Apple just did something it almost never does.
It admitted it couldn’t build fast enough alone.
The Apple and Google Gemini AI partnership — confirmed in January 2026, reported at approximately $1 billion per year — is being read as a product story about Siri’s long-overdue upgrade. It isn’t. It is a partnership story. And it is one of the most consequential and instructive ones in years.
In The New Rules of Partnerships, I argue that in the AI era, ecosystem competition has replaced firm-level competition as the primary battleground. The critical capabilities — models, data, compute, distribution, regulatory positioning — now move faster than any single organization can assemble them alone. Partnering is no longer peripheral. It is the primary mechanism for assembling the scarce, heterogeneous assets that define competitive position. Apple licensing Google Gemini as the engine underneath Siri is the most visible confirmation of that thesis the technology industry has produced in years. If it can happen to Apple, it can happen to anyone.
WHAT EACH PARTY BRINGS — AND WHY THE ECOSYSTEM LOGIC IS SOUND
The commercial logic of this deal is grounded in something fundamental: each party brings what the other genuinely cannot build on the required timeline, and each retains what the other cannot be allowed to touch.
Google brings frontier model capability — Gemini’s ability to handle complex language understanding, multi-turn reasoning, world knowledge, and summarization at a level that Apple’s own models have not yet matched. This is not a gap Apple could close through incremental investment. It is a capability frontier that Google has reached through years of research, infrastructure, and data scale that represent a genuinely different asset base.
Apple brings something equally irreplaceable: distribution into a billion devices, a privacy architecture that remains the most trusted in consumer technology, a user experience discipline that no AI company has replicated, and a brand relationship with consumers that Google would not have been able to access through any other means. Google gains a strategic position inside the world’s most valuable consumer ecosystem. Apple gains an AI capability that closes the gap with rivals at a speed that internal development alone could not have achieved.
The structure reflects what good ecosystem partnership design looks like when both parties understand what they are doing. Separate identities. Interdependent capabilities. Apple controls the customer relationship and the privacy framework. Google provides the model layer. The customer experiences seamless intelligence. Neither party dissolves into the other.
This is the ecosystem architecture principle at the centre of The New Rules of Partnerships made visible at the largest possible scale: find what you cannot build, find who has it and needs what you have, structure the exchange with enough rigour to make it durable. Both Apple and Google have done the first two. The third is where the work begins.
BOTH ARE GOOD DEALMAKERS. BOTH HAVE MADE EXPENSIVE MISTAKES.
Apple and Google are among the most sophisticated commercial organisations on earth. They have each executed partnerships and acquisitions that have defined industries. But their partnership histories also contain failures that are directly instructive for how we should assess this arrangement.
Apple’s joint ventures with IBM in the 1990s — Kaleida and Taligent, both designed to create transformative platform technology — were total failures. Both companies staffed the efforts with people they didn’t want yet didn’t want to fire, and internal teams at both parent companies worked on competing initiatives simultaneously. The ventures lost mandate, lost talent, and were quietly dissolved after consuming years and tens of millions of dollars. The diagnosis from inside was unambiguous: the joint venture model itself was the problem — neither party was dedicated to its financial success, and people from the investing companies broke the cardinal rules of joint venturing at various times.
A decade later, Apple committed nearly $578 million to GT Advanced Technologies in a sapphire supply partnership that collapsed in bankruptcy within a year. Both parties failed to do what was needed to manage the relationship — Apple was not providing the human capital and knowledge required to make the technology viable, and GT was afraid to ask for help. The structural failure was one of governance: no early warning system, no performance visibility architecture, no escalation path that allowed the problem to be surfaced and addressed before it became irreversible.
Google’s own partnership history contains equivalent lessons. Its attempts to build content licensing arrangements with the entertainment industry collapsed repeatedly over misaligned incentive structures and inadequate governance of what each party could expect from the other. Its social and platform partnership initiatives of the early 2010s produced similarly instructive failures for the same root cause: commercially coherent concepts with structurally inadequate execution architecture.
Both organisations know what partnership failure costs. That history is itself a reason for cautious optimism about how seriously they will have approached the structural design of this arrangement.
THE QUESTIONS A PRACTITIONER ASKS
Analysing this deal through the lens of The New Rules of Partnerships, five structural questions determine whether this partnership delivers or becomes a cautionary study:
1) Incentive alignment as Apple Intelligence matures. Google’s $1 billion per year is contingent on Gemini remaining the model of choice inside Apple’s ecosystem. As Apple’s own models improve — which Apple is actively pursuing — the incentive gap between the two parties on this question will widen. Is there a defined mechanism for renegotiating scope and economics as that capability gap closes, or does that conversation happen under pressure, without an agreed framework, from asymmetric positions?
2) Data boundary governance at the system level. The deal is structured so that complex queries route to Gemini while simple ones stay on-device. That routing decision creates a data flow boundary that must be defined operationally, not just in principle. What can Google observe, retain, or learn from its position inside Apple’s ecosystem? Who has audit rights, and how are they exercised? In the AI era, data boundary governance is not a legal formality. It is a core commercial term.
3) KPI ownership and measurement custody. What constitutes performance for each party, who defines it, and who measures it? In a relationship where one party provides the model layer and the other controls the user experience, attribution of outcomes is genuinely complex. Ambiguity here is not neutral — it compounds over time and becomes the source of the disputes that erode partnerships from the inside.
4) Decision rights when competitive interests diverge. Apple and Google compete for advertising revenue, developer relationships, enterprise customers, and now AI market position simultaneously. When a product decision inside this partnership serves one party’s competitive interests at the expense of the other’s, what is the governance mechanism for resolving it? Without a clear decision-rights architecture at the working level, every significant disagreement escalates to principals — who are busy, increasingly irritated, and negotiating in public whether they intend to or not.
5) Exit architecture. This is the governance question that most partnerships address last and most inadequately. What are the provisions for an orderly transition if Apple decides to reduce or eliminate Gemini’s role as its own models mature? Are those provisions structured to allow a managed wind-down, or will exit require a renegotiation that neither party can afford to conduct openly given their competitive relationship?
WISHING THIS PARTNERSHIP WELL — AND MEANING IT
Apple and Google do not need my analysis to build a good deal. They have the legal sophistication, the commercial experience, and the institutional memory of their own past failures to draw on. What I find genuinely encouraging about this partnership — beyond the commercial logic, which is sound — is what it signals about where the AI era is taking even the most fiercely independent technology organisations.
The company that spent a decade preaching vertical integration as competitive doctrine has concluded that the pace of AI development makes ecosystem partnership strategically necessary. That is not a concession. It is a correct reading of the competitive environment. And it is the argument at the centre of The New Rules of Partnerships validated at the highest possible level of the industry.
In the AI era, the organisations that win will not be the ones that build everything. They will be the ones that partner most intelligently — identifying the right counterpart, designing the right structure, governing the outcome with the rigour the relationship requires, and maintaining enough strategic optionality to adapt as the ecosystem evolves.
Apple and Google have identified the right counterpart. They have designed a structure whose commercial logic is coherent. The governance is where the partnership will succeed or fall short — and that part of the story is still being written.
I wish them well. Genuinely. The ecosystem they are building together, if it holds, is good for consumers, good for the AI industry, and good evidence that ecosystem competition has permanently replaced firm-level competition as the defining dynamic of the technology era we are all navigating.
Watch the governance. That is where this story gets written — and where every partnership story always does.
Randy McGraw is the founder of M2 Ventures and the author of The New Rules of Partnerships. He has been the lead day-to-day architect of more than $2.3 billion in commercial partnerships and joint ventures across Japan, ASEAN, and the US. M2 Ventures advises corporate clients on partnership architecture, negotiation, and governance.