Open innovation in the age of AI

Two decades ago, Henry Chesbrough’s concept of open innovation reframed how companies thought about innovation. It challenged the idea that the best ideas came from within a single organization and showed how deliberate flows of knowledge – both into and out of the firm – could unlock value. Since then, open innovation has become mainstream (Holgersson et al., 2024).mCorporations large and small have partnered with startups, launched crowdsourcing campaigns, and tapped into global talent pools.

Now, artificial intelligence is rewriting that script. AI is no longer just another tool in the innovation manager’s toolkit. It is transforming the scope, speed, and even the necessity of traditional open innovation practices. In some areas, AI enhances what companies already do, making it faster and more precise. In others, it enables entirely new approaches that were previously unthinkable. And in certain cases, AI threatens to replace well-established practices altogether.

Enhancing the familiar

Much of open innovation has always depended on the ability to find the right ideas, partners, and resources. Traditionally, this search was constrained by time, geography, and human bandwidth. AI removes many of those constraints.

Consider the external search for innovation opportunities. Instead of relying on chance encounters at conferences or scanning a handful of industry publications, companies can now deploy AI systems to analyze millions of data points – from academic papers to online forums – to surface unmet needs and emerging trends. Reddit’s recent move to license its community data for AI training illustrates how vast, unstructured online discussions can be mined for insights. AI makes it possible to find promising ideas hidden in places that human analysts would never think to look.

Partner identification is undergoing a similar shift. Bill Joy’s maxim – that most of the smartest people work for someone else – is truer than ever, but now AI can map where those people are, what they know, and how they might complement your capabilities. Platforms like Monocl use machine learning to connect organizations with relevant scientific expertise worldwide, cutting months from traditional scouting processes. AI can identify potential connections. Humans must judge which to pursue.

The bottleneck of idea evaluation is beginning to loosen. Crowdsourcing platforms and innovation contests often generate thousands of submissions – far beyond what internal experts can assess in depth. Research shows that early-stage filters are far more effective at weeding out weak ideas than at reliably identifying winners (Dahlander et al., 2023). AI mirrors this pattern. It excels at screening large volumes of ideas to eliminate low-potential options, but human judgment remains critical for recognizing and nurturing the true outliers. This division of labor plays to the strengths of both: machines handle scale and efficiency, while humans provide contextual insight and foresight. The result is a more streamlined process and better allocation of scarce expert attention.

Finally, AI can improve access to specialized resources. In sectors where expensive, highly specialized equipment sits idle for much of the time, AI can help match users with facilities, schedule usage, and even forecast demand. This not only increases the return on these assets but also opens opportunities for smaller players who could never afford to own such equipment themselves.

Enabling the new

Some of AI’s most profound impacts come not from making old processes better, but from making entirely new ones possible.

AI is creating markets that didn’t exist before. In the music industry, for example, Italy-based IK Multimedia has developed AI tools that can “clone” the sound of rare guitar amplifiers. Musicians who own these amplifiers can now sell digital versions of them through the company’s TONEX platform, allowing others to use the sounds in recordings or live performances. Owners monetize assets that would otherwise gather dust, while buyers gain affordable, instant access. The platform’s success shows how AI can create new ecosystems where none existed before.

Business models are also evolving. Recorded Future, a company that tracks cybersecurity threats, uses AI to sift through open web and dark web data, transforming publicly available information into proprietary intelligence for its clients. This model blurs the line between open resources and private advantage – a hallmark of modern open innovation.

Federated learning offers another example of AI enabling new forms of collaboration. This approach allows multiple organizations to train a shared AI model without ever exchanging raw data. In healthcare, hospitals can build diagnostic models together while keeping patient data private. In finance, banks can collectively develop fraud detection systems without revealing sensitive customer information. Such arrangements solve one of open innovation’s oldest problems: how to share value without compromising competitive or regulatory constraints.

Replacing the old

Not all AI applications will coexist peacefully with traditional open innovation. In some cases, they threaten to replace it outright.

Idea generation is one such area. Large language models can now produce ideas that human evaluators rate as more creative than those from many professionals. This doesn’t eliminate the need for people in the innovation process – implementation, refinement, and context still matter enormously – but it does reduce the value of traditional idea-sourcing methods like suggestion boxes or open calls.

Synthetic data is another potential disruptor. By generating data that mirrors the statistical properties of real datasets, companies can simulate sensitive information without risking privacy or intellectual property. In autonomous vehicle development, for example, synthetic traffic scenarios can replace vast amounts of real-world driving data, accelerating development while avoiding the legal and logistical hurdles of data sharing. In such cases, the rationale for partnering to obtain real data weakens significantly.

Multi-agent systems – networks of autonomous AI entities that negotiate, coordinate, and adapt in real time – may reduce the need for traditional inter-firm coordination. Imagine supply chain optimization conducted not by teams of human managers but by AI agents representing suppliers, manufacturers, and logistics providers, continuously adjusting to shifts in demand and supply. This is not a distant vision; companies like Amazon are already experimenting with such systems.

A co-evolutionary future – and a leadership imperative

AI and open innovation will not simply coexist; they will redefine each other. AI will make open innovation faster, broader, and more data-driven. Open innovation will, in turn, feed AI diverse inputs and ensure its responsible development.

For leaders, the question is no longer whether to adapt, but how – and how quickly. Competitive advantage will accrue to those who redraw their innovation boundaries before rivals do.

If your competitors are using AI to scan every corner of the market while you rely on quarterly reviews, you won’t see them coming. If they replace slow data-sharing agreements with synthetic datasets while you are still negotiating approvals, they will move twice as fast. If their autonomous agents are already negotiating, designing, and refining while you wait for the next steering committee, they will set the pace – and you will be forced to follow.

I often tell my students that being a second, third, or even fourth mover can still win. But not this time. Leaders must act now. Decide where AI will enhance existing practices, where it will enable entirely new models, and where it will replace outdated approaches. Waiting for the dust to settle is not an option. By then, others will have written the rules – and taken the market. The leaders who win will be those who move with speed to seize opportunities, with discipline to cut what holds them back, and with judgment to turn AI into advantage that lasts.