When all the top tech companies seem to moving in a pack toward artificial intelligence, Apple has stood startlingly apart. Its infrastructure investments haven’t ballooned. The presence of AI in its products is, comparatively speaking, minimal. And when Mark Zuckerberg came knocking with huge paychecks for Apple’s talent, Tim Cook didn’t do all he could to retain it.

There are two ways to look at this state of affairs. One is that Apple is in disarray, its AI products don’t work because it has been caught napping on the next great tech revolution and is hemorrhaging talent as a result. Another is that Cook is exercising restraint as others in Silicon Valley lose their heads.

This week’s Apple event in Cupertino, California, will be the usual affair of incremental changes to its core lineup, plus a new, thinner iPhone “Air.” The more interesting developments for the future of the iPhone are taking place behind closed doors as Apple seeks to reinvent Siri for the AI age. The company needs to reverse the virtual assistant’s status as a byword for dumb AI. No other Apple brand comes close to receiving this kind of mockery. When AI captured the world’s attention, Siri’s hopelessness became a serious problem.

Make no mistake, Apple’s first choice would have been to solve the matter in-house and use its own engineering talent to make Siri smarter. That hasn’t happened — or it hasn’t happened yet, at least. Instead, the company is now looking to outsource the task by bringing in a new tutor.

That tutor could be Google and its Gemini AI. The two companies are testing and fine-tuning a custom-built version of Google’s model that runs on Apple’s servers and would be used for summarizing, Bloomberg News’ Mark Gurman reported. We’re still weeks from any final decision, and Apple still may decide to go it alone. But the talks reveal at least that Apple is thinking pragmatically. It may not make much sense to sink billions of dollars into building its own AI when, as the leading hardware maker, it has the power to go out into the marketplace and choose whatever models it considers to be well suited. It can use the dominance of the iPhone to help push for the best possible terms, playing potential partners against one another, much in the way it squeezes those responsible for its components and manufacturing. As well as talking to Google, Apple has considered using Anthropic’s Claude, Gurman wrote — but the firm wanted too much money.

Outsourcing in this way might seem embarrassing. Apple’s shortcomings in AI have been a significant contributor to it being second only to Tesla Inc. as the worst performing stock in the Magnificent Seven this year. And yet it needn’t be seen as a weakness.

Even trying to continue to go it fully alone would come with risks. It would be expensive and extremely risky — Apple might not ever pull it off with the talent and infrastructure it has. It would mean wading deeper into the wild scramble for chips, power sources and data center real estate.

That leaves two backup options. No. 1, as has been hotly speculated, could be to buy an AI company. Under consideration has been Perplexity and Mistral. Either would come with a staggering upfront expense and would need significant continuing investment. Further, the top talent that would make any target appealing may not want to become a cog at a behemoth like Apple, while those that do stay might struggle with the culture — one key reason huge acquisitions aren’t something Apple typically does and something Meta Platforms Inc. is learning after several of its big AI hires already quit. And, of course, Apple runs the risk of simply backing the wrong horse and then being stuck with it.

The second option, then, is outsourcing, to see AI as something that can be bought-in as needed until such time as Apple can do it better (or at least as good) itself. A Google-made AI model might adequately suffice much in the way Apple used Intel chips in its Macs until it was happy its own designs were up to the task. As time goes on, and the AI market matures with new use cases, Apple can adapt as necessary.

As Plan B’s go, this outsourcing is an attractive and versatile way forward. In the current climate, you might even call it thrifty.

There is one snag, however: The companies that Apple might look to partner with on AI see themselves as Apple’s fierce competition. At Google’s Pixel event last month, part of the (muddled) sales pitch was stressing how easy it would be for users to ditch their iPhones. And it was only three months ago that Jony Ive and Sam Altman were getting pally in a San Francisco cafe, talking about how their new AI device would shake up the status quo of how we use devices.

Will Apple help its competitors by partnering with them on AI? Sure, maybe a little. But by waiting much longer to implement AI into the iPhone, Apple would be opening that window of competitive opportunity far sooner. What matters most for the future of the iPhone is that its AI functions work and are useful. If that can be achieved, the vast majority of users won’t give a second thought as to who built the underlying AI model for Siri, just as they pay little attention to who designed the iPhone’s modem or screen. Partnering may not have been Apple’s preferred choice, but it’s still a good one.