WEBVTT

00:00:00.041 --> 00:00:02.460
We are here at AWS re:Invent

00:00:02.460 --> 00:00:05.338
and the future of Cloud and Gen

00:00:05.422 --> 00:00:08.425
AI and the overlap is a key focus.

00:00:08.550 --> 00:00:12.804
With me is an expert on both matters,
Filippo Scognamiglio

00:00:13.013 --> 00:00:15.598
He is a Managing Director
and Partner at BCG.

00:00:15.598 --> 00:00:18.476
He is also Global
Head of Cloud Advisory.

00:00:18.476 --> 00:00:21.312
Filippo, thanks in advance for the time.
Thanks for having me.

00:00:21.312 --> 00:00:24.315
So this year has been a breakout year.

00:00:24.315 --> 00:00:26.609
Gen AI: super hot topic.

00:00:26.609 --> 00:00:30.071
What effects are you seeing on the cloud
industry?

00:00:30.155 --> 00:00:32.741
Well, first and foremost,
tremendous amounts of investment.

00:00:32.741 --> 00:00:36.953
Microsoft clearly has a leading
partnership with Open AI and they've been

00:00:37.037 --> 00:00:41.207
making great inroads and building
capabilities on their cloud platform, Azure.

00:00:41.291 --> 00:00:45.503
Amazon and Google have been racing
to catch up, and potentially even leapfrog

00:00:45.503 --> 00:00:47.297
what Microsoft has been doing and pouring

00:00:47.297 --> 00:00:50.258
tremendous amounts of investment
in this space.

00:00:50.258 --> 00:00:53.928
So at BCG, you have introduced a standardized

00:00:53.928 --> 00:00:56.931
pricing index for the cloud industry.

00:00:57.098 --> 00:00:58.683
What are you seeing?

00:00:58.683 --> 00:01:00.143
Well, lots of surprises.

00:01:00.143 --> 00:01:02.562
In fact, it's our first attempt
at creating the equivalent

00:01:02.562 --> 00:01:05.398
of a Big Mac index,
but for cloud resources specifically

00:01:05.398 --> 00:01:08.818
to make them comparable
in a space where historically, prices

00:01:08.818 --> 00:01:12.155
really were not that comparable
considering their complexity.

00:01:12.238 --> 00:01:15.784
In looking at these comparisons, we found
actually quite a lot of surprises.

00:01:15.825 --> 00:01:20.205
I think the biggest one is that cloud is
not more affordable in emerging economies.

00:01:20.288 --> 00:01:23.541
Matter of fact, it is more expensive
in emerging economies,

00:01:23.625 --> 00:01:26.503
the cheapest location
to run cloud workloads actually happens

00:01:26.503 --> 00:01:29.923
to be the US, which is not what
we expected when we initiated this research.

00:01:30.006 --> 00:01:31.382
Yeah that’s surprising.

00:01:31.382 --> 00:01:36.387
So when you think of the year ahead
and for those who are looking to utilize

00:01:36.429 --> 00:01:40.391
GenAI, I have to ask you an age
old tech question,

00:01:40.391 --> 00:01:43.394
is it better to build
or is it better to buy?

00:01:43.478 --> 00:01:47.398
When it comes to web scale, large language
models; buy is definitely the choice

00:01:47.398 --> 00:01:50.443
right now. Trying to replicate
what some of these companies have done

00:01:50.443 --> 00:01:54.531
in your own local environment
is prohibitively expensive.

00:01:54.614 --> 00:01:55.323
I do believe that

00:01:55.323 --> 00:01:58.326
in the future
we will find more specialized models

00:01:58.326 --> 00:02:02.163
that can operate at a smaller scale
for specific purposes and use cases.

00:02:02.247 --> 00:02:06.376
And with that class of models, I think
we'll find an opportunity to build,

00:02:06.376 --> 00:02:12.006
in addition to relying on the quote “bought” models
from cloud service parts themselves.

00:02:12.090 --> 00:02:13.508
Filippo, thank you very much.

00:02:13.508 --> 00:02:16.636
Filippo Scognamiglio joining us there.