Christopher Young , Executive Vice President, Business Development, Strategy and Ventures, Microsoft Interviewer : Kristin Stoller, Editorial Director, Fortune
Category
🤖
TechTranscript
00:00All right, good afternoon, everyone.
00:01Chris, good to be here with you.
00:03Thanks for having me.
00:04Of course, so I want to start off
00:06by talking about investments.
00:07Obviously, Microsoft is a big investor in the AI space,
00:11but I want to talk about embodied AI.
00:12So some of the recent investments Microsoft has made
00:15is Robotic Startup Figure, Self-Driving Car Startup Wave.
00:18Where do you see the biggest potential for embodied AI,
00:21and how does that fit into Microsoft's
00:23bigger investment strategy?
00:24So what you just described are two great examples
00:27of autonomous systems.
00:28And a couple of years ago, we did some work,
00:30and we said, you know, this was even pre-generative AI.
00:32We felt that autonomous systems were really
00:34going to be some of the future applications that would really
00:37change how we lived our lives, how organizations function.
00:41But when we look at that space, we
00:43think it's going to be a combination of self-driving
00:46cars or self-driving vehicles, to put it more broadly
00:49speaking.
00:50I think it's going to be less about transporting people
00:52around and a lot more about transporting goods and services
00:55around.
00:55Ultimately, long-haul trucking would be an example of that.
00:59Or just think about all the vehicles
01:00that operate in small warehouse contexts
01:03or even in a shipyard as you move goods off a ship
01:06onto trucks and the like.
01:08And then on the embodied side, look,
01:12I think humanoid robotics, it's going to happen.
01:15That'll be something that'll be quite commonplace,
01:18I think, particularly in categories of work
01:20where it's very dangerous for people.
01:23Think about offshore oil rigs or oil drilling in Antarctica.
01:29This is going to be a lot of different opportunities,
01:31I think, for us to bring new types of embodied technology
01:35into those kinds of categories.
01:36And it's pretty exciting.
01:37Yeah, definitely.
01:38And speaking of investments, obviously, we
01:40just heard from OpenAI, Microsoft, a big investor.
01:43So where does that relationship currently stand?
01:46Are you partners?
01:47Are you competitors?
01:48How do you find a coexistence with them?
01:50We're great partners.
01:52And it was even mentioned by Jason a little bit earlier.
01:54We invested in OpenAI a number of years ago,
01:57even pre-pandemic.
01:59Over time, we've had an amazing partnership.
02:01If you think about all the great things
02:03we've been able to do together as partners, if you look back,
02:06I mean, it's been a great partnership.
02:08And I would say one of the more unique partnerships
02:11you even would see in technology.
02:14Obviously, we have a lot that we're doing together.
02:17You even see Mike at our Ignite event a couple of weeks ago,
02:20we announced a number of new innovations.
02:23All of our co-pilot innovations are
02:24built on with OpenAI technology as part of it.
02:27And so we're really excited about the partnership.
02:29And we continue to do a lot together.
02:31Speaking of Ignite, there was a lot
02:33about announcements around co-pilot,
02:35including co-pilot actions.
02:37Tell us a little bit more about this
02:39and what excites you the most in terms of co-pilot's capabilities
02:41going forward.
02:42So three big things at Ignite.
02:45We had many announcements, I think 80 announcements.
02:47But three big ones, you just mentioned
02:49co-pilot actions, really templates,
02:51or really being able to bring more agentic functions
02:54to co-pilot.
02:56The second one was agents, which is ultimately
02:58going to be the extension of how people do their work
03:01and integrating that into a process flow.
03:03And the third is really Azure AI Foundry,
03:05which is going to enable developers
03:08to build a lot of their own co-pilot-like capabilities,
03:11build their own agents.
03:13So we were really proud of all the announcements.
03:15But those are some of the big categories
03:16that we announced a couple weeks ago.
03:18I want to come to the audience soon for questions.
03:19But first, I want to go back to what you said about Azure.
03:21Tell us about what the advancements in that have been
03:24and what excites you about it for the future.
03:26So first of all, think about Microsoft
03:27as a platform company.
03:30We are bringing a diversity of capabilities
03:32to help developers build their own tools, their own co-pilots,
03:36their own agents into Azure.
03:39And that's really what Azure AI Foundry is all about.
03:42And you're seeing a lot of that.
03:44I would say if you look back, we were really
03:46one of the first companies through GitHub co-pilot
03:48to bring capabilities for developers
03:51to do more code development.
03:53Now what we're doing is bringing capabilities and tools
03:55for developers to build agents, to build co-pilots,
03:59and to do that using a diversity of models.
04:01We have a lot of different models on our catalog
04:03and are allowing developers to use what works best for them.
04:07Makes sense.
04:07Any questions from anyone over here?
04:11All right, good, because I have a lot.
04:13I'll keep going.
04:14So earlier last month, Microsoft unveiled a new open source
04:18multi-agent AI system called Magnetic One.
04:21How is that different from all the other multi-agent systems
04:24out there today?
04:25And where do you see the most commercial success there?
04:27So what you're referencing is some work
04:29that came out of our Microsoft Research Organization.
04:32And they published a paper on multi-agent one,
04:35which is an agentic framework.
04:37And if you think about it, it's really
04:39based on a lot of the theory that agents
04:42will function in teams in many ways just like people do.
04:45I mean, you'll have a lead, for example, a lead agent that
04:49will direct some of the specific work that
04:51will happen by very specific functional agents.
04:53And part of the reason for that is
04:55as you go through any completion of task,
04:57whether it takes a period of time
04:59or whether it has multiple facets to it,
05:01there will be changes.
05:02There will be adjustments that need to be made.
05:05And the lead agent will direct that work,
05:08similar to the way a team leader might lead work in a project.
05:12And so it mirrors that concept.
05:15But if you think about it this way,
05:18ultimately agents will need to function
05:20in concert with one another because there
05:21will be many different agents at work
05:23in an organization that might be even at work
05:26on behalf of an individual.
05:28And so having a lead agent that directs
05:31the functionality of other agents
05:32is really the core concept of that.
05:34And you've seen a lot of this.
05:36You saw a lot of this Ignite.
05:37We're focusing on extending the value of what
05:40we're doing with Copilot into the world of agents
05:42and the multi-agent framework is really a part of that.
05:45Makes sense.
05:46Another announcement this year that I personally
05:48found fascinating was the announcement
05:50around Three Mile Island.
05:51So Microsoft announced plans to reopen it to support AI.
05:55Can you talk to me a little bit about what
05:58the future of the nuclear industry is as it relates to AI?
06:01And then also, how does that relate to sustainability?
06:04I'd love to hear your thoughts.
06:05So look, the way we look at nuclear is
06:08it is a clean energy source
06:11that will become part of our overall portfolio
06:14of energy sources that we use for delivering
06:17our data center capacity around the world.
06:19And we expect there's gonna be a diversity of energy sources.
06:23We've made certain commitments as a company
06:26to being carbon neutral and even carbon negative in 2030.
06:30And so nuclear is a part of that portfolio
06:33and that's really what that announcement was.
06:35It's part of a much bigger strategy
06:36that we've got as a company.
06:38Gotcha, interesting.
06:39And how are you looking about sustainability
06:40as it relates to it as well?
06:42So the way we think about sustainability
06:43is number one, it's a must.
06:46And we expect that there's gonna be a lot of different,
06:50in fact, there's a recent blog that we published
06:54which are some of the trends that we see.
06:56And one of the big trends that we see
06:58is really the ability to deliver more and more
07:01compute capacity with a lot better efficiency.
07:04So for example, better efficiency around
07:07how we build data centers in the first place,
07:09better efficiency on how we do cooling in data centers.
07:12That's one of the big ways in which power gets used
07:15in a data center context is that the cooling take,
07:17it takes a lot of energy.
07:19Being more water efficient in how we run
07:21our data centers as well.
07:22So we have a series of different methods
07:26that we're employing in our data centers.
07:29Some of them we've been doing for a while, others are new.
07:31We continue to work with our partners,
07:33all of our partners like NVIDIA, AMD,
07:35even the in-house silicon that we're building at Microsoft.
07:38All of that is with an eye towards making our data centers
07:41more efficient, more effective, and ultimately part
07:44of a more sustainable delivery mechanism
07:47across what we do around the globe.
07:50If you had to pick three big bets that you would make
07:54on AI in 2025 or big trends, what would those be?
07:58Well, I'll tell you what we're investing in right now.
08:01Because as was mentioned earlier,
08:03one of my responsibilities is for our venture arm,
08:06M12 at Microsoft.
08:08And so a couple of key categories
08:09that I think are really important.
08:11One is, broadly speaking, AI infrastructure.
08:14But if I were to break that down into some core components,
08:17one aspect of that is going to be cyber security.
08:19Near and dear to my heart,
08:20I used to be in cyber for a long time.
08:23But for example, we've invested
08:24in a company called Hidden Layer.
08:26And their focus is protecting models
08:29against being manipulated by third party actors.
08:32Because the models themselves will become
08:34attack surface for malicious actors.
08:37And so that's just one example.
08:39We're also seeing need, for example,
08:41for data governance and data compliance, data protection.
08:45So if you think about this way,
08:47as we bring more AI capabilities into organizations,
08:51those AI agents, the AI copilots
08:53are gonna want to access information.
08:56You need better governance, you need identity capabilities.
08:59And so we're seeing a resurgence.
09:01We've invested in a number of companies.
09:03Reliance AI is one we recently invested in
09:05in that space that's in the data governance category.
09:08We also have a number of our own products at Microsoft,
09:10first party products that are in this category.
09:13And then we're even seeing the need
09:14for new databases in some cases.
09:16Like one of our recent investments
09:18was a company called Neon.
09:19It's a serverless Postgres database.
09:21And we're seeing a lot of demand
09:23for different database models,
09:25different data architectures as it relates to developers
09:28building new applications that are AI first in nature.
09:31And so there's so much happening right now.
09:33But that broad category of capabilities
09:36that are AI infrastructure enablers
09:39is a big area that I think is important to focus on.
09:43That's where we're making a lot of bets.
09:44And then some of the other ones you just said a moment ago,
09:47embodied AI.
09:49The last one I'd add is I do think
09:51there's a lot of opportunity in,
09:53generally speaking, in the healthcare space.
09:55All the way from patient care and delivery
09:57through drug discovery and development.
09:59A lot of exciting companies that are using AI.
10:02One example I would share is we recently invested
10:05in a company called Inveda.
10:07Inveda, what they do is they're using generative AI
10:10to essentially study what happens in nature
10:13and look for the natural remedies
10:15that nature has for certain ailments.
10:17Like eczema is one that they've got actually
10:19a few drugs and trials to treat eczema
10:22using natural remedies that occur in nature.
10:25And so that's just one example of many.
10:28We've invested in companies that are using AI
10:31as a better mechanism for diagnosing
10:33or assisting pathologists and radiologists
10:35in the diagnoses of cancer
10:38by being just really good at the screening aspects of this.
10:41And so I think there's a lot to be done there
10:43across the board in healthcare.
10:45And it should hopefully raise the bar
10:47for health outcome for all of us.
10:48Yeah, and I love that you have
10:50this cybersecurity background too.
10:51I think it's so fascinating.
10:52So I'm gonna ask you on the flip side of things then.
10:55What is your biggest fear right now
10:56as it relates to cybersecurity and AI?
10:59I mean, there's near-term fears
11:02and there's long-term fears.
11:03I mean, look, my wife would tell you
11:05that being married to somebody like me,
11:07you have a certain degree of paranoia as your baseline.
11:11Same with a journalist.
11:12But I would tell you, look,
11:15I think there's a real need for responsible AI.
11:18It's one of the reasons why at Microsoft
11:19we've invested really heavily in really testing
11:22and a lot of the use cases around
11:25looking for what we think could be model manipulation,
11:29misinformation, disinformation.
11:31You know, there's a lot of focus on that
11:32the last couple of months here
11:33and just even in the United States,
11:35given the election, people were on the,
11:37I would say, really on the lookout for that.
11:39And I think we came through,
11:41all things being equal came through
11:43without a lot of too many big alarms on that front.
11:46So I think we have to be mindful though
11:48that everything that we do with models for good
11:51can be flipped and turned to be done,
11:53be used by malicious actors.
11:55And so just making sure we have the right security models,
11:58it's why responsible AI is really important.
12:00It's one area we've invested heavily in at Microsoft
12:04and we encourage all of our own customers,
12:06our partners to have their own investments
12:08in responsible AI for that reason.
12:10You know, in the long run, we'll find new things,
12:13we'll have new challenges we face,
12:15but I actually am pretty confident
12:17that we're making a lot of the right decisions
12:18and we'll work through challenges as they arise.
12:21Yeah, wise words.
12:22Well, thank you so much, Chris.
12:23I appreciate it.
12:24Thanks for joining us.