You are watching a classic platform moment play out in real time. Apple can sell more iPhones and still have the bigger story happen somewhere else. That somewhere else is Services. When a company with a massive installed base posts strong Services growth alongside strong device demand, it is telling you something important about where leverage lives. Hardware creates distribution. Services captures compounding value. As a product leader, you should read that as a reminder that the best product strategy is often not a single feature. It is the system that turns usage into durable profit.

The Services angle matters because generative AI is creating new spending categories at speed. People are paying for subscriptions, they are buying add ons, and they are spending time inside AI experiences that did not exist two years ago. A meaningful share of that consumption rides on mobile ecosystems. If you sit at the center of payments, identity, and distribution, you can monetize AI demand even if you are not the company training the top model. That is the platform trick. You do not need to be first in capability if you are first in access to the customer.

There is also a counterweight you should keep in mind. AI changes how people discover information. If users ask an assistant instead of typing a query, traditional search volume can soften over time. That is not an academic concern. Search and advertising are giant profit pools for the companies that control default pathways and attention. When leaders get cagey about breaking out search performance, you can assume they are tracking the same tension you are. The short term can be positive because AI app revenue expands the ecosystem. The long term can be negative if assistants reduce the old behaviors that used to fund the machine.

Now bring Siri into the frame. If Apple powers Siri with a third party frontier model while keeping the Apple interface and privacy posture, it is making a product decision that many teams should study. Users do not reward internal purity. They reward outcomes. They want an assistant that understands context, plans tasks, and does not embarrass them in front of other people. If your product has a trust deficit, you do not get to wait for your internal roadmap to catch up. You have to ship credibility fast.

This is where the idea of good enough becomes a strategic weapon. The market does not require Siri to be the absolute best assistant in the world for Apple to win. Apple needs Siri to clear a threshold where customers stop laughing and start relying on it. Once you clear that bar, the rest of the Apple stack does the work. The device, the ecosystem, the defaults, the integrations, the polish, and the habit loops create retention. That is why buying capability can be rational. You are purchasing time to value, not just model output quality.

From a product management perspective, best is multi dimensional. Output quality matters, but it is rarely the only thing that matters. You also care about latency, uptime, data handling, safety controls, and how quickly you can integrate. You care about vendor support, change management, and the reality that your launch date is fixed even when your engineering plan is not. You care about the total cost at scale, not the demo cost on day one. If you can get a model that is good enough, reliable, and cheap enough, you can spend your differentiation budget elsewhere.

Vendor selection also looks different when you are shipping to hundreds of millions of users. You are not just choosing a model. You are choosing an operating relationship. You want a partner that can handle custom infrastructure requirements, enterprise grade security processes, and inevitable escalations. You want a partner that can keep improving without breaking your user experience. You want a partner that will not collapse under the weight of your roadmap. In that context, a large vendor that can do the messy integration work can outperform a technically superior vendor that cannot.

There is another subtle advantage in keeping the supplier in the background. If you own the interface and you do not market the underlying model, you preserve optionality. You can switch suppliers later. You can mix and match capabilities across different tasks. You can route some requests to your own models and others to external systems. You can do this while keeping the user experience consistent. The user keeps trusting Siri, not the name behind Siri. That is a powerful branding moat.

But you should not ignore the lock in risk that comes with success. Once you fix Siri with a strong external model, you raise the baseline expectation. Suddenly good enough means something higher than it used to. Now imagine you try to replace the external model with an internal one. You are asking customers to accept regressions in a feature they now rely on. You are asking your organization to take a working system and swap it for a system that has not been tested under real market pressure. This is one of the hardest product moves to pull off because you get punished for any downgrade, even if the long term plan is sound.

There is also a strategic risk if AI becomes the primary user interface. If assistants become the layer people interact with more than apps, then the model becomes closer to the product. In that world, the supplier is not just a component provider. The supplier influences capability, pace, and differentiation. If you outsource too much, you can wake up one day and realize you still own the hardware, but you do not own the new interface paradigm. That is the long term risk that hides behind a very reasonable short term decision.

You should also see the supplier incentives clearly. A vendor like Google benefits from being embedded in the dominant mobile ecosystem. If it can deliver a strong model at a relatively low price, it reduces the motivation for Apple to invest aggressively to compete at the frontier. That is a familiar playbook. Offer a deal that is good enough and convenient enough that the customer chooses partnership over competition. It can be mutually beneficial for years. It can also shape the competitive landscape in a way that is hard to reverse.

So what should you take away as a PM, even if you do not work at Apple. First, separate differentiation from requirements. Spend lavishly where you can create durable advantage. Be disciplined where you cannot. Second, measure time to value as a real strategic variable, not a scheduling detail. Third, treat supplier relationships like product bets because that is what they are. Fourth, if you plan to bring something in house later, define the migration path early and build guardrails against dependency creep.

The healthiest way to think about this is not as a moral debate about building versus buying. Think of it as sequencing. Sometimes you buy to stabilize the experience and restore trust. Then you invest internally where you can truly differentiate, such as privacy architecture, on device performance, user interface design, and ecosystem level integration. If you choose this path, you need a plan for what you will own long term. Otherwise you are not buying time. You are renting the future.

Here is the question you can use in your next roadmap review. If your AI supplier raised prices, slowed improvements, or became unavailable, would your product still have a credible path forward. If the answer is no, then the supplier is not a component. The supplier is a strategic dependency. And if the assistant becomes the next interface, that dependency becomes the story.