Microsoft CEO Nadella on AI Wave and Tech in 2024

00:00

Thank you for joining us. We talked about maybe moving this event

00:03

to the Caribbean next year. You like some cricket and have some good

00:08

weather. That sounds great.

00:10

If you want to lob a question up here for satire.

00:13

Please scan the QR code and you could submit a question.

00:16

But I want to start with a question that I know you're just going to love, which

00:21

is acknowledging this remarkable moment at the end of last week where the market

00:26

capitalization of Microsoft surpassed Apple and Microsoft became the most

00:31

valuable public company in the world. And it just reminded me of a moment in

00:36

2010 when Apple's market cap exceeded Microsoft's and Steve Jobs sent an email

00:42

to employees and and he said stocks go up and down.

00:45

But he wanted to recognize an extraordinary moment.

00:47

And he said, and remember, Apple is only as good as our next amazing product.

00:52

I wanted to know, and I suspect the answer is no, whether you acknowledged

00:56

to Microsoft employees this moment and if you had, how you would finish that

01:00

sentence. Microsoft is only as good as what

01:05

I mean, you got jobs added, right? The I think if I had to sort of pick up,

01:12

we are only as good as our ability to execute, prosecute our mission because

01:18

in some sense, there's no God given right for companies to even exist

01:23

forever. Right?

01:24

They have to serve a social purpose. And our mission is to empower every

01:32

person and every organization on the planet to achieve more.

01:34

And I feel that if we are building products, services that speak to that

01:39

mission in the sense it's relevant to people and organizations, that's, I

01:43

think, pretty unique about Microsoft. We think about people as first class,

01:46

but we think about institutions and organizations.

01:49

People build to our class them as also first class.

01:53

And as long as we are building things for that, then I think we have a right

01:56

to exist. Did you take even a moment personally

01:59

here with employees to recognize, you know,

02:04

in my whatever, 32 years at Microsoft, we have gone up and down.

02:09

And that's why I think the most important thing to focus on is and this

02:12

is, after all, a lagging indicator, not a leading indicator.

02:15

So the last thing you want to do is to fixate on the stock price, which we know

02:19

means nothing in terms of what happens tomorrow, especially in our industry,

02:23

which really has no franchise value, quite frankly.

02:25

I mean, the problem in some sense is for all of us, whether we can bet it all on

02:31

what comes next, which means it's very, very hard.

02:35

Right. And speaking of what comes next, you

02:37

guys have been fairly aggressive and out in front on adding AI features and

02:41

capabilities to your products. And yesterday you made an announcement

02:45

where you're expanding the rollout of the co-pilot tool,

02:51

fueled partly by your partnership with Openai in the Microsoft products like

02:55

Outlook and Word and Excel. Talk a little bit about that and how

02:59

widely do you expect those tools to be used?

03:02

Yeah, I mean, if I step back, if you sort of look at

03:09

what has happened even in the last 16 months, right.

03:13

You have to go back to November of 22 when tragedy first came out.

03:17

I think that was the moment which I like to describe as the the mosaic like

03:22

moment. In fact, interestingly enough, it was

03:23

November of 93. A year after I joined Microsoft when

03:27

Mosaic first came out. And I think that's the first product

03:32

that we all could relate to and get a real sense of what this generation of I

03:39

can do in our lives. But for me, maybe the product that

03:44

really helped crystallize the potential was GitHub copilot, which probably came

03:49

six, eight months before that. And especially when we, you know, we

03:54

scaled from GPT 3 to 3 five. That's around the time when we felt that

04:00

if we can take something like software development, which is let's call it the

04:05

most elite knowledge work there is and have a tool that allows a software

04:10

developer in fact bring joy back to software development, keep them in flow,

04:14

get them to finish tasks. That to me solidified.

04:18

In fact, the idea that you can have a copilot for

04:23

pretty much every human task. And so we are been on that journey and

04:29

we anchor on to real things. The one thing the breakthrough is the

04:32

user interface. In fact, 70 years of computing history

04:35

has always been about can you build that most intuitive user experience?

04:38

That's kind of it led to graphical interfaces or the, you know, the

04:43

multi-touch phone or what have you. But now with natural language, you

04:47

ultimately in some sense have arrived at that point where it's not about us

04:50

understanding computers, but computers understanding us.

04:54

So that's one major breakthrough. The other breakthrough is

04:58

we now have a new reasoning engine. Which is a neutral reasoning engine,

05:01

because the other 70 year history of computing was how we digitized people,

05:06

places and things and try to make sense of it, reason about it.

05:10

And so we now have a new capability. So you put these two things together, a

05:13

new user interface that's much more intuitive,

05:17

you know, grounded in natural language, multimodal multitouch and multi-domain

05:22

and the reasoning engine, pretty much every software category.

05:24

What is productivity? What is an operating system, what's a

05:27

browser? They all in some sense collapse.

05:31

And so that's why I do ask copilot, just as maybe in the past we were known as an

05:35

office company or a Windows company or a cloud company.

05:39

I think going forward we will be we have a co-pilot, we have a co-pilot stack in

05:43

Azure, which is all the APIs. And that's sort of what our core focus

05:48

is right now. Tell us

05:50

let's talk a little bit about the relationship with Open API and how we

05:54

should understand it. Copilot is powered by open air.

05:59

We saw some instability in the relationship back in November, which

06:06

you seem to have now come through. Is is Microsoft outsourcing what you're

06:12

describing as a core capability going forward?

06:14

Yeah. So I think

06:17

if you sort of step back, in fact, it's probably helpful to understand I grew up

06:23

in a microsoft which sort of had these massive

06:28

partnerships. The first partnership that at least I

06:31

joined was around Intel and Microsoft. I don't think Windows would have existed

06:37

without Intel, and Intel wouldn't have had the success without Windows.

06:41

Subsequently, in fact, it's interesting. I worked on a SQL Server product with

06:46

SAP. In fact, I don't think SQL Server

06:48

database would have existed without SVP. SAP success.

06:52

Often being able to support SQL Server also helped them a lot.

06:56

And so in the same way I think of Openai and Microsoft.

06:59

So I'm used to constructing. In fact, a lot of people talk about

07:03

organic development, which of course is the core People talk about M&A, but sort

07:07

of not as much as talked about how much enterprise value gets created by

07:11

partnering effectively. So that's the spirit with which I think

07:16

about Openai is so there's a whole lot we do.

07:19

So when you say outsourcing, who's outsourcing, what to whom is the real

07:23

question, right? So we build the compute.

07:26

They then use the compute to do the training.

07:29

We then take that, put it into products. And so in some sense it's a partnership

07:33

that is based on each of us really reinforcing what everyone each other

07:38

does and then ultimately being competitive in the marketplace.

07:42

There are there's room for I call it horizontal specialization.

07:46

There is room for vertical specialization.

07:48

Sometimes some business models are in vogue.

07:51

I'm a big believer in horizontal specialization, especially if you can

07:56

vertically integrate everything. You have to worry about being over

08:00

indexed and overreliant on a company. A partner whose ultimate goals and

08:06

mission might be different from Microsoft's look.

08:09

I mean, you don't go into any partnership.

08:11

First of all, there is independence in a partnership.

08:13

They are two different companies, answerable to two sets of different

08:17

stakeholders with different interests. So therefore you have to then create a

08:21

commercial partnership in it that is mutually beneficial.

08:24

So that's why I think partnerships where you enter into partnerships where one

08:28

side is trying to take advantage of the other is not long term stable.

08:31

But if two partners can and that's sort of why I go back to the history of

08:36

enterprise value that was created with partnerships that at least I've been

08:40

involved in across my career at Microsoft.

08:42

So, yes, I think you have to sort of I feel very, very good about the construct

08:47

we have. I feel at the same time very capable of

08:52

controlling our own destiny. So it's not like that.

08:55

We are single threaded even today on Azure and this is not about even open

09:00

the I. It's all about reflection of what our

09:02

customers want. Every customer who comes to Azure, for

09:05

example, in fact our own products is not about one model.

09:08

We care deeply about having the best frontier model, which happens to be, for

09:12

example, today, good for. But we also have mixed strong as a

09:17

model, as a service inside of Azure. We use our llama in places.

09:21

We have PI, which is the best solution from Microsoft.

09:26

So there is going to be diversity of capability and models that we will have

09:30

that we will invest in, but we will partner very, very deeply with opening

09:34

on what is the right operating model for a company like Open.

09:39

I mean, currently it's a cap for profit company owned by a nonprofit.

09:43

A very unorthodox arrangement probably contributed to some of the drama and

09:48

instability in November. Have they figured it out yet?

09:51

Are you comfortable now that you've got a partner with a stable operating model?

09:56

You're talking to Sam later? I am, and I will ask him that as well.

10:00

He's for your opinion. He he needs to answer that question.

10:04

And his board, obviously, and I'm sure they're working through it.

10:07

Look, I. I always say this, which is we invested,

10:11

we partnered when they were whatever they were and whatever they are today.

10:15

Right. In terms of being a capped profit,

10:18

nonprofit, what have you. So I'm comfortable.

10:21

I have no issues with any structure. What we just want is good stability.

10:27

And and as I said, you don't even need look, I'm not even interested in a board

10:31

seat or we don't need we're definitely don't have control.

10:34

We have no we just want to have a good commercial partnership and we want to be

10:38

investors in the entity in even the way they're structured.

10:42

So what I would like is good governance and real stability.

10:46

That's it. You have a a board observer non-voting.

10:50

We were I was joking backstage. It feels like having somebody at the

10:53

back on the back of the bus without a seatbelt.

10:56

Yeah. I mean, it

10:58

is so it doesn't matter to me, right? I mean, the board seat is not the

11:03

critical path at all for us. What is most important, as I said, is we

11:07

just want a board that cares about open the eye on the open side, and that's all

11:12

we care. I mean, that's all we can ask for.

11:15

And we just want stability in the partnership so that we can then continue

11:20

to invest in it. That's it.

11:21

I'm always a little cautious in these sort of hype cycles in in Silicon Valley

11:27

that certain technologies are being kind of overhyped over promised.

11:31

You know, so far, for example, with the integration of GPT into Bing, have the

11:37

results met your expectations or has it or has it been over?

11:40

I don't know if you sort of take the combination of sort of chat app usage

11:46

and even being usage or capacity usage. I think at this point you have to ask

11:51

yourself your own user habit, right? How many times do you go, for example, I

11:55

think the real question here is the largest software business there is is

11:59

search, as we know. And the question is, is that stable?

12:04

I think like all big things, it'll take time.

12:07

But I think at this point, the idea that you go to one of these agents that gets

12:13

you to the answer quicker is pretty clear.

12:16

It doesn't mean that such as we know for today, I mean, Google obviously is super

12:20

strong. They have the defaults on Apple.

12:21

They have their own Android. Default Chrome on Windows is the largest

12:25

browser share and what have you. So it's a it's a structurally a

12:28

fantastic position that they have. But that said, I think such as we know

12:33

if it is going to change and the web as we know if it is going to change.

12:36

And so we have a real opportunity, whether it's with Bing, but also even

12:40

independent of building, for example. That's why I think about copilot is a

12:43

real product, right? To me, the the relationship we all will

12:47

have with computers is going to be now with an agent

12:52

which will be on all your computers. And to me, that I think is going to be

12:57

the defining category of this next generation.

13:01

I want to change gears and ask you about this year's elections.

13:05

I think 70 democratic elections around the world, more than half of people on

13:10

Earth will have access to vote in an election.

13:13

Donald Trump yesterday won the Iowa caucuses.

13:16

I know it's sort of fashionable to say that these are the most important

13:20

elections of our lifetime. I'm curious what you think is at stake

13:25

for Microsoft, particularly with the US election and for the safe stewardship of

13:30

A.I.? Does this feel momentous to you?

13:34

I mean, you know, if I step back from me, I

13:39

mean, the one thing that to your core question as a multinational company, you

13:45

know, the one thing that I'm always grounded on is we are also an American

13:49

company. So the state of the United States,

13:53

politically, economically, socially and its stature in the world across those

13:59

dimensions matters a lot. So I think that because that's your

14:04

passport, when we show up anywhere, at the end of the day, we are an American

14:07

company. And to the degree to which America has

14:11

the relationships, they welcome, that they welcome the companies that are born

14:17

in the United States. So that's, I think, the fundamental

14:19

thing in that context, obviously, in our democratic process, having that process

14:24

be well administered, that people have trust in it, I think is super important.

14:30

So we've always been through the years we have done a lot around what does it

14:34

mean to support the democratic process, whether it's the support for the

14:39

parties, it's the support for the election process itself.

14:42

Of course, the thing I it's not like this is the first election where

14:47

disinformation or misinformation and election interference is going to be a

14:52

real challenge that we all have to tackle.

14:55

We as a company have to do our best work right, whether in the.

14:58

Context. Me, I we have lots of initiatives around

15:01

content IDs and other things that will then help us, you know, at least vouch

15:06

for the veracity of any content out there.

15:09

And that's, I think, the work that we need to do.

15:11

But ultimately in the democratic process, essentially ensuring the

15:16

integrity of elections is one of the fundamental challenges we have to face

15:20

up to whichever administration takes over.

15:27

They will probably make it more difficult for Microsoft to do business

15:30

in China. How how are you thinking about the

15:33

technology you develop in China and how long do you think you'll be able to

15:37

employ engineers working on technologies like A.I.

15:40

in China? So a couple of things.

15:42

I mean, China is not a large business for Microsoft.

15:45

In fact, we if you sort of look at a of panel, even I think this is one of those

15:49

things that is probably not as well understood is it's fundamentally we do

15:55

business in China in order to support other multinationals going into China.

16:00

So this is the German automakers or American automakers or, you know, CPG

16:05

firms or what have you around the world who depend on having commonality of

16:09

infrastructure between the rest of the world and China depend on Microsoft.

16:14

And that's why in some sense, we have to be in China in order to support our

16:17

customers. And the same is true of Chinese

16:19

multinationals going outside of China as well.

16:22

And so that's really our business. So there is not a domestic business that

16:26

we have that to speak of in terms of human capital.

16:31

And the way I look at it and say is the greatness of the United States has been

16:36

all about being able to attract the best talent.

16:39

We definitely want people to come to the United States.

16:42

We want them to work in the United States, but we also want to tap into

16:46

human capital around the world to be able to contribute to what is

16:49

essentially American intellectual property at the end of the day.

16:52

Right now, when I look at some of the machine learning papers and so on,

16:56

there's as much being written in Mandarin as it is written in English.

17:00

And so to the degree to which we believe that somehow knowledge creation doesn't

17:05

have boundaries and in fact, the worst mistake we could be making is to somehow

17:09

shut ourselves up. I mean, the lesson of history, at least

17:12

as I read it, is that the worst mistake any civilization, any society can make

17:16

is to somehow shut yourself off from knowledge that's being created

17:20

elsewhere. So to us, if that are great, there's

17:23

great talent in China that wants to work at Microsoft contributing to essentially

17:28

on American companies intellectual property, we welcome it.

17:31

We at the same time, we are very clear about sort of, hey, this is our

17:34

intellectual property. We are definitely not going to have any

17:37

collaborations that are not in alignment with my aim in the United States's

17:43

national interest. Okay.

17:45

I want to go back to Microsoft's investment in Open A.I., which is being

17:48

scrutinized now by the EU and others. It feels to me like to the extent that

17:54

there was a holiday for Microsoft in terms of antitrust scrutiny, the holiday

17:58

has ended. But do you feel like Microsoft going

18:01

forward will now be more limited in not only the kind of acquisitions it can do,

18:06

but now the kinds of investments that it can make?

18:09

Look, I mean, I think it's inevitable that, you know, regulators everywhere,

18:14

antitrust folks are going to look at, you know, whatever a company of our size

18:18

and scale does. And so that's why I think in this

18:22

context, I all I say is if we want competition, it I against, you know,

18:31

some of the players who are completely vertically integrated.

18:34

I think partnerships is one avenue of, in fact having competition.

18:38

So I'm sure the regulators will look at it and say, is this a pro-competition

18:44

partnership or not? And to me, I think it's a no brainer.

18:48

I mean, if you don't even like to think about this, right, if Microsoft had not

18:51

taken the highly risky bet, I mean, this is not all conventional wisdom, but when

18:56

we made those investments in B, when we backed even open A, I went

19:02

all in on a particular form of compute that led to all of these breakthroughs.

19:09

You know, it would have not been fun. We wouldn't have had what we have.

19:12

And more importantly, you know, the incumbents would have been the winners,

19:15

right? It took 21 months for Microsoft's

19:19

acquisition of Activision to be approved.

19:21

And then almost I think it was a couple of weeks later, Adobe's attempted

19:25

acquisition of Figma was, was or was going to be rejected.

19:28

They walked away from it. Is the era of big deals over.

19:34

I think this is where perhaps looking at

19:40

I've been I talked to any antitrust folks.

19:44

I always ask them, have they talked to venture capitalists?

19:48

Because I feel the best way to make sense is not by

19:52

size or any one company and things. What is a big deal?

19:56

What is a small deal? There is conventional.

19:59

There are no fly zones. At every week you meet will tell you

20:02

where the no fly zones. All you gotta do is track because at the

20:05

end of the day, you want new entrants, right?

20:07

That's the core of making sure you have vibrant competition is that there is

20:12

room for new entry and new innovation, and that's where venture money is

20:17

focused on. And so by category you look at that,

20:21

you'll take productivity, right? Sort of a place where we have some great

20:24

success in the past. But think about the number of new

20:28

companies that have been born even in the last ten, 15 years.

20:31

Right. Zoom Slack notion.

20:33

I mean, you wake up and there's a big company.

20:36

Why is that? That is because there's a real

20:38

opportunity for new entrants. Then you say, okay, how many new search

20:41

engines have been born? None.

20:45

And so to some degree, that I think is the analysis.

20:48

The unit of analysis is as simple as looking at where is venture money going

20:53

in which categories and gave me that was up to, you know, you know, we beat out

20:57

gaming. We love gaming.

20:58

In fact, Flight Simulator was created before even windows, but we were number

21:03

three, Number four. And with now Activision, I think we have

21:07

a chance of being a good publisher, quite frankly, on Sony and Nintendo and

21:11

PCs and Xbox. And so we are we are excited about that

21:15

acquisition closing. I'm glad we got it through, but I think

21:19

each category by category is the view. I want to quickly ask an audience

21:23

question and perhaps US audience member went to CBS where there were a lot of I,

21:28

of course was a theme and there were a lot of AI powered gadgets that were

21:32

displayed. Do you feel like we're coming to the end

21:36

of the smartphone era? And what is an AI successor to the

21:41

smartphone look like, and does Microsoft eventually play in that category?

21:44

Yeah, I mean, I think about CBS this year was very interesting.

21:49

Obviously, I thought the the the demo of the rabbit OS and the device was

21:55

fantastic. I think I must say after Jobs

22:00

sort of launch of iPhone, probably one of the most impressive presentations

22:03

I've seen of capturing the the vision of what is possible going forward for what

22:10

is an agent centric operating system and interface.

22:15

And I think that's what everybody's going seeking

22:19

what device will make it and so on. It's unclear, but I think it's very

22:23

clear that computing I go back to that droid, if you have a breakthrough in

22:27

natural interface where this idea that you have to go one app at a time and all

22:33

of the cognitive load is with you as a human does seem like that can be a real

22:39

breakthrough. Because, you know, in the past when we

22:42

had the first generation, whether it was Cortana or Alexa or Siri or what have

22:46

you, it was just not it was too brittle where we didn't have these transformers,

22:52

these large language models, whereas now we have, I think, the tech to go and

22:58

come up with a new app model. And once you have a new interface and a

23:01

new app model, I think new hardware is also possible.

23:04

And is that an opportunity for Microsoft or are you moving away from hardware?

23:07

I mean, look, I mean, if always it's an opportunity for us.

23:12

And so, I mean, we make hardware today, we have surface devices, we make mixed

23:16

reality devices. The biggest hardware business we happen

23:19

to have is our cloud. We stream from the cloud.

23:23

So therefore, I think you'll see us exercise the full stack of it.

23:28

Last question, since we're out of time on a topic that I know is near and dear

23:32

to your heart, which is cricket, you're sponsoring a league here in the

23:36

US. How do you convince Americans who can

23:40

barely get through a soccer game to fall in love with cricket?

23:44

Well, I mean, you look, you know, there's room in the United States for

23:48

all kinds of things. But quite honestly, I mean, to us,

23:54

in fact, the interesting since you brought up cricket actually did the next

23:57

World Cup of T20is in the United States and looking forward to India and

24:00

Pakistan. There's an India-Pakistan game in New

24:02

York. Will you be going?

24:03

I hope so. If I can get tickets, that is, I think

24:06

you can probably swing a ticket. But if you look, it's you know, it's

24:11

it's a it's a sport that obviously for our South Asians, it's a big deal.

24:15

It's a religion for us. And so we're obsessed about it.

24:17

And I love the sport and I'm glad that it's not being played even in the United

24:21

States. In fact, originally the first test

24:23

match, I think was in US Canada. But after the American Revolution, I

24:27

think the one I think the US won and but for the US we rejected cricket as a

24:34

British sport after the American Revolution, but we can bring it back.

24:38

Satya, thank you very much. Thank you so much.