Unpacking the AI Economy: Part 1
How Specialist Partnerships Will Dominate Enterprise AI
It’s 2005, and I’m sitting in an executive office of the Queensland State Government on George Street in Brisbane. The conversation is free of the usual rigour and constraint I’m used to in these offices. We’re discussing a concept that feels both revolutionary and daunting: open source software.
Across the table, a couple of senior bureaucrats listen intently as we try to explain its transformative potential. It is software that isn’t locked behind licenses. It encourages co-opetition and collaboration. It offers flexibility governments and enterprises desperately need. It will attract businesses to the state. It will drive innovation and economic growth.
It wasn’t the first time we had met to discuss this topic. In hindsight, the challenge was never about selling the idea of open source, though that was difficult at times. It was unpacking the ambiguity that surrounded such a new concept. The “what’s in it for me?” type of questions. Open source was perceived as unstable, risky, poorly understood, and lacking commercial clarity in its business model. It sounded abstract to decision-makers used to dealing with proprietary systems. No one could explain it to their mums.
Ultimately the pathway forward came, not from discussions about its potential, but in breaking down the ecosystem in which it would operate. It came by highlighting how open source could create value and wealth through clear market segmentation. It came when the economic pathways were understood. Foundational platforms like Linux, applications like Apache, and the service models offered by companies like Red Hat became the market we know today once everyone knew their place in the overall operational framework. Only then it could be applied to real problems at scale.
Those few weeks taught me an important lesson about picking winners and losers in tech. It was that when a new concept feels overwhelming, and most of them do, clarity only comes from segmentation and specialisation. It was that the meaning and impact of big tech trends can be transformative for some, but different for all.
The open source market achieved this clarity by evolving from a vague, monolithic idea into a thriving ecosystem of interdependent players each excelling and building a specific niche. It then revolutionised enterprise IT.
Fast forward to 2025, and history is repeating. This time, the concept is artificial intelligence. And once again, we’re only just beginning to unpack this ambiguous yet transformative technology.
What we can say today, with complete clarity, is that success in AI will not be measured by the obvious dominance of a few goliaths. It will only come through specialisation, segmentation, and the development of a democratised ecosystem tailored to the needs of individual industries, sectors and businesses.
So after years of foundational development by the tech giants (you have an enduring thanks), this is the story of how the AI economy is moving into its second act, just as open source did nearly two decades ago.
The AI Economy is a Market Not a Monolith
To better understand the AI economy, it helps to draw parallels with the open source economic model, where the market was driven by the interplay of foundational components like the “Kernel,” Distributors”, and contributions from an active “Community”. Similarly, the AI economy is not a singular technological phenomenon. It is a dynamic and interdependent market with distinct roles and participants (see image1).
In the open source world, the “Kernel” served as the foundation for everything, with distributors packaging it for broader use and communities building, refining, and expanding its utility.
In the AI economy, Suppliers like OpenAI play a similar foundational role. As the creators of LLMs, which serve as the raw, versatile "kernel" of AI, and GPTs, which are specialised implementations of these models, they provide the core technology that powers countless downstream applications and innovations, much like Linux underpins the software world.
Similarly, NVIDIA can also be considered a supplier. But instead of foundational models, it supplies the enabling hardware, the GPUs and accelerators, that make the training and deployment of these AI models feasible.
Beyond suppliers, the AI market includes two other essential roles, Enablers and Intermediaries. Enablers provide the platforms and tools, needed to scale and adapt foundational technologies for broader use, akin to how distributors like Red Hat tailored and supported Linux for enterprises.
Companies like ServiceNow (through NowAssist) and Microsoft (through CoPilot) and Salesforce (through AgentForce) serve as enablers in AI by offering enterprise solutions that make foundational models accessible and practical for diverse applications.
Intermediaries, meanwhile, act as the facilitators of the AI market, bridging gaps between suppliers and buyers. They align AI solutions with the specific needs of businesses, much like how open source communities and integrators helped tailor Linux to unique use cases. These intermediaries include consulting firms, integration specialists, and marketplaces that connect supply with demand and amplify the market’s overall effectiveness.
This layered structure highlights that the AI economy, like open source, is far more than the sum of its parts. But more than that, it highlights how its success, and the success of the different players within the market, relies on the interplay of specialised roles to each drive innovation in its domain while depending on the contributions of others.
By viewing the AI economy through this lens, the crucial insight becomes clear: the partnerships that span across the distinct layers of the ecosystem will ultimately define which vendors become dominant. There are too few of these at the moment.
For suppliers like OpenAI, currently akin to the "kernel" in the open source model, their legacy will be determined by how effectively they enable and integrate with a thriving ecosystem. Theirs, and as a major investor, Microsoft’s, success hinges on fostering relationships that extend their foundational technology into diverse applications, industries, and use cases.
For enablers like ServiceNow and Salesforce, success will depend not on partnerships with other enablers like Microsoft, but on building practical, diverse connections both upstream and downstream.
That is, enablers must actively seek partnerships with industry-specific suppliers (for example a public-sector or jurisdictional LLM), and intermediaries that can tailor these AI capabilities to real-world needs. By embedding themselves into these targeted, cross-layer partnerships, they create the potential for exponential value rather than incremental improvement.
Partnerships within the same layer of the AI ecosystem (e.g., Supplier-Supplier or Enabler-Enabler) are valuable and widely adopted, but they inherently lack the broader impact and transformative potential of cross-layer collaborations.
To truly drive innovation and unlock the vast potential of the AI economy, both suppliers and buyers need to step outside these familiar, insular networks and form partnerships across different layers of the ecosystem.
This approach opens up new avenues for economic innovation and democratises AI by breaking down the barriers created by mega vendors.
It also ensures that the AI solutions that will add the greatest impact aren’t confined to the gated communities of large enterprises but become accessible and transformative for businesses of all sizes and industries, and reach into every corner of the business in every corner of the economy.
Bottom Line: The future of the AI market is going to be shaped by partnerships between dominant AI giants and emerging or lesser-known solution providers, where innovation and market momentum are converging. These are the press releases and vendor communications I’m watching out for.
I’ve written a follow-up piece diving a bit further into this. It was originally part of this article, but I’ll be publishing it as Part 2 next week. Stay tuned and make sure to subscribe or follow so you don’t miss it!
I have not addressed the Regulators and Skills sections in detail, as their relevance is largely self-evident. However, it is worth noting that segmenting regulators provides greater clarity in commercial relationships, and that distinct AI skill sets and talent pools are associated with each market segment within the model.