not to be crossed for SMEs (7). You can also approach the Competition Authority or contact the bodies that govern your sector of activity. In the event of coopetition, do not hesitate to seek the assistance of a lawyer specializing in competition law.

Complying will allow you to better manage potential risks and to be part of a truly ethical and responsible approach.

Defining the scope of the agreement and the monitoring procedures

It is essential to define the scope of the agreement, the roles, and responsibilities and how to exit the agreement if it is no longer relevant. It is important to structure the agreement in such a way that neither partner becomes dependent on the other.

There are simple cases where the agreement is limited and easy to put in place. For example, where one partner becomes a non-core supplier to the other with the aim of sharing costs but not know-how. Or where one partner takes responsibility for the collaboration and the other is protected by a contract that includes performance guarantees and penalties if targets are not met (provided you can establish these thresholds!).

Sharing the gains

Collaboration must necessarily create value for all participants. And the distribution of gains should be fair as far as possible. This becomes more difficult when the cooperation is unbalanced or when one partner is already paying the other for certain resources. The difficulty is even greater when the agreement includes three or more partners and the services exchanged do not offset each other.

A simple solution may be to allocate costs/benefits proportionally according to sales volumes.

Identify the mode of collaboration


The workspace of a collaboration is an element that seems trivial but is in fact a central choice in the balance of the relationship. If this space is on the premises of one of the partners (subspace of its data hub for example), this will necessarily be a source of tension: the one who owns the collaborative space will be able to manage access to it more and potentially benefit more from what is produced there.

For example, we often see imbalances in partnerships between large groups and startups, with the latter often being "sucked in" by the large group, which has more technical, legal and financial resources to offer a collaborative space, and ultimately to buy the startup's assets.

Trusted collaborative spaces should make it possible to solve this problem, by being a neutral and secure place that guarantees exchanges between partners and their traceability. It is necessary that technologies emerge and make possible the growing opportunities for collaboration between organizations.

Open your mind before you open your systems

Alliances between partners or competitors are a growing phenomenon. Nevertheless, every collaborative agreement has a strong emotional dimension. It takes a step forward to open up and accept that there are indeed multiple winners in your market. Managers must be open to this duality between cooperation and competition. Here it is a question of doing both instead of thinking exclusively.

Between these two opposing perspectives, several studies argue that companies are better off seeking the benefits of both coopetition and cooperation. Coopetition can enable new revenues. The advantages of cooperation are access to scarce and complementary resources.


The data and AI maturity of companies is a reality, the 2021 study by Mckinsey (9) reports that 56% of all respondents say they have adopted AI in at least one function of their company. However, skills are scarce and difficult to recruit and retain. An organisation may not be able to possess the full spectrum of skills, resources and data needed to innovate on its own.

According to IBM's Global AI Adoption Index 2021 (10), 90% of IT professionals say that the ability to implement their AI projects regardless of where the data is located is key to the adoption of this technology. When 33% pointed to a lack of expertise or knowledge as a barrier to AI adoption and 29% identified the increasing complexity of data and the existence of data silos as a barrier.

Cross-company collaboration is one response to this labour shortage. This strategy allows companies that lack expertise to access new opportunities. In addition, there are two other interesting effects: the mutualisation of project costs and the carbon footprint of AI models. For a Machine Learning model with 100 hours of training and inference, it is a 33% reduction in CO2 emissions when 2 organisations carry out this calculation together.

The collaboration of data, human and economic resources through a neutral and secure platform that guarantees exchanges between partners and their traceability is the key to the emergence of a collective intelligence necessary to solve the challenges of tomorrow. Inter-company collaboration will be an essential lever to remain competitive.

After defining the collaboration strategy (agreement, model, governance, etc.), the technological aspect remains to be grasped. Gaia-X was intended to support this dynamic: in our opinion, this project is a failure today, having failed to federate and concretize the praiseworthy principles initially defined.

At Vallai, we are developing a technological platform to meet these challenges and enable organizations to fully seize the opportunities of data collaboration.