It was going to be different this time.
The failure of then-Microsoft CEO Steve Balmer in 2007 to recognize the disruptive power of smartphones and app stores to change the world would cost the tech giant billions in failed efforts to play catchup — remember its Nokia acquisition in 2013 for €5 billion? And billions more in missed opportunities until it finally shuttered its smartphone business in 2019.
Bing, with its 3 percent share of the search market, has largely flatlined since Microsoft used the name to rebrand its floundering search engine business in 2009. Consumers don’t even use it in markets such as the EU, where regulators have forced remedies intended to give it an advantage over Google’s Chrome. As search engines go, it’s not very good.
In 2019, Microsoft, with OpenAI, saw an opportunity to deliver a grand slam home run success with the most significant consumer-facing technology innovation of the last three decades: LLMs and Generative AI.
The $1 billion investment the tech giant made in OpenAI that year was one that CEO Satya Nadella thought could level up Microsoft’s own AI internal shortcomings, bolster its cloud business and — importantly — close the AI gap with Google. Google, which has used AI since 2001 to improve search, expanded its AI focus in 2017 with the launch of the Google AI division focused on, among other things, innovating the use of large language models and the massive data set that is Google search.
| Microsoft, with OpenAI, saw an opportunity to deliver a grand slam home run success with the most significant consumer-facing technology innovation of the last three decades: LLMs and Generative AI.
In 2021, a year after OpenAI CEO and co-founder Sam Altman changed the platform’s status from not-for-profit to a capped for-profit entity, Microsoft added $2 billion to the OpenAI kitty. But it wasn’t until its $10 billion investment in January of 2023 — two short months after OpenAI’s public launch — that Microsoft was able to change the narrative about the importance of its role in the explosive GenAI movement and the commercialization of the GPT use cases that it hoped would follow.
Microsoft’s January 2023 $10 billion investment in OpenAI seemed to have caught Google, Apple and Amazon a bit flat-footed, and each of them began to pour even more billions into rival LLMs to fast-track their progress. Google made a $2 billion investment in Anthropic in October, Amazon made a $4 billion investment in Anthropic in September, and Apple was even forced to say that it’s investing $1 billion every year in its own GenAI R&D. Siri, why didn’t you tell them to do that sooner?
Until Friday evening, November 17, 2023 it seemed that Nadella and Microsoft, with OpenAI, were sitting comfortably in GenAI catbird’s seat with a big stake in an AI platform that was leaps and bounds ahead of most others.
Friday evening’s decision by the OpenAI board to oust co-founder and CEO Sam Altman, and the subsequent resignation of co-founder and president Greg Brockman, made Microsoft’s perch wobbly. And set in motion a course of events that makes its GenAI future more, not less, cloudy.
Microsoft and OpenAI’s investors were unsuccessful at putting Humpty Dumpty back together again and a new interim CEO was named on Sunday. Emmett Shear, the former Twitch CEO, will lead the company. Altman and Brockman were given positions at Microsoft to lead a new “advanced AI team.” OpenAI engineers and team members are said to be confused, some outraged, at the whipsaw of events over the last 72 hours at the world’s most promising GenAI startup. This confusion could slow Open AI’s innovation to a crawl as its technologists mull their options and leave for stabler, richer pastures. Particularly as the fundraise for their stock payouts hangs in the balance.
How OpenAI unraveled so abruptly is a headscratcher; the details are remarkably stunning. It’s surprising that Microsoft, a nearly three-trillion-dollar company with lots of sophisticated lawyers, didn’t see the proverbial writing on the wall. Shame on me, too, for not looking more closely at the “about us” fine print at OpenAI.
As is well known by now, Microsoft poured $13 billion over four years — $10 billion of that in the last eleven months — into an entity over which it has no control, nor a seat on the Board. This document describing OpenAI’s governing structure lays it all out.
Before today and after the ousting of Altman and resignation of Brockman, the OpenAI board consisted of four people, including its third co-founder. They were tech-heavy and public policy-focused — and remarkably light on the business skills essential to monetizing and scaling Generative AI.
But none of that should be surprising given its stated mission: the nonprofit’s principal beneficiary is humanity, not OpenAI investors or customers. It’s not clear from reading the board manifesto what “humanity” as the principal beneficiary means, nor how the board would weigh lost jobs from AI against lives saved from medical advances, for example.
Like most publicly-traded companies worried about investor lawsuits, Microsoft lays out the myriad risks that harm its bottom line. Amazingly, Microsoft’s most recent 10-K fails to flag the risk that the technology foundation for its entire AI future — OpenAI — has a governance structure over which Microsoft, as a 49% equity holder, has no control nor seat at the table, and where the board is supposed to make decisions to benefit humanity without any framework other than their own. A board that gets to decide for the world what is good and what is bad, where bad seemed to be making money from the many incredible innovations that its technology would support.
| You don’t need to be a corporate governance expert to find this more than a bit odd.
Nor, apparently, a board that felt no duty of responsibility to give its largest investor more than a five-minute heads up that they were going to fire the CEO — who founded it and is widely credited with driving the advances — and potentially crater Microsoft’s business. A termination that was not for any malfeasance, which the board has publicly stated, but because they didn’t like how Altman was thinking about monetizing the OpenAI GPT asset.
You don’t need to be a corporate governance expert to find this more than a bit odd, regulatory considerations notwithstanding.
Yet it was apparently a tradeoff — a roll of the dice — that Microsoft, its board, and everyone else who put money into OpenAI was willing to accept. Provided, of course, that they got a seat at the conference table, with superstars Altman and Brockman at the helm.
Except now they’re not. That leaves Microsoft with a dysfunctional OpenAI board and an interim CEO to sort things out. And Altman and Brockman leading a team at Microsoft that presumably includes working with the firm that just fired them.
What a mess.
As I write this piece, the situation remains very much in flux despite the illusion that all is right in the Microsoft/OpenAI world. Microsoft and OpenAI investors were unsuccessful at persuading the OpenAI board to reconsider and reinstate Altman and Brockman. Altman, who has tweeted how much he loves OpenAI and its team, and his co-founder are now working for Microsoft, in charge of a new AI initiative. It is unclear what that is, how they and Microsoft will work with OpenAI, and what the structure of that working relationship is with Microsoft.
But Altman and Brockman have all the leverage because they have a track record of building and scaling LLMs and a vision for how to monetize this incredibly disruptive technology.
Microsoft, on the other hand, remains vulnerable because they have very little. Whether they get a board or board observer seat as part of the new OpenAI structure is unclear, but they have few levers to force change. They could threaten to hold back funding, but turning off the tap could also backfire, since Microsoft has embedded the OpenAI model into every one of its products — including Azure in 2021. Microsoft needs OpenAI to be successful and its LLM to be a strong and continual source of innovation to be viable contender in the GenAI world — and reap the profits from the AI-powered Azure cloud that OpenAI can provide.
The biggest risk, however, may be the stability of OpenAI as a foundational model in the near term.
| Microsoft needs OpenAI to be successful and its LLM to be a strong and continual source of innovation.
The ousting of a well-respected and brilliant founding team without warning over a lot of vague mumbo jumbo about inconsistent communications has rattled the OpenAI team. Nadella’s next big headache is very likely to trying to stop the potential OpenAI brain drain as employees flee for competing LLMs with a less capricious overlord and more stable team. Or to one of the other BigTech players with their own LLM and big Gen AI visions and a checkbook to match.
A brain drain that could soon be followed by developers who now worry about OpenAI’s long-term viability and refocus their efforts on other platforms.
It will take months, and maybe many of them, for the current destabilization of the OpenAI business to return to normal, for trust to be restored, for a new board to be created and installed, and for whatever governance changes occur to be socialized and implemented across both businesses — OpenAI’s and Microsoft.
In the meantime, AI dollars and talent may flow elsewhere.
On Friday night, Nadella did the only thing he could do: he reaffirmed his commitment to OpenAI, the remaining team and the business. On Sunday night, he did the only thing he could do: he kept Altman and Brockman in the Microsoft tent.
Wall Street is still processing the news. In afterhours trading on Friday night, the stock was down nearly three percent. We’ll see how it opens today.
Timing is everything, right?
Last week I began working on a piece about OpenAI and its potential to create the first voice-activated AI-first operating system, at scale. And with it, Microsoft’s opportunity to monetize it in a way that they failed to do with Windows and mobile in the 2010s, leapfrogging the incumbents who thought they had it locked.
A few weeks back, OpenAI announced that it was launching an app store — the same news, ironically, that seemed to rattle the OpenAI board and set Friday’s disastrous series of events in motion. For an app store to ignite, it needs users and developers, and OpenAI has a lot of both. The company claims 100 million weekly users after only its first year. There are hundreds (if not thousands) of developers using GPT and its derivatives to create new apps that would ultimately live in the app store.
An app store also needs devices and distribution — and an innovative way to engage with its content. News broke in September that Jony Ive was in talks with OpenAI to create an AI-powered smartphone with $1 billion in funding and support from OpenAI’s Altman — and I am sure other connected devices after that. Around the same time, OpenAI announced GPT was voice-activated in five languages and with five different voices.
| It’s early days, but one can imagine that it will only get better.
I also began hearing more and more about people talking to GPT in a very different way than they talk to Alexa or Google or Siri. As in having a conversation with GPT, in the same way they’d have a conversation with a friend or acquaintance. They described those conversations as smart, even intuitive, relevant, even a little serious. If you are curious about how one of these conversations sounds, scroll through this article to the short demo. Spoiler alert, it’s early days, but one can imagine that it will only get better.
Well, I guess now that all depends.
The idea of a connected economy, powered by an always-on, ambient network of connected endpoints accessed by voice and powered by AI, is something that I first wrote about in October of 2016 – the year that Google launched Allo, five years after Apple launched Siri and two years after Alexa introduced herself via the Echo device that first sat on our kitchen counters and cracked bad jokes. It’s a concept, one of seven that I identified in 2019, as a trendline that would define the decade of the 2020s and influence the digital transformation of nearly every sector in some form or another.
The idea of a smart, powerful, intuitive voice assistant is something that, according to PYMNTS Intelligence research, 60 percent of consumers say they want, and nearly 30 percent of consumers say they’d pay to have. That’s because, today, nearly two thirds of consumers use voice to complete simple tasks and believe that voice is the easiest way to manage complex interactions — those unexpected downstream effects that happen when flights are late or kids miss the bus, or emergencies happen.
|A smart, powerful, intuitive voice assistant is something that 60 percent of consumers say they want.
Big Tech with their AI-powered voice-activated assistants have been around since Siri in 2011 on the iPhone, Alexa in 2014 with the Echo, and Google Assistant since 2016. Yet none of them have managed to turn voice into a significant ecosystem that can drive revenue. And when compared to OpenAI and GPT, those voice assistants seem not too smart. Even Amazon with Alexa, which I thought was a slam dunk to be that third operating system powered by voice, is redirecting its efforts to LLM in an effort to reclaim its smart voice assistant magic and a shot at an AI-powered commerce ecosystem activated by voice. But all of these early voice platforms are to ChatGPT what the Palm Pilot was to the iPhone.
The immediate crisis at OpenAI may have been addressed with Altman’s and Brockman’s new position at Microsoft, but there is a lot we still don’t know about that arrangement. However, it tells me that there must have been something amiss at Microsoft to have made a multi-billion dollar bet on an entity with such an unstable governance structure over which they had virtually no control.
And that these are still early days in the new AI revolution, with lots of unknowns about who will come out on top in a few years. And how these powerful and smart new ecosystems will evolve.
Over the last seventy-two hours, Nadella’s star at Microsoft might have just gotten a little dimmer.
Google’s and Amazon’s just got a little brighter. Maybe even Apple, too.
OpenAI’s crisis is now their even bigger GenAI opportunity.