How AI changed Microsoft and the world in little over a year
Microsoft had its ‘a-ha’ moment with AI barely 18 months ago, transforming the tech titan and setting it up to become the world’s most valuable company.
Microsoft’s sprawling headquarters in Redmond, about 30 minutes’ drive northeast of Seattle, is an unlikely scene for a revolution.
Squat, university-style buildings are spread over 502 acres, surrounded by Douglas fir trees on gentle slopes which stretch into the sky. It’s a scene that hasn’t changed much since Bill Gates was in charge — known for flying economy, even for 10 years after he became a billionaire because the service in first class was not much better.
Yet it was in these not so flashy surroundings a group of Microsoft’s senior leaders realised they would not only change the company, but the world.
In August 2022 chief executive Satya Nadella and many of his executives gathered in a meeting room at Redmond for a demo of the new GPT-4 — a powerful AI platform developed by then untested start-up OpenAI, which Microsoft poured billions of dollars into.
Scott Guthrie, vice president of Microsoft Cloud and AI Group, had been talking up the technology for years but it wasn’t until this demonstration that he faced an “a-ha moment” which would set one of the world’s oldest tech companies to become the most valuable. Crucially, it gave the green light for the breakneck development of powerful new AI tools.
Guthrie and other senior leaders watched in awe as the chatbot performed tasks such as solving biology problems and passing university exams.
For Guthrie, a self-described amateur historian, he was amazed at how the AI platform could provide balanced descriptions of the likes of Alexander the Great, Napoleon or Lincoln.
“We didn’t see ChatGPT, but we saw a very early version of an interface where you could ask a question and get back a response,” Guthrie tells The Australian.
“I was asking it questions about things in history where there’s two perspectives, or multiple sides, and it wasn’t sort of binary one way or the other way. I was asking questions about, ‘was this person good?’ … trying to trip up the model. And it was giving very good answers.
“That was a little bit of an a-ha for us of realising that we’d kind of broken through on a key advancement in AI and had built a model that was capable of a high degree of sophistication.
“That led to us really changing our entire Microsoft portfolio.”
Three months later OpenAI would launch ChatGPT, amassing more than one million users in five days. 17 months later, Microsoft would surpass Apple as the world’s most valuable company, with a market capitalisation of $US3.08 trillion — catapulting it back to the cutting edge of tech innovation.
Before the GPT-4 demo, AI had been touted as changing society but had largely failed to create much hype or enter the mainstream. This was despite Guthrie saying in 2019 the technology, while disruptive, would make people’s lives better overall.
“There’s a little bit of a pre-ChatGPT, post-ChatGPT era,” Guthrie said this week.
“AI has been around for a long time, and people have used machine learning and AI, but it’s often been a case where you take a large data set and train a very narrow, specific use case to build an AI model, like fraud or risk or customer recommendations. It often required a lot of work, and it was for a very specific task.
“What’s changed has been the ability to use these large language models, like GPT, which have been trained on lots of data, and can start to look at tasks beyond the specific and look much more broadly.”
Guthrie was in Sydney this week to demonstrate Microsoft’s suite of AI products to its Australian and New Zealand customers. He spoke for more than an hour on stage to more than 2500 people at the International Convention Centre before embarking on back-to-back meetings behind closed doors.
He showed how artificial intelligence ‘copilots’ are now embedded across Microsoft’s products, including Word, PowerPoint, Teams, Excel and Outlook, and how companies could create their own AI chatbots or GPTs by tapping into Microsoft’s Azure OpenAI.
“The pace of AI innovation at Microsoft has been unlike anything in my career. All of these AI products that you’ve seen here today were created in 14 months.”
Microsoft has reportedly invested $US13bn in OpenAI over a series of rounds during the past five years, although the company has never disclosed the exact quantum of its investment. The relationship has fuelled Microsoft’s aggressive adoption and expansion of AI, allowing it overtake rivals such as Alphabet and Meta — which own Google and Facebook, respectively — and win investor confidence.
Guthrie, who is best known for wearing a red polo shirt — because it’s “lucky” — is reluctant to talk about Microsoft’s stratospheric market cap rise but says the past 18 months have been “crazy”.
“We don’t focus on the market cap internally. We’re very much focused on customers — how do we make customers successful? We know if we do that, we’ll also be successful.”
Australian use cases include Commonwealth Bank saving customers almost $250m in nine months by deploying AI to prevent scams and mistaken payments. Others such as Telstra have harnessed the technology to slash follow up customer calls by 20 per cent by producing concise summaries. This reduces the need for customers to, frustratingly, repeat information, allowing their queries to be resolved more quickly.
Cabot’s, meanwhile, has created a customer chatbot to answer questions about its wood care products.
But Microsoft’s AI-fuelled ascendancy hasn’t been without its challenges. In March last year, several tech executives, including Elon Musk — former OpenAI board member — called for a pause in the rampant development of artificial intelligence.
A six month moratorium, Musk and others including Apple co-founder Steve Wozniak said, would allow industry to ward off potential harms associated with the technology.
It was also viewed as a veiled swipe at attempting to put the handbrake on Microsoft and OpenAI. Two months later, Musk said Tesla and Twitter could partner with an AI company, potentially replicating Microsoft’s relationship with OpenAI, calling for a “third horse” in the AI race.
But the greatest threat to Microsoft’s AI strategy erupted — with CEO Satya Nadella receiving only 20 minutes warning — last November when OpenAI’s board sacked chief executive Sam Altman.
Nadella’s masterstroke had not only been nurturing a relationship with Altman during the past five years, but changing Microsoft’s culture to be more accepting of outside ideas and eradicating a ‘Hunger Games’ type mentality from within the company, in which staff were even trained to have arguments with one another.
In other words, the idea of partnering with a start-up like OpenAI would have been unthinkable before Nadella, with Microsoft believing it could build such a tech platform themselves.
A potential implosion at OpenAI by removing Altman threatened to undo the work and spark an embarrassing $US13bn setback for Microsoft.
So Nadella set to work. His ideal outcome was restoring Altman as Open AI’s CEO. But, like playing chess, his first move was to offer Altman a job at Microsoft, which increased the tech giant’s leverage. Scores of other OpenAI employees threatened to leave and join Altman at a new Microsoft AI division — a revolt which led to the restoration of Altman at OpenAI as well as Microsoft’s AI strategy.
But for Guthrie, the rapid deployment of AI boils down to one thing: customers. Although he reports directly to Nadella, he spends a lot of time travelling the world to meet Microsoft customers and work out how software engineering can help solve their problems.
“Obviously, we’re living in a time of high interest rates around the world. And so the cost of capital is also high. I think there’s a real demand for productivity, because at the end of the day, if you’re struggling to hire people, and if interest rates are high, you desperately need productivity if you’re going to grow your business,” he says.
“There’s lots of interest in AI, and I think ChatGPT is showing the world the potential.”
But a lack of skilled tech workers threatens to hamper further adoption of using AI as a way to lift output.
“That’s where a lot of our Copilot strategy at Microsoft has been based on … how do we provide tools that people can roll out and implement and be successful with, with very good return on investment, and without having to require tremendous AI expertise,” Guthrie says.
“They can use the same platform that we did to build our copilots to build their own and be successful as well. That’s been kind of the strategy, and it’s resonated pretty well, I’d say, everywhere around the world.”
As for Microsoft’s Redmond campus, it’s changing too, with a major overhaul underway which will add 17 new buildings and 6.7 million square feet of revamped work. There will even be a cricket pitch. But some things won’t change, like ‘Lake Bill’, the pond the office of its economy-flying founder once overlooked.
More Coverage
