Digital sovereignty in the age of AI | WEF26 | Global Stage — Transcript
Full transcript
- 0:00Talal Al Caesi, you are the group chief
- 0:02global affairs officer at G42. I have to
- 0:06say it's impossible to have almost any
- 0:08conversation here in Davos this year
- 0:10without geopolitics coloring it. And
- 0:12when I think about digital sovereignty,
- 0:14which is a conversation happening both
- 0:16on the main stage and in many sideline
- 0:19events, is this push toward sovereignty
- 0:22being motivated by that uncertainty?
- 0:24>> It absolutely is. And I think it's not
- 0:26just geopolitics that is involved here.
- 0:27whether it's geocconomics and
- 0:28geotechnology in the world we live in
- 0:30today. I come from the world of
- 0:31geotechnology and today increasingly it
- 0:34is an intertwined system right so I
- 0:36think what we're trying to do at G42 is
- 0:38really help propagate intelligence to
- 0:40provide equitable access to intelligence
- 0:42globally and so that's what we we stand
- 0:44for as a company
- 0:45>> yeah and so let's just level set for
- 0:48everybody when we talk about digital
- 0:50sovereignty means different things to
- 0:52different people to consumers to
- 0:54countries to companies so why don't you
- 0:56define it as you understand it
- 0:58>> so you know sovereignty is a matter of
- 1:00having a level of control over
- 1:01everything from data to operations to um
- 1:04deploying systems and being able to
- 1:06leverage those systems so that they can
- 1:07underpin because this digital
- 1:09infrastructure is really what underpins
- 1:11the enablement of AI native societies at
- 1:13large. And so when when we talk about
- 1:15digital sovereignty, we do things at
- 1:17least in in G42 where we create
- 1:19innovative models that allow us to
- 1:21deploy that infrastructure in one
- 1:23country but be able to serve it in
- 1:24others like the digital embassy program
- 1:26that we announced a couple days ago here
- 1:28at Davos.
- 1:28>> So explain how that works. So that's
- 1:30sort of like treating your digital world
- 1:33as though it has a legal status
- 1:34>> in a sense. I mean it's kind of
- 1:36analogous to an embassy, a regular
- 1:37embassy, right? There's a Vienna
- 1:38convention that provides sovereignty
- 1:40over the land and the building that an
- 1:42embassy hosts and similarly in the
- 1:44digital world you have to think
- 1:46creatively about what that really means.
- 1:47So there are a couple of elements. Legal
- 1:49status is one of them. But then you have
- 1:51technical and policy controls that can
- 1:53allow um you know the companies that are
- 1:55having their data hosted outside of
- 1:57their borders. So you're losing data
- 1:59residency but gaining sovereignty
- 2:00through those controls and the G2G
- 2:02agreement on top of it that then
- 2:04solidifies the commitment in terms of
- 2:07protecting that uh other count's data
- 2:09sets.
- 2:10>> So obviously conversations about data
- 2:12sovereignty, digital sovereignty in
- 2:14general are not new. They weren't
- 2:16created by AI nor were they created by
- 2:18the current geopolitical tensions
- 2:20indeed.
- 2:20>> What about AI is putting this
- 2:22conversation on steroids? So it's a very
- 2:24good distinction that you drew because
- 2:26in the cloud world you did have to have
- 2:28multiple different cloud regions and
- 2:30there were only a handful of cloud
- 2:32providers predominantly from two
- 2:33different countries, right? And so it
- 2:35was much harder to enable sovereignty
- 2:37unless you had a private cloud, a
- 2:39disconnected cloud in every country that
- 2:40could be airgapped as they call it so
- 2:43that you can protect the imperatives.
- 2:44But that's inherently limited from a
- 2:46feature set standpoint and a scalability
- 2:49one by definition. So with AI, it's a
- 2:52little bit different because you could
- 2:53potentially have a world in which in
- 2:55like in the UAE where we're deploying a
- 2:565 gawatt AI campus, that's not going to
- 2:59be for 5 GW worth of compute and tokens
- 3:01that are manufactured in that AI factory
- 3:04to be domestically consumed. A lot of
- 3:06that is going to be exportable tokens.
- 3:08But then how do you inspire confidence
- 3:09in the countries you're exporting to
- 3:11that that data that they are processing
- 3:13through our systems are also going to be
- 3:16protected? And that's where we come up
- 3:17with those innovative solutions. That's
- 3:19like this hybrid model of I'm a little
- 3:21bit sovereign, I'm a little bit in the
- 3:22cloud. How do you make decisions about
- 3:25what to put where? Well, you have data
- 3:27classifications and you know the
- 3:29ultimate use cases here are going to be
- 3:31public sector and regulated industries
- 3:33because they are of a level of
- 3:35sensitivity not top secret that still
- 3:37requires a completely disconnected
- 3:38setting in any country but you want to
- 3:40reduce that footprint to a point where
- 3:42the plethora of other workloads that can
- 3:44benefit from the AI revolution and the
- 3:46AI compute that's being deployed can can
- 3:49uh uh you know benefit from that type of
- 3:51deployment in a sovereign way. So it'll
- 3:53be your genomics data, energy data,
- 3:55financial services, the public sector,
- 3:57government services. Those are the types
- 3:59of things that we target through these
- 4:00types of deployments.
- 4:01>> Yeah, cuz I was going to say obviously
- 4:02the technology is moving so much faster
- 4:06than the construction of the
- 4:08infrastructure or even the regulation
- 4:09around it can, right? So for a country
- 4:12that's not yet there with its own AI
- 4:15stack
- 4:16>> um how can they participate in the
- 4:19advances that are happening right now in
- 4:21healthcare in finance in digital
- 4:23government right now and not wait for
- 4:26that data center to get built
- 4:27>> and that's also a very good point
- 4:28because it is prohibitive I mean these
- 4:30data centers that we talk about if you
- 4:32look at it on a per gigawatt level these
- 4:34are about a $50 billion capex when all
- 4:36things are included and so not every
- 4:38country is going to be able to even
- 4:39deploy those data centers it's not a
- 4:40matter of how long it will take to
- 4:42deploy them in the places that they'll
- 4:44end up going. But how do you accelerate
- 4:46time to value and allow there not to be
- 4:49a digital divide in a sense and that's
- 4:51where these types of deployments allow
- 4:53us to um provide that service from afar
- 4:56in a virtual sovereign way that allows
- 4:58the company the regulators the
- 5:00government institutions in those other
- 5:02countries to embrace those technologies
- 5:04and adopt them early. So you talked
- 5:06about the digital divide. I mean this
- 5:07has been a omnipresent topic right for
- 5:11the past year. We saw from Microsoft's
- 5:14AI diffusion report the end of 2025 one
- 5:17in six people on the planet are already
- 5:19using generative AI. That's a huge
- 5:21adoption rate for a relatively new
- 5:24technology. But that also means five and
- 5:25six are not right. I think we all know
- 5:27what the problem is. Are you hearing any
- 5:31real world beyond just identifying,
- 5:33okay, we know there's a divide? How do
- 5:35we put proverbial boots on the ground
- 5:37and start to do something about this?
- 5:39>> Absolutely. We're not only I'm not only
- 5:41hearing about these things, we're
- 5:42actually doing something. What we
- 5:44defined a few Davises ago, I think it
- 5:45was two, uh we coined the term the
- 5:48intelligence grid, somewhat analogous to
- 5:50an electric electricity grid, right?
- 5:52>> And the idea is that intelligence will
- 5:54eventually need to be served like a
- 5:56utility. People need to be able to plug
- 5:58in and get that. But unlike electricity
- 6:00where till today you have 770 million
- 6:02people without access to electricity and
- 6:03there is that divide. What we need to do
- 6:05is be able to serve countries where even
- 6:07data centers wouldn't ever be able to uh
- 6:09operate whether there's it's an economic
- 6:12issue or an energy issue and that's
- 6:13where we're able to then provide that
- 6:15access immediately through that
- 6:17intelligence grid and enable AI native
- 6:19societies with or without a data center
- 6:21physically located within their borders.
- 6:23>> 5 years from now what does success look
- 6:25like for AI diffusion in your mind?
- 6:27Well, diffusion in general, I think this
- 6:30should propagate to the point where you
- 6:32are pricing intelligence in a very
- 6:34similar fashion to pricing electricity.
- 6:36So, you will see prices go down, access
- 6:38go up, uh, diffusion start to
- 6:41proliferate in a very safe and
- 6:42responsible way. uh but more importantly
- 6:44I think it's the use cases that these
- 6:46unlock and drive in particular in
- 6:48healthcare and education and other areas
- 6:50where we want to move the needle on some
- 6:52of those bigger issues like cancer
- 6:54research genomics and how you would then
- 6:56benefit from those types of data sets by
- 6:58having countries be able to access those
- 7:00technologies without physically having
- 7:01to have everything in their country ala
- 7:05thank you so much for stopping by
- 7:06>> my pleasure thank you for having me
- 7:07great to see you [music] You too.
- 7:12[music]
About this transcript
This page contains the full transcript of Digital sovereignty in the age of AI | WEF26 | Global Stage by GZERO Media, generated from the public captions YouTube serves with the video. The transcript has 1,441 words across 219 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.