The Impact of AI on the Tech Ecosystem: From Chip Makers to Software and Regulatory Challenges

AI's transition to reasoning will boost accelerated computing, particularly inference and time-based inference.

### The Future of High-Bandwidth Memory and AI's Impact on the Tech Ecosystem
* 💻 High-bandwidth memory is becoming increasingly important as US chipmakers focus on **advanced semiconductor technologies**, driven by advancements in **AI over recent years**.
* 🚀 The deployment of **advanced semiconductor content** has reached **unprecedented levels**, with significant contributions from companies like OpenAI, Google, and Anthropic.
* 🧠 As the industry transitions from **pre-training to reasoning capabilities** in AI, this shift is expected to **enhance the demand for accelerated computing solutions**, including GPUs and custom ASICs.
* 📈 The year 2025 is anticipated to be pivotal, as companies begin to **monetize the substantial investments made in data centers** over the past few years, leading to **growth in the software ecosystem**.
* 💡 The software market has expanded significantly, with AI bringing new opportunities that could **increase the total addressable market (TAM)** for software and services.
* ⚖ The tech landscape is likely to see a narrowing of winners and losers, as companies that **effectively leverage AI** will experience **accelerated growth**, while others, like Adobe, may struggle.
* 🏛 Regulatory challenges are expected to persist, particularly for large-cap tech companies, impacting their **merger and acquisition opportunities**, as exemplified by **ongoing legal scrutiny** in cases like TikTok.
* 🔍 Overall, the tech ecosystem is poised for significant changes as AI evolves, with a clear distinction emerging between companies that **adapt successfully** and those that **fall behind**.

What do you make of high-bandwidth
memory? The focus on US chipmakers to get into
that space? Yeah.

I mean, we've witnessed probably the largest
deployment of advanced semiconductor content ever in our history.
And the reason that's happened is

because we've advanced AI over the last
couple of years. And, you know, those advancements
continue. I mean, we've really laid the tracks the
last couple of years with retraining of all these models from OpenAI, from
Google with Gemini, Anthropic, and others.

And what's really interesting now is
we're actually making the transition from
from what we've seen with these models to reasoning.
And that's a really big deal because now we're going to start seeing the handoff
in a pretty material way from

pre-training and training to actually
inference and time-based inference. And so that should continue to be quite
good for accelerated computing. It's just the mix might get a little
different over time. But as it relates to, you know,
GPUs, custom ASICs from the hyperscalers,

high-bandwidth memory, optics, you know,
it should continue to be quite strong. Janus Henderson has some exposure to
Micron, but I'm not seeing it, for example, in your Global Technology
Fund. For example, I'm interested, therefore, more about what
you think about the shift that we've

Yeah, that's absolutely right.
And, you know, I actually think 2025 is going to be a really interesting year
because you can't put the investment into data centers that we put in over
the last couple of years unless you

start to monetize it.
And we're actually starting to see the breadcrumbs, particularly over the last
couple of quarters from a lot of companies.
And, you know, you hit it on the head. You know, the application software
ecosystem, whether that's publicly

traded vendors that are starting to
actually talk about specific contributions to their revenue models, or
a ton of, you know, startups, obviously, that are, you know, racing to,
you know, disrupt the existing vendors and take advantage of this multi-trillion
dollar opportunity.

And what's really interesting is,
you know, the software industry, you know, over the last 20 years, you know,
you grew to, you know, I don't know, six, seven hundred billion dollars in aggregate as it expanded
the software TAM.

But as you really start thinking about
what AI can do, this jump to reasoning and where we're going from here, it
brings not only the software TAM into play, but this, you know, larger,
you know, services TAM that can be attacked

by AI.
And that should be really good for the software ecosystem.
How are we going to start to see the wheat from the chaff here, though?
Throughout 2024, in many ways, all boats have risen until the second half,
and then we started to pick winners.

against losers. Are we going to see that continue? Look, we take a case study of Adobe just not managing to deliver when it comes to how much generative AI is going to get to their bottom line compared to Palantir, that has just exploded ever higher because of the opportunity.

Yeah, well, that's that's where we're at. And I mean, we've actually seen a narrowing in the semiconductor ecosystem in terms of the winners as we look at picks and shovels. And that's to be expected. Every time there's a big theme like

this, you get a rising tide; it lifts all boats, and then you get a narrowing of it. And, you know, and I think that's what we're going to see across, you know, the rest of the tech ecosystem, including

software. In '25, as you said, we'll separate the wheat from the chaff. The companies that are actually having meaningful contributions from AI, they're going to get higher multiples. Their growth is going to accelerate, and it's going to be good news. And then there are going to be companies

that are clearly struggling to get to the right side of time, as you mentioned, and Adobe is, you know, it's an interesting case study because it's been a heavily debated stock in the investment community for the last couple of years as investors have gone from Adobe being on the right side of time, as

it relates to AI, to potentially being on the wrong side of time. Talking of the wrong side of time, many are trying to guess what the right or wrong side of the next administration is going to be. This is going to affect, for example,

some internet names. I know that you focus in on Danny. And at the moment, we've got breaking news. The US Supreme Court is going to hear arguments over a TikTok divestment. We're going to dig into that in a moment.

But we do see regulatory implications for a lot of companies coming. How do you navigate that? Yeah, well, not too differently than how we've navigated it for the last handful of years.

And what I mean by that, we've had a very hostile backdrop for the last four years. So, is it going to get worse? It's probably hard to get worse. Is it going to get better? It might get a little bit better in certain areas, but I expect it to continue to be a difficult environment for large-cap tech.

You know, that diminishes the opportunity for them to do M&A. And then, it also, you know, we will continue to see aggressive postures from the DOJ, like we've seen with the Google case.

A Narrowing In Semiconductors

Welcome to the Bloomberg Technology YouTube channel. Subscribe for original series and mini-docs on game-changing technologies, profiles on innovators, live conversations with experts, and breaking technology news & analysis from Bloomberg TV.

Bloomberg Technology

Discussion on the impact of advancements in AI, particularly the shift towards reasoning and its implications for accelerated computing and various hardware components.

Significant deployment of advanced semiconductor content driven by AI advancements.

Transition from AI models to reasoning, leading to a shift towards inference and time-based inference.

Continued growth expected for accelerated computing, with a potential shift in the mix of hardware components like GPUs, custom ASICs, high-bandwidth memory, and optics.

Focus on high-bandwidth memory and its role in the evolving landscape of accelerated computing.

Monetization of data center investments and the rise of the application software ecosystem in the context of AI.

Expectation of increased monetization of data center investments made in recent years.

Growth in the application software ecosystem, with both established vendors and startups contributing to revenue models.

AI's potential to expand not only the software TAM but also the larger services TAM.

Evaluation of companies in the AI landscape and the impact of regulatory implications.

Discussion on separating successful companies from struggling ones in the AI space, using Adobe and Palantir as examples.

Narrowing down of winners in the semiconductor ecosystem, following the initial rising tide that lifted all boats.

Companies with meaningful AI contributions expected to experience higher multiples and accelerated growth.

Regulatory implications and their impact on companies, with a focus on the hostile backdrop for large-cap tech and the TikTok divestment case.

Navigation of the regulatory environment, anticipating continued difficulties for large-cap tech and aggressive postures from the DOJ.