The most powerful technology companies simply cannot stop talking about artificial intelligence, and in particular, the “generative AI” flavor that can create human-like text, images, and code.

During calls after this week’s earnings reports, Alphabet CEO Sundar Pichai and his team said “AI” 66 times. Microsoft CEO Satya Nadella and his execs said it 47 times. And on Wednesday, Meta CEO Mark Zuckerberg and the Facebook executive team said the magic phrase 42 times, according to a CNBC analysis of transcripts.

But Apple barely talks about artificial intelligence, and you shouldn’t expect to hear much about it during the company’s earnings next week.

Its sober approach to the new technology contrasts deeply with its rivals, which are stoking excitement and elevating expectations every chance they get.

During May’s Apple earnings call, CEO Tim Cook only said “AI” twice, and that was in response to a question. During Apple’s two-hour software launch event in June, it never said the phrase, although it announced several new features powered with AI.

Apple execs instead use the phrase “machine learning,” which is more popular with academics and practitioners. Apple execs also prefer to talk about what software does for the user, such as organizing their photos, improving their typing, or filling out fields in a PDF, as opposed to the technology that makes all that possible.

Apple’s approach to AI as a core underlying component instead of the future of computing represents a way to present the technology to its consumers. Apple’s AI works in the background. And the company doesn’t yell about it the way some of the other companies do because it doesn’t need to.

Google, for example, has announced its plans to revamp its search engine using an AI model called Search Generative Experience. Microsoft’s biggest new initiative is a $30-per-month “Copilot” subscription that integrates generated text or code from partner OpenAI’s ChatGPT into Word, Powerpoint, and other apps. Meta’s most recent investment in AI technology is its own large language model it calls LLaMA, which could underpin new kinds of social media chatbots or automatically generate online ads.

Meanwhile, Apple still makes the bulk of its money from iPhones, which generated $51.3 billion of its $94.84 billion in revenue during the company’s second fiscal quarter. Why talk a big AI game?

Besides, mega-cap tech companies signaled to investors earlier this week in earnings calls that the rollout of AI products could take a while.

In Microsoft’s case, Nadella tempered investor expectations for Copilot, signaling that growth would take time, and said that its rollout would be “gradual.”

It could take until next year before investors understand how the Copilot subscription affects the company’s revenue. “In the second half of the next fiscal year, we’ll start getting some of the real revenue signal from it,” Nadella said.

Google and Pichai say that the company’s text-generating AI models will make its search engine better and could even answer questions that normal Google search can’t. From a business perspective, Pichai said, generative AI used for creating and serving ads will “supercharge” the company’s existing ads business, and there are “opportunities” for new kinds of ads with AI-generated search.

But Pichai still said it’s still “early days” for the new AI-powered search, and later, when pressed about how SGE might increase usage of the search engine, and therefore increase revenue, he said the company was experimenting.

“I think we are definitely headed in the right direction, and we can see it in our metrics and the feedback we’re getting from our users as well,” Pichai said.

Zuckerberg was effusive about AI technology and its applications in virtual reality, ad targeting, and recommending content from accounts users don’t follow.

He was particularly optimistic about a concept called “AI agents,” where software would be able to message business customers automatically without a human involved, or act as a coach, or be a personal assistant.

Still, Zuckerberg admitted he didn’t know how many people would use the new AI features.

“The reality is, we just don’t know how quickly these will scale,” Zuckerberg said. He said Meta was debating internally how much it should spend on servers for AI.

The slow rollout of revenue-generating AI products from Big Tech matters because many people in the technology industry believe that new foundational technologies go through a “hype cycle” based on research from analysis firm Gartner.

When a new technology is introduced, according to the hype cycle model, it gains lots of attention and investment as it reaches a “peak of inflated expectations.” But, as the deployment of the tech moves slower than initially expected, enthusiasm and investment dry up, in a “trough of disillusionment,” before maturing and becoming productive.

For now, shovel-makers and people seeking investment capital are benefiting from the AI boom. Nvidia stock has risen 220% so far in 2023 as investors have realized its GPUs are essential for the technology. Venture capital investment in AI startups has boomed, and many of those dollars are going to Nvidia for computer capacity, and to cloud providers for access to AI models.

But if everyday consumer applications for AI don’t catch on, then many AI companies could slip into the trough of disillusionment again. Analysts found earlier this month, for example, that downloads for OpenAI’s iPhone app slowed earlier this month after launching in May.

Some analysts are starting to understand that an investment opportunity based on new AI products won’t be immediate and that the costs could stack up.

“We cautioned investors that that process of translating early demand to large-scale implementations and recognized revenue will be a multi-year trend rather than an instantaneous flip of a switch,” JPMorgan analyst Mark Murphy wrote this week.

“We recommend investors invest elsewhere until Metaverse, Reels, Threads, Quest and Generative AI investments become accretive (if ever) to META’s [return on invested capital], rather than dilutive,” Needham’s Laura Martin wrote in a note.

UBS analyst Lloyd Walmsley wrote this week that Generative AI was still an “overhang” over Google.

“Management expressed optimism around the ability to solve for ‘deeper and broader’ use cases with Search Generative Experience (SGE), but we do not believe the company is out of the woods with management still describing monetization as having a ‘number of experiments in flight including (for) ads,'” Walmsley wrote.

When Apple reports its earnings next week, analysts will likely press it on its plans for AI, given the industry-wide obsession, and especially after a recent Bloomberg report that said the company was developing a ChatGPT-like language model internally.

Last month, Apple announced new iPhone keyboard software that uses the same transformers architecture as GPT, showing that it has substantial internal development of AI models. It just doesn’t like to talk about products that aren’t out on the market yet to stoke investor anticipation.

Apple is unlikely to discuss AI at length next week as its mega-cap rivals did this week. During Apple’s earnings call in May, when asked about the technology, Cook quickly moved the conversation back to the company’s products and features.

“We view AI as huge and we’ll continue weaving it in our products on a very thoughtful basis,” Cook said.