From pharma companies to automakers, industrial and service companies are entering into dependent relationships with global tech firms as they seek to advance their futures with AI. So far, however, the track record is poor. In 2024, nine out of ten partnerships between industrial and tech companies were failing to meet their goals, according to BCG research.
In the era of agentic AI, when autonomous AI systems can act, negotiate, and transact on a company's behalf, the failure rate is likely to rise further. Partnerships based on agentic AI face new and complex challenges, such as governance and liability, that can strain relationships between companies whose histories, worldviews, and decision-making speeds may be very different.
Our work with clients demonstrates that executives can significantly improve their chances of success by choosing the right partnership archetype for a particular collaboration and by avoiding six common pitfalls during setup and operation of the partnership. Recognizing and sidestepping these pitfalls can be the difference between a partnership that transforms an industry and one that joins the long list of failures.
The Indisputable Logic of Partnerships
Despite the high failure rate, the underlying logic for tech-industry partnerships is sound because they can solve a number of fundamental strategic problems. (See Exhibit 1.)
A leading industrial company can supply deep sector knowledge, a relevant brand, and customer relationships. It can provide domain-specific data that is essential to train AI and, more generally, power innovation. In addition, an Industrial company can ensure relentless focus on value creation, steering projects away from “interesting” technical challenges or tokenmaxxing, which measures success in AI usage rather than revenues and margins. A tech player complements these strengths with frontier AI expertise and infrastructure, an agile culture, and the ability to manage and harness data at scale.
Critically, the value unlock today is no longer about deploying or customizing off-the-shelf tools. It comes from co-innovation: both parties committing the engineering talent, capital, and roadmap required to rewire core products and workflows together. This sharing of risk, governance, and upside distinguishes a partnership from a more straightforward vendor model, in which the industrial partner pays the tech player and receives a clearly defined product or service in return.
This cooperative effort drives embedded delivery, in which tech player engineers operate inside the industrial company to close the last mile between AI ambition and production, a long way from traditional tech licensing. The new team can be quite large: in the auto industry, one three-year co-development agreement created a collaboration team of more than 1,400 specialists across six countries. However, rewiring an industry in this way, makes negotiations over governance, platform lock-in, and IP ownership even more difficult.
The Four Tech-Industry Partnership Archetypes
The realities of these complex, sometimes asymmetric, relationships quickly become apparent when the partnering companies move from theory to practice.
In our work with clients, we see four archetypes in common use. No archetype is superior to the others in all settings. To select the right one, the partners must candidly assess their goals, which may be only partially aligned, and agree on which model delivers the optimal outcome.
The AI Transformist
This archetype applies a tech player's agentic AI and platform capabilities to transform an industry's core workflow or business model. Typically, benefits materialize only after deep embedding in the industrial partner. For example, Australian life insurer TAL created a strategic partnership with Microsoft to build a suite of internal AI-powered tools, including a knowledge base that TAL says saves seven minutes per question and has earned 93% positive user feedback.
The Tech Co-creator
This archetype aims to drive tech innovation, such as a breakthrough product feature. Deep tech development may include, for instance, the creation of new AI models. The auto industry has widely deployed this model as manufacturers seek to develop autonomous-driving capabilities. The industrial partner may be perform multiple roles:
- It may serve as the integrator, as in the case of the various automaker partnerships with Wayve, which has developed a self-driving platform.
- It may function as a co-developer, committing heavy in-house capability to create a new platform that partners can sell to other industry players. BMW’s extensive partnership with Qualcomm follows this pattern.
- It may act as the lead developer, turning to tech players for vital enabling technology such as semiconductors. Tesla’s deep relationship with Samsung for AI chips, connectivity, and more is an example.
The Venture Launcher
This is a more ambitious archetype and may involve creating new corporate entities or equity stakes. For example, French drugmaker Sanofi is collaborating with pharmatech company Owkin to build novel AI agents that will automate and accelerate Sanofi’s drug discovery and development. When the partnership started in 2021, Sanofi invested $180 million in Owkin stock.
The Cross-Border Integrator
The rationale for this archetype is to ease access to markets or navigate geopolitical tension. For example, Volkswagen’s partnership with Chinese EV specialist XPeng builds EVs in China that combine VW’s engineering and production know-how with XPeng’s advanced software platform, which includes advanced self-driving. In some cases, industrial companies select a partner on geopolitical grounds; for instance, some large European companies have formed partnerships with the French AI firm Mistral because they want to align with a local, European player.
The Six Pitfalls of Tech-Industry Partnerships
Industrial companies have become adept at procuring technology. But creating and operating a tech partnership—especially one with ambitious, transformational goals—is a far more demanding undertaking.
Choosing the most appropriate archetype is a good start. But with pursuing a productive partnership that has a multiyear lifespan and depends on partners with different strengths and goals requires caution, foresight, and good judgment. From working with clients, we have found that six pitfalls are most frequent and dangerous. Each leads to disappointing results, wasted investment, and—worst of all—a missed opportunity to create breakthrough innovation. The first four are design-time pitfalls; if any provisions are ill-conceived at the agreement signing stage, everything downstream of them inherits the problem. The next two are execution pitfalls. Unless management is alert, these problems may slowly emerge over months or even years and undermine even a well-planned partnership.
A Hazy Value Proposition
Every sector has seen a landmark partnership agreement between an industry giant and a global software company that generates a lot of excitement and then... nothing. Often, the underlying problem is a hazy value proposition, and the solution is better planning at the outset. The initial timeline should include near-term and interim milestones as well as the big vision to be realized over three to five years. Typical first-year targets include building the foundation and achieving quick wins, followed by goals for years two and three that expand this foundation and begin driving broader use cases.
Misaligned Goals and Incentives
It is normal for the two parties in an asymmetric, tech-industry partnership to start with very different objectives. The key to success is to defuse conflict, accept tradeoffs, and reach alignment. For the industrial company, a three-step process brings clarity:
- Align internally first. Conflicts between functions within the company—the CTO pushing for scale, the CFO holding the line on costs, business units wanting quick wins—need to be settled before talks with the tech partner begin, rather than surfacing midway through partnership negotiations. This preparatory work also helps define red lines for the partner negotiations.
- Model the tradeoffs. Rather than debate options in the abstract, partners should put numbers against two or three realistic deal structures. For instance, they might pit a narrow use-case license against a full co-development agreement, and score each on speed to value, control retained, and IP exposure. This amounts to letting the model pick the structure, instead of deferring to the person in the room who argues loudest.
- Address conflicts in governance. Provisions in contract agreements can deal with difficult issues such as lock-in. Options may include a mandated multicloud or multimodel clause, data-portability rights, and periodic price benchmarking against market alternatives.
- Think about exit-value logic. Goals aren’t the only things that that differ between partners. In most partnerships, the tech player’s exit logic involves selling data and know-how from the industry partner across the sector, using the industry partner as a reference. Rather than being surprised later, the industrial partner should build this intended outcome into the negotiation—for example, by securing a time-limited exclusivity window before the tech partner can offer specific intellectual property (IP) to a named rival, with lower ongoing fees in return for being a named reference account.
Unclear Data and IP Ownership
Going into a partnership, most industrial companies fear that the proprietary IP or data that they share in a tech partnership will eventually give rivals a competitive advantage. And they are right to be concerned: in many cases, the tech partner will want to commercialize its share of the IP across an industry, not just maintain it with a single player. The resulting risk-and-reward equation is complex. (See Exhibit 2.)
The starting point for the two partners should be a full specification of how each can independently use the new IP and on what commercial terms. Three other actions are also essential:
- Fully declare preexisting, background IP so there is no confusion about what is new and what has been contributed but already existed.
- Differentiate between hard IP, such as patents that require formal ownership, and soft IP, such as know-how protected by nondisclosure agreements.
- Include termination clauses that specify what happens to the IP if the partnership ends.
Conflicts over data ownership can be a trap, draining momentum from the venture and creating a win/lose dynamic that undermines cooperation. The focus should instead be on carve-outs: the industry player should prevent the data and IP from being used where it really hurts (named rivals, specific regions, key use cases) but otherwise give the tech player a free hand. Taking this approach locks in the protection that matters and is easier to win.
In addition, the industry partner has options for balancing risks and reward. Weighing these options may tie back to archetypes. For example, opting for a series of use-case tech partnerships will reduce data risk but is unlikely to have the industry-defining impact of a comprehensive alliance with a global software or AI player.
Unclear Governance
Governance is not simply a matter of writing the appropriate partnership agreement. It is a multilayer, living process:
- CEO alignment must stay on track. An annual CEO review can showcase the company’s continued commitment to the partnership, both internally and externally, and can provide an opportunity to realign the partnership vision if needed.
- Executive sponsors should meet quarterly. Sponsors should review progress and KPIs, and monitor investment inputs and IP outputs.
- The partnership leader from each player should meet twice a month to ensure alignment. These representatives can allocate resources, assess joint efforts, and track KPIs in detail.
- Within the partnership, the managers of each value stream should meet monthly. The managers should identify blockers, agree on solutions, and share lessons learned across streams.
Unclear governance brings an additional risk: companies may keep a partnership alive even when it no longer delivers value. Paradoxically, effective foundation work, such as aligning KPIs and driving a sense of wide ownership, may increase the reluctance of key stakeholders to accept that the alliance has served its purpose and should come to an end. To reduce this risk, company leaders should schedule reviews that include impartial stakeholders. Reviews should also be triggered when key KPIs are not met.
Failure to Activate Results
As the partnership starts to deliver results, the industrial company will usually want to activate the new IP by building it into products and services. However, business units may have different priorities or lack the skills to leverage advanced technologies such as agentic AI. In addition, they may have no incentive to deploy IP that originated outside their domain. The regional CEO of a global technology company told us recently, "If the front line does not see a path to revenue, the whole thing fizzles out.”
The solution to this problem starts at the top, with enthusiastic executive sponsors. But they must take action to drive activation down into the organization. (See Exhibit 3.)
Aligning the KPIs of business units so that they have an incentive to use the partnership IP is vital. Executives should also consider creating a dedicated partnership office to coordinate and activate the partnership’s resources and outcomes.
Failure to Bridge Cultures
The head of strategy at a global software firm told us about a key meeting where his team showed up with its share of the work fully done while “our industry partner had not done anything, as decisions were still going through approval processes at different layers of the organization.”
This highlights a key fault line that may quickly emerge when tech-industry partnerships get underway. Despite frequent talk of cultural mismatches, the real issue is the speed of decision making. The tech player decides in days; the industrial partner needs to complete an approval process measured in weeks.
To resolve this mismatch in latency, the industrial company usually must do the hard work. Three options are often helpful here:
- Build flexibility into the agreement. Partnerships that start with pilots typically require fewer approvals.
- Use sandboxes to reduce technical risk. The greater structural protection can translate into a need for fewer sign-offs.
- Structure the partnership to decouple from business as usual. One possible model is the speed boat methodology, which allows the partnership to accelerate away from the slow-moving mothership. The company should delegate a carved-out decision mandate and budget to the partnership lead in advance, so the lead doesn't have to reenter the mothership approval chain for every call.
The Strategic Value of Tech-Industry Partnerships Will Increase
Although the array of pitfalls is daunting, industrial companies are right to actively seek new partnerships that bring the power of leading-edge technology—particularly agentic AI—into their sector. Increasingly, AI will be woven into the core fabric of every product and workflow, changing the parts of the value chain that capture the most significant profit. The impact will be transformative. In the BCG survey noted earlier, 94% of tech and industry stakeholders said that they expected these partnerships to become more common.
Certainly, there are risks. These partnerships require huge effort from both sides, and in-depth planning is essential to keep the venture on track and ensure that the output delivers commercial benefit. But the bigger risk for the industrial partner would be to proceed without a plan, thinking that tech magic sprinkled on their products and services will future-proof their business.
The overall picture is still highly dynamic, but companies that choose an appropriate archetype and avoid the six pitfalls will reduce their risk significantly and position themselves to move quickly up the learning curve. Although the tech player will probably eventually offer the IP developed by the partnership to the market, the industrial partner will shape it from the start, ensuring that it fits with its vision of the industry. Just as important, the company will gain access potentially years ahead of its rivals, a vital competitive advantage given the time-compounding nature of AI development. For those who plan wisely, the advantages of partnership can be industry-defining.
