I would argue that we are indeed standing on the summit of peak “phone as hardware”: While Apple’s newest iPhone offers some impressive hardware features, it does not represent the beginning of the next 10 years of the smartphone, as Apple claims.

The better question, then, is where do phones go from here?

To understand the future of phones, it helps to look at the history of phone innovation. We have seen this movie before. When focal dimensions of innovation change, incumbents often get left behind. More specifically, as we shift from hardware-based innovation to differentiation around AI-driven technologies, market leaders like Apple should be on high alert.

Innovation in technology product categories tends to proceed along a specific dimension—a “vector of differentiation.”

Players pursue innovation along a vector of differentiation until the vector runs out of steam. This happens for two reasons: limits to innovation along the vector of focus and the ability of competitors to catch up with market leaders. When that happens, the focus of innovation shifts to a different vector and new market leaders emerge. We have seen this pattern several times in mobile phone innovation over the past three decades.

The physical dimensions of the phone constituted the first vector of mobile phone differentiation. Phones were big, clunky bricks of technology until the small, sleek StarTac arrived in 1996, establishing Motorola as a market leader, soon joined by Nokia. Innovation continued along this dimension, with phones getting smaller and smaller.

With the advent of the Blackberry, Palm device, and others in the mid- to late-1990s, the emphasis shifted to data capabilities, especially email and text messaging. Consumers loved those button-based keyboards, and Blackberry snatched the crown of market leadership from Nokia and Motorola.

Fast-forward to 2007, when the vector of differentiation shifted once again with the debut of Apple’s iPhone. Now it was about display and apps. In a revolutionary move, Apple eliminated the physical keyboard to maximize real estate for glass. It also created the App Store, a thriving ecosystem of applications that contributed to Apple’s breathtaking market success. RIM, the maker of Blackberry, was never able to make the transition to a media-centric phone and slipped into oblivion.

Innovation along the display and media vector led to larger and larger phones, ironically the reverse of the “make it smaller” innovation in the earliest mobile phones. Recent models like Samsung’s Galaxy S8 Plus have pushed the limit of this dimension—a return to brick-worthy phone surface area—and we have seen the rise of bezel-less phones (those with no borders around the screen) and those with more powerful processors, better displays, and more powerful cameras. Samsung and Apple have led innovation in this space.

Now, the vector of differentiation is shifting yet again, away from hardware altogether. We are on the verge of a major shift in the phone and device space, from hardware as the focus to artificial intelligence (AI) and AI-based software and agents.

This means nothing short of redefinition of the personal electronics that matter most to us. As AI-driven phones like Google’s Pixel 2 and virtual agents like Amazon Echo proliferate, smart devices that understand and interact with us and offer a virtual and/or augmented reality will become a larger part of our environment. Today’s smartphones will likely recede into the background.

As we have seen, when the vector of differentiation shifts, market leaders tend to fall by the wayside. In the brave new world of AI, Google and Amazon have the clear edge over Apple. Consider Google’s Pixel 2 phone: Driven by AI-based technology, it offers unprecedented photo-enhancement features and deeper hardware-software integration, such as real-time language translation when used with Google’s special headphones.

Similarly, the Amazon Echo enables natural conversations through the Alexa virtual agent. Next-generation devices will use AI and deep learning to recognize our voices, faces, and emotions. We will move from touch to touchless interactions and we will move from software apps to AI-powered skills. Just like the App Store, Amazon has created an Alexa skills store for third parties to offer skills that enable the Echo to do everything from set your kitchen temperature to play Jeopardy with you. “There’s a skill for that” will apply to more and more consumer needs.

The shifting vector of differentiation to AI and agents does not bode well for Apple.

The advent of Amazon’s skill store and similar innovations speak to the need to create an AI-rich ecosystem where hardware, software, and third-party contributors work in concert to enhance consumer experience across life domains. Amazon is making rapid progress along this vector of differentiation, as are Google (with its TensorFlow open-source platform for AI apps) and even Microsoft.

Sheets of glass are simply no longer the most fertile ground for innovation. That means Apple urgently needs to shift its focus and investment to AI-driven technologies, as part of a broader effort to create the kind of ecosystem Amazon and Google are building quickly. However, Apple is falling behind in the AI race, as it remains a hardware company at its core and it has not embraced the open-source and collaborative approach that Google and Amazon are pioneering in AI.

The history of mobile phones suggests that when vectors of differentiation shift, so does market leadership. Apple has only to look at former dominant businesses like Motorola, Nokia, and Blackberry to understand how quickly a leader can fall from the peak in this market, and do its best to avert this outcome.

Mohanbir Sawhney is the McCormick Foundation professor of technology at the Kellogg School of Management at Northwestern University. He has no investments of the companies mentioned in this article.