Earlier this week, Microsoft Corp. named artificial intelligence pioneer Mustafa Suleyman chief of its consumer AI business and hired most of the staff from his Inflection AI startup. A day before, Bloomberg reported that Alphabet Inc.'s Google was in talks to license its Gemini AI engine to Apple Inc.
The moves suggest that despite pouring billions of dollars into partnerships, investments and product development, Microsoft and Google are struggling to figure out how to capitalize on generative artificial intelligence. Neither company is moving fast enough to field consumer products that generate revenue and grab market share, and, despite their size and power, they remain vulnerable to being disrupted. Even as engineers labor to perfect the large language models undergirding the technology, the companies are forming alliances and scouring the world for talent and promising startups.
Forging a leading position in generative AI requires each of the tech giants to assemble various ingredients"-computing power, top-of-the-line AI models, trustworthy and easy-to-use products and ways of getting them to people. None of the tech giants has all of the ingredients. Google, once the pioneer in large language models, keeps releasing products with worrying errors and biases. Microsoft got a head start with exclusive access to many of OpenAI's ground-breaking models, but has never been skilled at building exciting consumer products besides video games. Apple is years behind in AI and planning to announce a comeback strategy in June, but it does have the iPhone and app store.
"Even large tech companies and cloud providers, with their vast resources, cannot innovate the entire generative AI ecosystem single handedly," said Ido Caspi, a research analyst at Global X ETFs. "Companies must constantly be on the lookout for the best talent and technology to fill gaps in their portfolios."
Microsoft Chief Executive Officer Satya Nadella isn't satisfied with his team's efforts to create consumer products, according to two people familiar with his thinking. Over the past year, the company has baked AI into the Bing search engine, Windows, Office and other products"-creating various digital assistants under the new Copilot brand. Yet the new Bing has made few gains against search market leader Google, and the other products are works in progress. Though OpenAI has given Microsoft a first mover advantage, the startup is focused on the underlying technology not developing products for the software giant.
That's why Nadella recruited Suleyman to be executive vice president and CEO of Microsoft AI, along with Inflection AI co-founder KarA(c)n Simonyan, who will become chief scientist of the new unit. Nadella wanted someone to unify and oversee product development and believes he found that person in Suleyman, with whom he bonded during conversations about how AI could be tailored to individual users.
"Really all we're trying to do is craft a true end-to-end product experience, so the user feels that there is a seamless, fluid, human-like conversational interaction," Suleyman said in an interview Monday. He likened the work to sculpting and said "that art requires you to know the real sweet spot of when a piece of technology is ready and how to dress the experience so that it has an essence and a character which is familiar and accessible and trusted."
Nadella also knows that because AI will completely transform the industry, even Microsoft's flagship products aren't immune to disruption"-whether by a peer like Google or a startup that hasn't even been invented yet. "These businesses are big and they're all going to get re-litigated, right?" Nadella said in an interview. "So even the incumbents can't take it for granted."
Nadella's faith in Suleyman isn't universally shared. During his time at Google's DeepMind, Suleyman wasn't focused on consumer products. At Inflection AI, he oversaw the development of a chatbot called Pi that was designed to interact with users in a more human and supportive fashion. But despite attracting a million active daily users, the startup never found a business case. Suleyman was also accused of treating DeepMind employees poorly. He has acknowledged making mistakes and says he learned from the experience. Still, it's unclear how well he'll mesh with Microsoft's existing team.
The proposed partnership between Google and Apple is still being worked out so there are few details of how exactly it would work. But assuming the deal comes to fruition, it could help each company with their AI efforts.
Since early last year, Apple has been testing its own large language model codenamed Ajax, according to people familiar with the situation. Some employees also have been trying out a basic chatbot dubbed Apple GPT. But Apple's technology remains inferior to tools from Google and other rivals, according to the people, making a partnership look like the better option. Apple could ultimately team up with another company, including OpenAI, or several.
Google has raced to weave AI into its products, but the company has been dogged by concerns that it's been too slow to capitalize on the shift in the market and has been playing catch up to Microsoft. Last month, the company pulled an image generator amid a flurry of criticism over inaccurate historical depictions of race. A deal with Apple would be Google's highest-profile partnership for Gemini to date and could be a major boon for the company's AI efforts. Apple has more than 2 billion devices in active use that could potentially become home to Google Gemini later this year.
Pushing Google's AI tools to as many devices as possible could help make them the ones users turn to reflexively, as they do with the company's search engine, creating a virtuous cycle whereby the programs become sharper and sharper with additional user data. Earlier this year, Google struck a deal with Samsung Electronics Co. to incorporate its Gemini technology into the South Korean company's phones, offering a glimpse of how Google could leverage its relationships with Android device manufacturers in the age of generative AI.
Industry observers expect tech companies to continue collaborating and investing in startups to augment their services and avoid being disrupted. "They need each other, especially in a space like AI which requires engineering expertise and lots of computing power," said Dan Wang, a Columbia Business School professor who teaches innovation and entrepreneurship. "It's a sensible way for the large tech companies to spread out the risk." No one, he added, "has really found a killer application just yet that people are willing to pay for."
