0:00

When you arrive in the all new Toyota

0:02

Crown, every entrance becomes a grand

0:05

one. With an available

0:07

hybrid max powertrain that says You

0:09

always arrive fashionably on time. Style

0:12

that says emphasis on the fashionably and

0:16

presence that says You

0:18

speak softly and everyone listens.

0:21

Introducing the Toyota crown, the car

0:23

that says so

0:23

much. Toyota.

0:27

Let's go places. Hey,

0:31

y'all. This is Josh Mucci, host

0:33

of the pitch. A Voxmedia podcast

0:36

where real entrepreneurs pitch real

0:38

tech investors for real money.

0:40

In this season, we're letting listeners like

0:42

you invest in the startups

0:45

on our show. Hear the next billion

0:47

dollar idea way before

0:49

anyone else. That's gonna of

0:51

sexy. We'll make the money, honey. Follow

0:53

the pitch for our

0:54

best season yet. New episodes

0:57

start dropping weekly on February eighth.

1:03

Hello and welcome to Dakota. I'm the I Patel

1:05

editor in chief of the verge, and Dakota is my

1:07

show about big ideas and other acrons.

1:09

We have very special episode today. I'm not actually

1:11

in the studio. I'm talking to Microsoft

1:14

CEO, Sachin on the Microsoft

1:16

campus in Red where just a few

1:18

hours ago, the company announced the next

1:20

version of the Bing search engine would

1:22

be powered by OpenAI. That's

1:25

the company that makes chat GPT. You'll

1:27

be able to chat with Bing, just like you can

1:29

chat with ChatXBT. There's also

1:31

a new version of the Edge web browser with

1:34

OpenAI ChatX. A window that rides

1:36

along with your web pages to help you understand

1:38

it. Novella has been very bullish in

1:40

AI. He's previously talked about AI being the next

1:42

major computing platform And today, he's

1:45

openly hoping to use that tech

1:47

to take on Google search. That's

1:49

right. While talking about how much AI can

1:51

change the game, such as said, most

1:53

importantly, we want to have a lot of

1:55

fun innovating again in search because

1:58

it's high time. That's a pretty

2:00

direct shot at Google. But the presentation was

2:02

in-depth. It showed how OpenAI running and being

2:04

an edge could radically increase your productivity,

2:06

make travel itineraries, post to LinkedIn,

2:09

and even rewrite code to work in different

2:11

programming language. After presentation, I was

2:13

able to get some time with Sasha and I had

2:15

a lot of questions. I wanted to talk to him about

2:17

his partnership with OpenAI and why he

2:19

thought now was the best time to go after

2:21

Google and search. I also wanted to know

2:23

how the money flows how do the people who make

2:25

all the content that gets scraped into the

2:28

AI model get paid for their work? This

2:30

is short interview, but it's a good one. Sasha was very

2:32

candid. Okay. Sachin CEO

2:34

of

2:34

Microsoft. Here we go.

2:49

Sasha Nadella, you are the CEO of Microsoft.

2:51

Thank you for coming on decoder today. Thank you

2:52

so much and late for having me. Microsoft just announced

2:54

a huge new version of Bing. It's powered by

2:56

a bunch of open AI technology. A

2:59

couple weeks ago, the company made a, what was called,

3:01

a multi billion dollar, multi year investment

3:03

in Tell us what's going

3:05

on. Well, I mean, today's announcement is all about

3:07

rethinking the largest

3:09

software category there is search

3:12

with this new generation of AI because

3:14

it's a platform shift and you get to sort of reimagine

3:16

pretty much everything. Right? Starting with

3:19

the core ranking. In fact, perhaps the most salient

3:21

part of today's announcement is we've had the

3:23

best gain in relevance

3:26

in the core ranking using some of these

3:28

large models. Second is, it's not

3:30

just not just a search engine. It's an answer engine.

3:32

Because we've always had answers, but with these large

3:34

models, the fidelity of the answers

3:37

just gets so much better. And

3:39

then we've incorporated chat right into search,

3:41

which is grounded in search data. So you can do

3:43

a natural language query, which is long,

3:46

you get a great answer, and then you can engage

3:48

in conversation with that as the grounding

3:50

or the context. So it's about basically

3:52

essentially bringing in fact a more sophisticated

3:55

model, larger model, next generation model

3:57

compared to chat GPT and grounding

4:00

it in search data. The other

4:02

last thing we also added was a co pilot

4:04

for the web. So so that you

4:06

in edge can be looking at any

4:08

website or any document on

4:11

a website like a ten Q for example.

4:13

And then do things like summarization. So

4:15

a whole lot of these features all coming together

4:18

essentially as the new

4:19

bank. A really interesting piece of puzzle

4:21

here is A lot of this is powered by OpenAI

4:23

and OpenAI's technology. OpenAI CEO

4:26

Sam Altman was on stage with you today.

4:29

You've been working with OpenAI for three years.

4:31

But you haven't acquired them. You've made a huge

4:33

investment into them. Why work

4:35

with an outside technology vendor for

4:37

the largest software

4:38

category? First of all, you got to remember the

4:41

relationship with OpenAI and our cooperation

4:44

with OpenAI has got many facets.

4:46

Mhmm. The most important thing is what

4:48

we've done over the last four years is

4:50

to actually build out the core

4:52

infrastructure on which OpenAI is

4:54

built. These large models the training

4:57

infrastructure and the infrastructure doesn't

4:59

look like just Nadella cloud. Right? So we

5:01

have had to essentially evolve

5:03

Azure. To be pretty specialized

5:06

AI infrastructure on which open AI

5:08

is built. And by the way, inception is

5:10

also using Azure. Character AI is

5:12

using Azure. There will be many others who will

5:15

also use Azure infrastructure. So we are very

5:17

excited about that part. And then, of course,

5:19

we get to incorporate these large models inside

5:21

of our products and make those large models available

5:23

as Azure AI. And in all of this, we both have an

5:25

investment return and

5:27

we have a commercial return. And so we

5:29

think we're well placed to

5:31

partner. Like, I will never assume that

5:34

great partnerships can't be both,

5:36

great returns for our

5:38

customers, shareholders, and Microsoft.

5:40

There's a lot of talk today in a presentation

5:42

about the values that are coming into Bing,

5:45

about the safety work that's being done, about

5:47

the responsible AI work that Microsoft has done

5:49

for years. How do you make sure that bridges

5:51

the gap to OpenAI, which

5:53

is not your company, but obviously very

5:55

tied very closely. And how do you make sure that

5:57

your products inherit all those values

6:00

even when you're working with an outside

6:01

company? Yeah. First of all, opening eye cares about

6:04

safety. I mean, in some sense, they're in dire

6:06

inception. Was about how to think about

6:08

safety in AI and alignment in

6:10

AI. And so we share that,

6:12

and we've had our principles. As we talked

6:14

about it today in your languages. Since twenty sixteen,

6:17

we published the principles. Since ever since

6:19

quite frankly, we've been very focused on what I'll

6:21

call the hard work of incorporating it. In

6:24

the engineering practice of building

6:26

products. Right? Starting with design. One of the things

6:28

think a lot about is when you have, let's say,

6:30

a new model coming, it's

6:33

probably most important to actually

6:35

put human in the loop versus design

6:37

the human out so that you can

6:39

in fact ensure that the human agency

6:42

judgment is what you

6:44

use to train the Nadella, to be aligned

6:46

with human feedback. So that's kind of what we're

6:48

doing. Like, when I look at Even what we're

6:50

doing in Bain is taking

6:52

it even one step further, to

6:54

even ground it in the context, which

6:57

is such So I always say, look, these

6:59

generator models just don't randomly

7:01

generate stuff. You prompted it. So

7:03

there's a whole lot you can do in the matter

7:05

prompt. And the sequence of

7:07

prompts you generate, which we can assist

7:09

with. So there's a lot of I'll call it product design

7:12

choices one gets to make. Of when

7:14

you think about AI AI safety.

7:16

Then you let's come at it the other way. Right? You

7:19

have to take real care on the pretrained

7:21

data. After all Nadella are trained on pre

7:23

trained data. What's the quality, the provenance

7:26

of that pre trained data? That's a place where we've

7:28

done a lot of work. Second, then the

7:30

safety around the model. Right? At runtime,

7:32

we have lots of classifiers around harmful

7:35

content, or buyers, which

7:37

we then catch, And then, of course, the takedown,

7:39

ultimately in the application Nilay. You

7:42

also have more of the safety

7:44

net. It. So this is all the ones to come

7:46

down to, I would call it, the everyday engineering

7:49

practice. And guess what? Search

7:51

is like that. Search is an AI product. It's

7:53

kind of interesting that we are now talking about

7:55

a new algorithmic breakthrough in

7:57

these large models. But we've

7:59

always had AI models for decades

8:02

now. And we've really built, you

8:04

know, our sense of what is authoritative, how

8:06

to detect authoritative, how to ensure

8:09

harmful content doesn't get

8:10

through, and those are all practices that will now be

8:12

used. So that leads me to, I

8:14

think, the value exchange of search

8:16

right now. In a traditional search model,

8:18

I ask Bing some question.

8:21

It might return some snippet, but usually

8:23

returns a list of links. I go visit a web page.

8:25

The creator of that webpage might capture some advertising

8:28

revenue or something else. Now you're just

8:30

answering the question directly and you've trained

8:32

the model on other people's

8:34

information. Other people's reporting. I'm

8:36

very biased

8:36

in favor of reporting. How do you make

8:38

sure that they get the value back? One

8:40

of the biggest things that is

8:43

different about the way we've done the design,

8:45

I would really encourage people to go look at This is

8:47

about look, look, at the end of the day, search is about

8:49

fair use. Ultimately, all this

8:51

content we only get to use inside of a

8:53

search engine if you're generating traffic for the

8:55

people who created. And so that's why if you look

8:57

at whether it's in the answer, whether it's in

9:00

chat, These are just a different way to

9:02

represent the ten blue links more in the context

9:04

of what they want. So the core measure,

9:06

even what SEO looks like,

9:09

If anything, that'll be the thing in the next multiple

9:11

years. We'll all learn. Perhaps there will be

9:13

new incentives in SEO to even generate

9:15

more authoritative content that then gets

9:17

in. Everything you saw there,

9:20

had annotations, everything was

9:22

linkable, and that'll be the goal, whether it's

9:24

inside a search, whether it's in the answer

9:26

or even in the

9:27

chat. But if I ask the new being, what

9:29

are the ten best gaming TVs? And it just makes me

9:31

a list. Why should I the user

9:33

then click on

9:35

the link to the verge, which has

9:37

another list of the time best gaming TVs? Well,

9:39

I mean, that's a great question. But even there,

9:41

you will sort of say, hey, who did these things

9:43

come from? And would you want to go

9:45

dig in? Like, that they even search today

9:47

has that. Like, we have answers. They may not be

9:50

as high quality answers. They just

9:52

are getting better. So I don't think of this as

9:54

a complete departure from what is

9:56

expected of a search engine today, which is

9:58

supposed to really respond to your query

10:00

while giving them the links that they

10:02

can then click

10:03

on, like ads, and search works

10:05

that way. The reason I

10:07

asked this is, obviously, when you say you're

10:09

taking on a larger software category in the world of search,

10:11

there's a dominant player in Google. If

10:14

Google stopped sending as much traffic from

10:16

its search engine results page to

10:18

publishers, to creators, to other websites, regulators

10:21

on the world would freak out because they have a dominant market

10:23

share. Bing does not have a dominant market

10:26

share. When you evaluate the risks

10:28

both IP risk, legal risk,

10:31

regulatory risks. You say, well, look, we don't

10:33

have the share. We can take a step forward in how we

10:35

present these results in a way that our competitor

10:36

cannot. That's not how I come at it.

10:39

I'm just curious. See, I come at it primarily

10:41

on today if you look at the search category,

10:44

it's great. It works fifty

10:47

percent of the time. It doesn't work for

10:49

the other fifty percent of the time. So I

10:51

think what really

10:53

I wanna do is to go back and say,

10:55

look, is there some new powerful technology

10:58

that can make a search a better product?

11:00

Without fundamentally changing how

11:02

search gets permission to even exist as

11:04

a product, which is other people's content

11:07

organized in useful ways so that users

11:09

can find them. To me, that is the category.

11:11

And so we will live and die by our ability

11:14

to help publishers get their

11:16

content to be seen by more

11:18

people. Up to now, you're absolutely right.

11:20

Google dominates this market by a

11:22

significant margin. We hope

11:24

in fact if anything, having two, you

11:26

know, or multiple searches. They're not just

11:28

us. There'll be others who'll be competitors. By

11:31

having more let's call it evenly spread

11:33

search share, will only help publishers

11:36

all get traffic from multiple sources. And

11:38

by the way, advertisers better pricing. And

11:40

so publishers will make more money. Advertisers

11:42

make more money and users will have great innovation.

11:45

Oh, then think about what a great day it'll

11:47

be. I'm eager

11:49

for there to be more competition and search I'm

11:51

curious about is if

11:53

more and more people are producing more and more AI content,

11:56

and that becomes the base layer

11:58

that you're training against So

12:00

if instead of me writing a story about

12:03

the Chinese spy balloon, I asked Bing

12:05

to write such a story and that gets fed back into

12:07

Bing, eventually the amount

12:09

of original content in the ecosystem begins

12:11

to

12:11

wither. Is that a feedback loop that you're worried about?

12:13

Absolutely. But the way I look at it and

12:15

say is what people sort of talk about.

12:17

Like, my daughter sent me this unbelievable example

12:19

the other day. She's taking some French lit class

12:21

and she said, hey, I was using this AI tool to

12:24

summarize what I was writing and it took me two hours.

12:27

Because she was doing meta prompts and

12:29

prompts and learned more about that

12:31

text than ever before. And

12:33

so I feel like a little bit let's give ourselves

12:35

a little to think about

12:37

what is original content. Because as

12:40

I said, AI just just doesn't generate

12:42

it. You prompted it. You have a draft

12:44

which you edit. Today, I mean, I would be unemployable,

12:47

but for the red squiggly and my

12:49

so word. Because that's what

12:51

helps me write anything. So I

12:53

think we've used and evolved to

12:55

use new tools. I think of it that way.

12:57

Right? I think it, yes, it'll create, you know,

12:59

some of the dragery of knowledge

13:02

work may go away, but doesn't

13:04

mean I won't enjoy. Like, in fact,

13:06

the best place in the life I feel it.

13:09

Is in GitHub co pilot.

13:11

Right? Coating I mean, it's not like

13:13

suddenly you're not coding. If anything,

13:16

you're more in the flow of

13:18

coding with some of these prompts. You

13:20

read more code. You accept the

13:22

more code. So I think it's just a different way

13:25

for us to perhaps enjoy

13:27

our knowledge work more.

13:29

That brings us to the the second product,

13:31

right, which is the co pilot inside of the

13:33

edge browser. If you

13:35

look at Bing, you have an opportunity now

13:37

to capture market share from Google.

13:39

If you look at Edge, you have an opportunity to

13:41

capture market share from Chrome. Potentially

13:44

Safari if you go to the iPhone. Is that

13:46

how you're seeing this? This is an inflection point. You

13:48

have a new technology. You have a lead with this

13:50

partnership with OpenAI. It's creating

13:53

an opportunity for you to go take share? Or

13:55

is it you're expanding the

13:56

category? And you think you can initiate in the digital

13:59

anyway? Sort of I start always not

14:01

from zero sum, but I sort of sort of look and say,

14:03

hey, how does the category expand? How

14:05

can we participate in that expansion? That's at

14:07

the kind of foundation level. But at the same time,

14:09

you know, there will be. Like, these are these are places

14:11

where the dominant browser is Chrome. I mean, forget

14:13

anywhere else on Windows. Yeah. Google

14:15

makes more money than all of Microsoft. Right? So

14:18

let's start there. So there's a huge opportunity.

14:20

For us, if we got got some additional share

14:23

well for whether it's our browser or our search

14:25

engine. And so that's kinda how I look at it,

14:27

which is let's build first a product that

14:29

is competitive in the marketplace that actually

14:32

serving user needs. And, like,

14:34

all the things in the early I'm also I'm not a

14:36

one platform guy. I'm, like, I want

14:38

us. I grew up in

14:39

Microsoft that So this is your big

14:41

change Microsoft.

14:42

Not really. There's The

14:43

Microsoft that I grew up in -- Yeah. -- because I have

14:45

always do that one. Yeah. That Microsoft

14:47

software, like, office was

14:49

on the Mac before even

14:51

Windows. So that's kind of the Microsoft

14:54

that I learned

14:55

from, and I'll always make sure that our

14:57

software is everywhere where users want

14:59

it. It's been a relative period of

15:01

calm between Microsoft and Google. There

15:03

was a previous period of I would say,

15:06

antipathy or more open antipathy.

15:08

Recently, you've partnered on things like

15:11

Android on some of your hardware. You've partnered,

15:13

I think, Microsoft three sixty five on Chromebooks

15:15

is some partnership that was recently announced. Do

15:17

you expect this new sort of head

15:19

on competition against their most important

15:22

product to change that

15:23

relationship. First of all, I mean, look, I have the greatest

15:25

of aberrations for Google and what they've

15:27

done. And, you know, they're unbelievable

15:29

company with great talent and, you know,

15:31

and have a lot of respect for Sundar and his team.

15:34

So therefore, I just want us to innovate.

15:36

Right? So there's always I mean, we compete today

15:38

Today was a day where we brought some more

15:40

competition to search. We've been at it. Believe

15:43

me, I've been at it for twenty years and I've been

15:45

at it for it. But look, at the end of

15:47

the day, they are the eight hundred pound gorilla on

15:49

this, which is what they are, and I hope

15:51

that with our innovation, they will definitely

15:53

wanna come out and show that they can

15:55

dance. And I want people to know that we made them

15:57

dance. And I think that'll be a great day. What

15:59

was the moment in the development of the product

16:01

where you said, okay, it's ready. We should

16:03

announced it like this with a pretty direct

16:05

shot at the eight hundred pound

16:07

gorilla. Was there a a light switch flip for you?

16:09

Was it committee decision? How that work?

16:11

So when I first saw

16:14

this new model because the model that you

16:16

saw today is the next generation model

16:19

that's so Is it on GPT

16:20

four? Let I'll let Sam

16:22

at the right time talk about his numbers.

16:25

Okay. So it is the next generation model and

16:27

it's been done as we said, we call it the Prometheus

16:29

model because as I said, we've done a lot to

16:31

the model to ground it in search. Right?

16:33

So the search use case is pretty unique, and so we

16:35

need it to ground it in that as well. So

16:37

when I first saw the raw model

16:40

back in the summer of I'd say twenty twenty

16:42

two is when I thought that this

16:44

is a game changer in terms of the search category

16:47

aside from everything else that I'm excited about

16:49

because I do care about Azure having

16:51

these APIs even. So we've

16:53

been at it. In fact, I'll never forget my

16:55

first query I did on the model, which I

16:57

think sort of for me as it growing up, you know,

16:59

I always felt Like, if only

17:01

I could read roomy, translated

17:04

into urudu, translated

17:06

into English, that is my dream.

17:09

I just put that in as one long

17:11

query and it was magical to

17:13

see it generated.

17:16

And I said, man, this is different.

17:18

And you could have I mean, I could have programmed

17:20

it, done some multi turn. That

17:22

was your first query? That was the query.

17:24

That's just one of

17:25

the classiest people I've ever met my entire

17:27

life. I mean, that's like a very complicated.

17:30

It was just one of the I've

17:31

already worked. I mean, people have a look

17:32

portrait. He's great, man.

17:34

I mean, I I buy it. My first query was like,

17:36

Are you alive? That's

17:38

what that's what I would have gone. So you you run

17:40

this query, right, to translate revenues

17:43

at Indian Howard. Version

17:45

into across two languages and you

17:47

receive a response and you think, okay, this

17:50

is a product

17:51

or this is a product with revenue possibility

17:54

or this is product with market share possibility?

17:56

Yeah. I mean, like, all the things one of the things

17:58

that I think about is Indian platform shifts, the

18:00

two things have to happen. You have to retool pretty

18:03

much every product of yours. Mhmm. Right? So

18:05

you gotta rethink it whether it's on the way

18:07

you build it, what it's core

18:09

features are like, it's kinda

18:11

like, you know, how Microsoft had to pivot

18:13

for the cloud. Right? Which is have to rethink exchange.

18:15

It was not an exchange server. It was exchanges

18:18

as service or what we had to do with, you

18:20

know, our server infrastructure. We had to rebuild

18:23

essentially a new core stack in

18:25

Azure. So every time which transitions,

18:27

you have to essentially rewrite it. That's kinda

18:29

how I I think about it. You

18:31

also have to think about the business model. Sometimes

18:34

these transitions are pretty harsh. I'll tell

18:36

you the last transition from having, you

18:38

know, the high shares were

18:41

business with great gross margins

18:44

and saying, hey, the new business is called cloud

18:46

and it's going to have one fourth of the margins

18:48

is like the new news -- Mhmm. -- was pretty

18:50

harsh. So we made it. Whereas

18:52

in here, I look at this, there are two things.

18:55

One is it's absolutely new tech, but

18:57

it builds on cloud. Right? So that's

18:59

one place where we already have relevance. And

19:01

so there is the next generation of cloud.

19:04

And second, in search, the

19:06

economics are interesting. Which

19:08

is we already have a profitable business,

19:11

but with very little share.

19:14

And so every

19:15

day, I just want a few users and

19:17

a little bit more gross margin. And

19:19

so, yeah, I did see I think a tremendous

19:22

opportunity for us to make some

19:24

real progress here. So the model

19:26

right now, the last slide being is a eleven

19:28

billion dollars a year revenue

19:30

business, something like that? Something like that. I think Amy

19:32

is going to

19:32

talk about it. I don't know how she wants to talk about

19:34

it. Yep. Incredible hobby. I wish I had an eleven

19:37

billion dollars of your hobby. You wanna grow that into a real

19:39

business. You wanna take share. But obviously,

19:41

the new technology does not have the same cost

19:43

structure. As the old

19:45

search query. Right? I'm sure that whatever

19:47

you're doing with OkAI, it's

19:49

more compute intensive, and then obviously you have

19:51

a partner sitting in the middle of it. And then

19:53

that monetization model is

19:55

still search ads. Right? It's direct response search

19:58

ads. But as you bring more and more content

20:00

on the screen, that model might

20:01

change. Or the price

20:03

of those ads might change.

20:05

It's so wonderful. Let

20:06

me just dig into what what you just said. You

20:08

said, okay. Here is the

20:10

largest software category, when you have

20:12

the smallest share. And what

20:14

you just painted out is an unbelievable pick

20:16

of incremental GM. If Steve Barber saw

20:19

that, he would have, like, lit up

20:21

and said, oh my god. That

20:23

very few times in history,

20:25

do opportunities like that show up

20:28

where you suddenly can start a new

20:30

race. Mhmm. With our base

20:32

where every day is incremental g

20:34

m for you. And someone else

20:36

has to

20:37

Nilay, to protect it all, every user

20:39

and all the g m. So I wanna

20:41

wrap up with two questions here. One, I just wanna come

20:43

back to this. I think you are going to

20:45

face a lot of scrutiny from

20:47

publishers, creators, other

20:49

website owners saying, hey, that is

20:51

our training you're already seeing it. Right? Getty

20:53

is suing a handful of the image generation

20:56

AI companies saying, hey, you're generating

20:58

results with our watermark in it. Right? This is

21:00

obviously ours. So I'm curious if you have

21:02

a view of the potential IP risk

21:05

on the downside or on the positive

21:07

side of how to grow and keep

21:09

the ecosystem

21:10

vibrant. I mean, on the search side, I'm very, very

21:12

clear. The search category is about fair use

21:14

that we can generate traffic back to

21:16

publishers. And so that's sort of we wanna

21:18

stay. Then

21:19

is that a KPI that you're keeping track of traffic

21:21

you're sending a hundred percent Like, I mean, that's the

21:23

only way our bots are not gonna be allowed

21:25

to crawl search if we are not

21:27

driving traffic. That I think is the core of

21:29

the category. In other places again, it'll

21:32

have to be really thought through

21:34

as to what what is the fair use. And

21:36

then sometimes, I think there'll be some legal cases

21:38

that will also have to create precedent. But

21:40

at the end of the day, don't think this any of

21:42

this can be done without a

21:44

framework of law that governs it,

21:46

and ultimately financial incent is

21:48

that benefit. If anything, I look at this and

21:50

say, God, is this something where the fact

21:52

that there's gonna be more com competition can

21:55

really help publishers get more

21:57

monetization. Advertisers get better

21:59

returns for their investment and users have more

22:01

choice. All

22:02

right. I want to

22:02

end with a question that I think is the most important question.

22:04

All you have described a transformational moment

22:07

in the largest software category in the world.

22:09

You've said it, obviously, there's a moment of increased

22:12

competition against the dominant player. What

22:14

was it like in the room when you decided to stick with

22:16

the Bing brand? Right? There had

22:18

to be a slide with like fifty options

22:21

I'm assuming it's Microsoft. There's some passionate

22:23

back and forth debate. Eventually, someone

22:25

decides. Was it you who decided? Yeah.

22:28

We wanted to call it Azure search

22:30

twenty twenty

22:30

Xbox live search. You

22:33

go or bring back clip a year. Yeah.

22:35

Yeah. No. Look, it's not a really interesting

22:38

enough. It is not much of a discussion. Because we felt

22:40

like, look, we've been we love being we

22:42

have been at it. I was there the day of

22:44

launch of Bing. I worked on

22:45

it, but it has a lot of baggage as a brand.

22:47

Look, brands can be rebuilt as long as there's

22:49

innovation. I I think the brands are

22:51

only as good as the product and as good as

22:53

the innovation. And so we're gonna go wicking.

22:56

And that was a your choice. Absolutely. Alright.

22:58

Well, Sasha, thank you so much for talking today. It was really

23:00

exciting to see all the new stuff and be here to see

23:02

how it grows the future. Thank you so much. Thanks

23:06

again to Sachin for taking the time to talk today

23:08

and thank you for listening to Dakota. I hope you enjoyed

23:10

it. As always, I'd love to hear what you think is You can

23:13

email us at decoder at the verge dot com or

23:15

hit us up directly on Twitter. We're at decoder

23:17

dot com. If you like decoder, please share with your friends

23:19

and subscribe wherever you get podcast. If you

23:21

really like the show, hit us with that five star reviews.

23:24

DeCoder is a production of the Virgin and part of the Voxmedia

23:26

podcast network. Today's episode was produced by

23:28

creating DCM, Jack McDermott, Dearn Pavic

23:30

and Becca Versace, who is edited by Kelly

23:32

Wright. The decoder music is by Breakmaster

23:34

cylinder. Our editorial director is Brook mentors, and our

23:36

executive director is Alan Ardon. We'll see you

23:38

next time.