Who Will Power the GenAI Operating System?

Silicon Valley’s 72-hour OpenAI drama was quelled last week with the return of Sam Altman as CEO to the firm he co-founded in 2015.  The board will be reconstituted with grownups who might have more business — and common — sense. Many open questions remain. A burning one: what role will Microsoft have in directing OpenAI’s strategy, leadership or governance? And another: how will OpenAI navigate between its “good for humanity” mission and the opportunity to massively commercialize its GPT platform?

Threading that needle may be among the company’s most challenging endeavors yet, as the new board is faced with balancing the views of a small number of AI doomsday extremists with the many more who see its potential to change the world, and the people living in it, in a profoundly positive way.  And to profit from the innovations that make humans and humanity better off.

It’s been reported that one of the issues leading to the ouster of Altman on November 17th was his plan to open an OpenAI app store, along with OpenAI’s involvement in a $1 billion AI-smartphone startup with former Apple design chief Jony Ive. That’s a development with implications beyond the debate over whether an app store would be good or bad for OpenAI’s mission and commercial intent.

One of the most significant implications of the digital transformation of the global economy is the inevitable shift from smartphones and apps to a distributed network of connected devices and smart ecosystems to access information and conduct commerce.  GenAI and LLMs will be important catalysts for that change. We will likely see a significant shift in that direction over the next 3-5 years.

GenAI and LLMs will be important catalysts for the inevitable shift from smartphones and apps to a distributed network of connected devices and smart ecosystems.

In a GenAI-first world, the consumer’s experience in accessing devices and information will change because consumers won’t always engage with LLMs and the apps developed using them with taps, swipes or clicks — and possibly not even with a device that’s always in their hand. That will, in turn, force a redesign of the devices and device ecosystem where “apps” are accessed by voice or another biometric trigger rather than a single device that is the app ecosystem’s centerpiece today.

A change in the business models that support how developers create and monetize “app” experiences is also inevitable as use cases and devices evolve — at a minimum, that could challenge existing operating system models. GenAI LLMs like GPT with scale that operate as apps inside of existing operating systems today will likely see the potential to break out and create their own to get more control over the customer, the experience, the data and the revenues.

Tomorrow’s Operating System for AI

For the last 16 years, since the introduction of the iPhone in 2007 and Android and the app stores in 2008, consumers have lived in the smartphone + app economy.

Today, consumers mainly use smartphones or tablets to access their apps, one at a time, on devices that run the iOS and Android operating systems. They find new apps and download them from the Apple and Google app stores.  It’s been reported that consumers have 80 apps on their phones, use 30 of them monthly and 9-10 every day.

Marketplaces and software platforms have aggregated many related activities into a single-app experience, saving consumers time and money and home-screen real estate. Ride hailing apps like Uber are now a one-stop shop for mobility and last mile delivery and can help users book travel. Amazon shoppers and Prime Members can buy food, retail and consumer products, have prescriptions refilled, book a virtual healthcare appointment, order food for delivery, and stream content from a single app. B2B commerce marketplaces like Joor help fashion buyers find each other and do business, operate virtual showrooms and trade shows and pay and be paid without hopscotching across different app experiences to do business.

Over the last 16 years, handset manufacturers have created more powerful handsets: hardware with better cameras, faster speeds, 5G and longer battery life. Phones that can make and receive calls — remember when they didn’t do that so well? — and stream video content, enable users to make video content and to play games. Operating systems have optimized the performance and security of the apps on those devices.

But the most important innovations — the ones that keep consumers buying phones  — are courtesy of the developers who create the new experiences that consumers like, download and want to use with them. Without apps, there is no consumer interest in using the device. Without consumer interest, there are no developers. Without developers, there are no apps. Without developers, apps or users, there are no sales.  Even the fanciest of hardware can’t overcome that platform dynamic.

A new GenAI operating system could power an always-on, connected and distributed economy, reimagining how consumers interact with each other and the many third parties with which they’d like to engage.

That’s what makes the idea of a GPT app store plus a new AI-powered smartphone interesting, and potentially powerful. It could become the foundation for a new GenAI operating system — one that would be optimized for many derivative GPT apps that exist today and will come in time. An operating system that could power an always-on, connected and distributed economy, reimagining how consumers interact with each other and the many third parties with which they’d like to engage. Reinventing the interface they use to do that.

But for an app store to ignite, it needs the same thing the Apple and Google Play app stores needed in 2008: a critical mass of developers and users. OpenAI claims to have a lot of both, with 100 million weekly users after only its first year and hundreds (if not thousands) of developers using GPT and its derivatives to create new apps that would ultimately live in a GPT app store.

An app store also needs devices and a vision for how consumers will interface with them to discover and use apps. It was Jony Ive and team who broke us all of our Crackberry habits with a device that wasn’t a better version of the mini PC we said we could never live without, but one that gave consumers an entirely new experience they never knew they wanted. It would take only two years for the iPhone to shatter the dominant position Blackberry once had — and another seven for Blackberry to have a zero market share.

We’ll know soon whether Jony Ive and the OpenAI team have another transformative megahit in the making, one that is compelling enough for consumers to give up their iPhones and Android devices right away or over time.  They’ve gotten much of the publicity. There are surely others operating in stealth mode or who can’t afford a publicist to share their scoop.

In the meantime, Apple, Google, Amazon and Microsoft are watching, too.

A week ago, the notion of a fractured OpenAI was a reality, as were ways to capitalize on the dysfunction. Now, with an OpenAI organization more or less reassembled, the focus intensifies on how to adapt the operating systems, developers and user to a world that will have GenAI embedded into many of the activities in which consumers now engage.

Who Might be Nervous

Apple and Google, as the only two mobile operating systems with any scale, each have a massive consumer and developer base and share of the smartphone market to leverage. The Android OS counts 3.3 billion users worldwide and 1.6 billion apps in the Google Play store and a 70% global market share. Apple has 1.96 billion iOS users and 1.46 billion apps in the App Store with a U.S. market share of 57% and 23% globally.  Both operating systems are optimized for smartphones and apps today, and both very recently have become laser-focused on adapting them for the GenAI future.

Today, Google seems to have a significant GenAI edge compared with Apple. It’s been investing in AI since the early 2000’s and has used it to power and perfect search for more than two decades. Google set up the Google Brain AI lab in 2011, acquired DeepMind in 2014, and published the pioneering paper on the use of transformers for powering neural networks. Having AI as a competency is fundamental to the success of Google’s core business.

Google introduced its GPT challenger, Bard, in March of 2023 and has invested in other LLMs since then, too.  It would be safe to assume that a plan to launch a GenAI app store, develop apps and create new devices over time is a strategic priority. It’s permissioned and tokenized identity and payments credentials, with data on consumer purchasing intent, could power a smart, AI-commerce ecosystem. Google has the developer community, users and devices to do that, at scale.

Apple seems far less well-positioned. Since the news of OpenAI’s launch of GPT, there have been non-specific announcements about its GenAI plans. Apple says it is working on its own GPT, but what’s been reported seems seems scattered, internally driven and slow. Siri isn’t that smart, and engineers have said that adapting Siri to GenAI will take time — weeks, months or even a year, depending on the complexity — given its tech architecture. A year seems like a lifetime in a GenAI world that is moving everyone forward at warp speed.

In the meantime, Apple, Google, Amazon and Microsoft are watching, too.

Apple’s GenAI and GPT complications come on top of more muted iPhone sales, political and economic challenges in China, and a future where services from apps in the app store are set to drive revenue and margin. It’s a strategy that didn’t look as risky before GenAI changed how consumers and third parties now wish to engage — but one that could force a rethink of how Apple monetizes its operating system. Currently, apps can’t have their own app stores in the Apple ecosystem. It is unclear whether Apple would allow ChatGPT to have its own app store — or if it did, how much it would try to tax those apps.  How this gets sorted may have a lot to do with Apple’s own iOS future, how regulation and court cases on Apple’s policies play out, and the sales it derives from apps and GenAI over time.

Amazon with Alexa seems more like the underdog. It was the first Big Tech player to imagine an AI-based operating system with apps (they call skills) to power a distributed network of devices activated by voice. Amazon claims that 500 million Alexa-powered devices have been sold — everything from Echo devices to smart doorbells, smart appliances, smart lights and blinds, even cars — and 130,000 Alexa apps/skills. They claim 900,000 developers are part of the Amazon ecosystem, using a fork of Android, the FireOS, to create their apps. Consumers use Alexa today to do basic chores and manage basic activities.  At the moment, though, it seems like an ecosystem that lacks energy and momentum.

Amazon has a massive amount of data on consumer purchasing behavior, along with how users engage with a voice assistant. Its own commerce ecosystem connects Prime Members to many of the activities that represent the consumer’s daily or frequent interactions, and products they want to purchase and have in the past. It needs to make Alexa smarter, more conversational, and more accessible on the network of devices that consumers have or want to own. Whether that means Amazon needs a totally different set of devices is an unknown — as is the company’s appetite for creating them, given recent cutbacks in  its hardware division.

All of this could change with the introduction of a conversational Alexa that will be available on Echo devices on an opt-in basis. The experience does not require the user to “Ask Alexa” to generate a response and can complete a variety of tasks in context. The demo I watched featured a conversation about a dinner party including making suggestions for the theme and what to cook, adding items to a shopping list, preparing an email invitation that was sent to the user’s phone, offering decorating themes, and setting timers for when to complete certain things related to the event. The flow seemed natural. If real life mimics demo performance, Alexa could become the smart personal assistant consumers say they want and have been looking for — and drive more interest from developers to be part of their ecosystem.

Microsoft,  doesn’t have a mobile operating system or control a network of mobile devices, with the exception of its tablets.  It also doesn’t have the sort of data that Google and Amazon have on consumer purchasing or a consumer-facing ecosystem with users to leverage. Yes, it has Bing — and maybe OpenAI will put the GenAI spring in Bing. But even if that happened, it isn’t clear that a better Bing would do much to enhance its consumer-facing GenAI prospects.

There’s Meta. It has created its own foundational LLM and is pouring billions into creating its own app ecosystem. It lacks its own devices and a commerce ecosystem, but with a massive global user base could emerge as a strong GenAI contender. And in real life, not the metaverse.

What’s Next

It is easy to forget that it’s been just about a year since OpenAI triggered the world’s obsession with GenAI. It’s not even a year since Microsoft announced its $10 billion investment in OpenAI that started the Big Tech GenAI race. We are — quite literally — at the beginning.

Like every other app store that’s been successful, consumers will lead the way.

GPT still hallucinates, and produces content that thinks we are still dealing with COVID when asked about current events. Google’s GenAI search is getting better with each search, but the current user interface is messy and hard to navigate. Humane AI launched a magnetic pin that uses voice and the palm of a person’s hand to access content, but I can’t imagine reading anything more than five words on my hand — can you?

But it’s evidence that innovators are thinking differently about how consumers, content and devices can interact inside of a voice-activated ecosystem that that isn’t smartphone-centric.

Like every other app store that’s been successful, consumers will lead the way. Developers will create experiences that wow them, save them time, eliminate friction and give them a really great reason to break with the habits they’ve perfected over the last 16 years. Or not. But the massive consumer uptake of ChatGPT at this early stage tells us a lot about where this might go. There’s going to be an AI-based app ecosystem — probably more than one. The question even GPT can’t answer: who is going to own these ecosystems, and who isn’t?