There is a version of the partnership playbook that most technology companies have been running for the better part of two decades. Co-market to build awareness. Co-sell to generate pipeline. Certify partners to ensure quality. Build for breadth. Lead with the vendor’s story.
The market has already rendered a verdict on how well that playbook is holding up. According to Zinnov’s State of Partnerships H1 2026 report, AI-native firms grew 148% in market cap between Q1 and Q4 2025. In the same period, GSI valuations fell 15%. That divergence is not a blip. It is the market repricing who creates value in the technology ecosystem and how. The conversations happening inside hyperscalers, platform companies, and the firms shaping technology services investment right now reveal an industry that has quietly outgrown the assumptions its partnership model was built on. The rules haven’t been officially retired. They’re just producing diminishing returns, and fewer people are willing to say it out loud.Â
The partner ecosystem is estimated to represent a USD 840 billion to USD 1 trillion opportunity by 2030. The organizations that capture it will not be the ones with the largest partner networks or the most established programs. They will be the ones that recognized early enough that the rules had changed and acted on it.
Here are five of them.
For years, the sequencing felt intuitive. Build brand awareness through co-marketing, generate pipeline through co-selling, and only invest in co-building once the relationship is mature enough to justify it. Build last. Market first.
Co-building is now the entry ticket. When a significant share of enterprise buying research happens before a customer speaks to a vendor or a partner, showing up with a campaign is not enough. A solution needs to already be integrated at the technical layer, with the platform, the data infrastructure, the security architecture the customer runs on, before any marketing investment can land. A customer who finds a partner through a campaign and then discovers that real deployment requires months of integration work is not a customer who stays.
The vendors seeing the strongest partner outcomes have flipped the investment order: co-development first, validated customer outcomes second, marketing those outcomes after. Jason McIntyre, VP, Consulting and SI Program, Scale Ecosystem at Databricks, puts it directly: “Where everybody wants to get to the pinnacle is going to be an investment in co-development.” Investment in co-marketing decoupled from proven delivery rewards the wrong behavior. The model gaining traction ties funding to activation milestones, pilot, deployment, outcome, rather than to pipeline volume alone. The spend follows the proof, not the promise.
The logic used to hold. The more competencies a partner carried, the more customer situations they could address, the more deals they’d be included in. Signing more partners, covering more use cases, building a bigger ecosystem, all of it felt like the right direction.
When a platform like Databricks works with over 5,000 partners, and every one of them has checked the same boxes, cloud certifications, AI competencies, industry solutions, breadth becomes the entry fee rather than the differentiator. The partners earning consideration are the ones who can point to a specific industry, a specific workflow, and a specific outcome they have already delivered for a customer in that space. Verticalization is where that proof lives, with partners who can carry a conversation from the IT function into the line of business, to a CFO, a head of operations, a business owner, winning engagements that generalists are not in the running for.
The market is rewarding depth in two forms: vertical specialization that earns a seat at the table with business decision-makers rather than just technology buyers, and proprietary IP, pre-built agents, fine-tuned models, workflow accelerators, that lets a partner demonstrate an outcome before the engagement begins. As Alyssa Fitzpatrick, GVP Partner Sales at Elastic, puts it: “Differentiation is less about checking all the boxes and more about what you do best. Double down on that. Because if you’re really good, you will succeed. But if you’re a little bit good at everything, there’s nothing that’s going to make you stand out.”
The bar for partner qualification has moved. Training, testing, and tiered credentials were built for a world where technology was stable enough that knowledge transfer was the primary barrier to deployment quality. That world no longer exists. Model capabilities evolve continuously, not in fixed release cycles, and a certification earned six months ago does not reflect what the platform can do today or what customers are now demanding.
The gap that keeps surfacing in enterprise AI deployments is between what partners can demonstrate in a controlled environment and what they can execute in a live enterprise, with real data, legacy systems, and a customer measuring return on investment before they expand the engagement. Michael Khoury, Vice President, Global Ecosystem Partners at Palo Alto Networks, captures the standard that is emerging: partners need to be not just deployed but successfully deployed, meaning they have adopted what they purchased and are actively extracting value from it. John Kain, Head of Financial Services Market Development at AWS, observed that many AI proofs of concept that fail to reach production were never designed to solve a business problem that would justify the investment of scaling. The ones that succeed share a foundation: a clearly defined business outcome, a data architecture built to support it, and a partner who understood both before the build began.Â
That combination is what enablement programs need to produce. As Aly Shivji, Group Vice President, Cisco, frames it: “Partner enablement is table stakes. Our partners are demanding not just enablement but co-engineering — tools, resources, APIs, MCP servers to really take advantage, to build their AI agents on top of. That’s where we’re trying to do things more than just partner enablement. It’s partner co-engineering and partner innovation.“
Almost every major technology vendor describes itself as partner-led. It appears in earnings calls, partner program documentation, and go-to-market frameworks. It has become, at this point, close to universal, which is part of the problem.
Partner-led, as it has been practiced, often means something closer to “partners carry a portion of our pipeline.” The vendor still controls the product roadmap, the customer relationship, the pricing, the narrative. Partners operate in the space the vendor allocates to them. When that space is generous, the model works. When the vendor’s own field team starts chasing the same accounts, partners notice, and they remember.
The term that actually describes what works is customer-led: the entire motion, vendor, partner, hyperscaler, ISV, organized around what the customer needs to reach a specific outcome. That reframe has real organizational consequences. It requires the vendor to be comfortable with partners owning the customer relationship, the delivery, and the outcome, not just carrying a quota number. Kristine Henley, WW Head, Public Sector Partner Core at AWS, describes what that looks like in practice: “When it comes to individual customer deals, completely hands-off. We allow the partner a good amount of trust and we base the funding on the results overall for the quarter.” The deal-by-deal co-sell motion, she notes, is not where the collaboration happens. It happens at the go-to-market stage, around which solutions to prioritize, what the better-together story looks like, and how to show up jointly in market.Â
Some of the fastest-growing cloud businesses of the last decade were built on an explicit structural decision to keep professional services deliberately small and push delivery ownership entirely to partners. The SI practices that followed were a direct result of that commitment being structural rather than rhetorical, with comp structures, field incentives, and investment allocation all pointing in the same direction. Any vendor serious about partner-led growth faces the same question eventually: does the operating model make it rational for a partner to invest, or are we asking them to absorb risk we are not willing to share? The comp structure and the field incentives answer that question. The program documentation does not.
The ecosystem has always had a clear hierarchy. Global SIs at the top, large managed service providers in the middle, regional and boutique partners below. Investment, co-sell priority, executive relationships, all of it flowed toward the top. The logic was scale: the biggest partners could touch the most customers, carry the most revenue, and absorb the most enablement investment. That hierarchy is being disrupted faster than most ecosystem programs have registered, and the market cap divergence in the opening is the clearest signal of why.
The partner archetype gaining ground is IP-led, agentically native, and priced on outcomes. These are companies two to three years old at most, small in headcount, but built around a narrow and focused IP play in a specific business area. They are showing up with models that produce verifiable results, priced on outcomes rather than time and materials, and they are winning deals against much larger incumbents because enterprise customers doing real due diligence are finding that output quality and cost structure both favor the smaller firm.
The financial logic behind this shift is straightforward. When a partner is selling time and materials and the customer is buying outcomes, every efficiency improvement AI brings to delivery reduces that partner’s revenue while the customer’s value stays the same or increases. Jeff Rich, Operating Partner at Sunstone Partners, framed the structural risk plainly: “Just because you were part of the SaaS transition and the cloud transition does not guarantee you are going to be part of the AI transition and have the right skill sets to help companies adopt change.” The partners building durable businesses right now measure success by whether outcomes show up in the customer’s revenue growth, gross margin, or EBITDA, not by certifications or headcount.Â
For ecosystem programs still structured around preferencing the largest, most established partners, this creates a real allocation problem. The partners who will drive disproportionate customer value over the next three to five years are not all on the current tier-one list. Finding them early, investing in them before they are obvious, and building commercial infrastructure that supports outcome-based engagement is the work most ecosystem teams have not yet started.
None of these rules failed because they were poorly designed. They were right for the conditions that existed when they were written. Those conditions have changed, comprehensively and fast enough that the rules need to change with them.
The partners gaining ground share a pattern: they know exactly what they are good at, they have built enough IP to show a customer an outcome before the contract is signed, and they have found the right ecosystem to amplify that. The ones losing ground are still checking boxes, still running broad, still waiting for a co-marketing campaign to do the work that a proven delivery track record should be doing.
The window to act is open. It is also narrowing.
THANK YOU
FOR YOUR RESPONSE
