5 Ways AI is Different Than Past Tech Trends
23:01 Share

Today on the AI Daily Brief, five ways AI is different than past tech trends. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. Thanks to today's sponsors, Blitzy, Plum, Vanta, and Agency.org. And to get an ad-free version of the show, go to patreon.com slash ai daily brief.

Hello, friends. Quick note before we dive in. We are digging deep into the legendary Mary Meeker's latest trends report, the first in six years, which is, of course, about AI. And you better believe this one is taking the entire episode. So we will not have our normal breakdown between headlines and main. Just a main today. Second thing I wanted to note is that while I mentioned before that we have...

a Patreon now for people who want to avoid the ads in the show, I'm actually opening up a new level that is just for ad-free and pricing it basically as low as I possibly can, which will be $3 a month. So if you are one of those people who doesn't want to click skip on the ads...

For just three bucks a month, you can get it ad-free over at patreon.com slash ai-daily-brief. Hopefully this gives a better set of options that fit different types of budgets. But with that, let's get into today's show. Welcome back to the AI Daily Brief. Today we are looking at the new artificial intelligence trends report from Mary Meeker. Now,

Now, if you are relatively new to the technology industry, this is a name that you might not have heard. However, for the old hands, you might remember a time when Mary Meeker's annual Internet Trends Report was the big analyst drop for the year. It started all the way back when she was a Wall Street analyst in the mid-90s and continued right on up into 2019.

It extended through her time on Wall Street, but then also into venture capital, where she was at Kleiner Perkins before starting her own firm, Bond. And these trend reports were very, very big picture and macro. They looked at things like key trends in technology categories to infrastructure build out to global adoption patterns. And so the relevant point of all of this is, one, this is someone whose process is extremely comprehensive and macro and wide viewing.

And two, this is the first time in six years that she's done that type of report, and it's entirely focused on artificial intelligence. All 340 pages of it.

Now, for our purposes, there is obviously no chance that we're going to be able to get into everything that this behemoth of a report, which is available for free, by the way, and I will include in the show notes, gets into. And so what we're going to do is try to frame this around five ways AI is different than past tech trends as shown by this report.

We'll also talk a little bit about the audience that she's writing for and some of the things that fall out of that, as well as where she thinks things are headed. And I'll share a bit about how almost instantly out of date the report feels, even as comprehensive as it is. But let's get into these five ways that AI is different than past tech trends, because I think they're going to give you a good picture of the overall report.

On a high level, I think if you could sum up what this report is trying to say, it's that yes, AI is different than past tech trends. It represents a different phenomenon and kind of needs to be engaged with in its own terms.

The key word for the presentation, honestly, is unprecedented. In fact, Meeker says that growth in AI is unprecedented in 51 different contexts throughout the report. That word was used to describe development, adoption, investment dollars, users, and essentially all aspects of AI. This is a seasoned investor who not only lived through the dot-com bubble, but guided professional investors on where to look. And her opinion is basically that the AI megatrend makes the internet look quaint.

Now, the first part of that is simply in terms of speed. The first way in which AI differs is just that the evolution and adoption are faster. In her overview, Meeker writes, To say the world is changing at unprecedented rates is an understatement. Rapid and transformative technology innovation and adoption represent key underpinnings of these changes, as does leadership evolution for the global powers. For some, the evolution of AI will create a race to the bottom. For others, it will create a race to the top.

The speculative and frenetic forces of capitalism and creative destruction are tectonic. It's undeniable that it's game on. And indeed, talking about this idea of it being faster, section one of eight in the outline is a question. Seems like change happening faster than ever, question mark? Yes, it is.

Now, part of this is because things were poised to be faster because there was more infrastructure and distribution specifically to be built upon. Meeker points out the technology compounds. And by the way, I should make a note here. If you are just listening to this one, I highly recommend watching it. There's a lot of charts that I'm going to be referencing. In some cases, just talking over without explaining a ton. The video version is available not only on YouTube, but also on Spotify. In any case, the point here is that AI is building on everything that came before it to move faster.

She points to 260% annual growth in data to train AI models since 2010, 360% annual growth in compute to train AI models since 2010, which led to 167% growth over the last four years in the number of powerful AI models, which is now leading to just unprecedented uptake.

Now, one of the remarkable charts comes on page 20, where she shows the growth of users, subscribers, and revenue for ChatGPT, which have recently hit an inflection point that makes even their initial insane growth look slow. Indeed, Meeker calls it hard to match ever. The time it took to hit 365 billion annual searches was 5.5 times faster for ChatGPT as opposed to Google. And beyond just consumer adoption, she also points to other areas like developer adoption.

Using the Google ecosystem as a proxy for the wider AI developer ecosystem, she points out that in just a year between May 24 and May 25, there were 5x the number of developers in the Google ecosystem, from 1.4 million up to 7 million. Now part of why adoption is increasing is that performance of AI is also increasing at an incredible rate. Meeker points to image generation increases, gains in realistic audio generation, and more.

And again, what this all adds up to is just more people using this technology faster than anything that happened before. One of the most interesting charts she shares is the speed it takes to get to approximately 50% adoption of a technology in U.S. households. She points that each cycle seems to ramp in about half the time, with the PC era taking 20 years to reach 50% adoption, the desktop internet era taking 12 years, the mobile internet era taking 6 years, and the AI era seeming to take about 3 years for this.

Point is, as Meeker points out, yes, everything is happening faster. A second thing that runs throughout this report is the notion that everyone is in on this. This is not just consumers. This is not just startups. It's those actors, but it's also enterprises. It's governments. Specifically, enterprises and big tech have been extremely fast off the jump in this technology era, as opposed to perhaps some in the past. She points out that to tech incumbents, AI adoption has been a top priority.

The mentions of AI in corporate earnings transcripts has gone nothing but up over the last few years. The proportion of S&P 500 companies mentioning AI during their earnings calls rocketed up from around 10% at the time of ChatGPT's launch to over 50% now. Meeker and her team point to some stats that honestly at this point feel pretty dated about adoption in the enterprise, but which still make the point that the C-suite is getting there extremely fast.

She points to a 2024 Morgan Stanley adoption survey where 75% of global CMOs were running initial tests or experiments with AI, with basically the rest of them planning on starting testing within 12 months. Now, one theme that is very not present in this is agents, which we'll talk about towards the end. But the point is, if you flip through, it's just story after story, stat after stat of how important AI is right out of the gate, even to enterprises.

One small interesting note is that she specifically points out that at least in the early goings, companies seemed to be more focused on growth and revenue benefits from AI as opposed to cost reduction. Again, referencing a Morgan Stanley study from 2024, when leaders were asked what areas they were targeting for improvement with Gen AI, the leading areas were things like production and output, customer service, sales, productivity, revenues, whereas the lower categories were things like admin costs, manufacturing costs, headcount, and hiring costs.

Given how much time we spend talking about this idea of efficiency AI versus opportunity AI, I think there's something promising there. Now, of course, the other dimension of enterprise and big tech being in on the trend is the amount of money that they're spending on the trend.

Meeker dedicates a big section to the CapEx spend trend, showing how going back to 2014, CapEx and data have increased at 21 and 28% a year respectively, among the biggest six or so US technology companies, a trend which seems to be increasing significantly at the moment. Today's episode is brought to you by Blitzy, the enterprise autonomous software development platform with infinite code context.

which, if you don't know exactly what that means yet, do not worry, we're going to explain, and it's awesome. So Blitze is used alongside your favorite coding copilot as your batch software development platform for the enterprise, and it's meant for those who are seeking dramatic development acceleration on large-scale codebases. Traditional copilots help developers with line-by-line completions and snippets,

But Blitze works ahead of the IDE, first documenting your entire codebase, then deploying more than 3,000 coordinated AI agents working in parallel to batch build millions of lines of high-quality code for large-scale software projects. So then whether it's codebase refactors, modernizations, or bulk development of your product roadmap, the whole idea of Blitze is to provide enterprises dramatic velocity improvement.

To put it in simpler terms, for every line of code eventually provided to the human engineering team, Blitze will have written it hundreds of times, validating the output with different agents to get the highest quality code to the enterprise and batch. Projects then that would normally require dozens of developers working for months can now be completed with a fraction of the team in weeks, empowering organizations to dramatically shorten development cycles and bring products to market faster than ever.

If your enterprise is looking to accelerate software development, whether it's large-scale modernization, refactoring, or just increasing the rate of your STLC, contact Blitzy at blitzy.com, that's B-L-I-T-Z-Y dot com, to book a custom demo, or just press get started and start using the product right away. Today's episode is brought to you by Vanta.

Vanta is a trust management platform that helps businesses automate security and compliance, enabling them to demonstrate strong security practices and scale. In today's business landscape, businesses can't just claim security, they have to prove it.

Achieving compliance with a framework like SOC 2, ISO 27001, HIPAA, GDPR, and more is how businesses can demonstrate strong security practices. And we see how much this matters every time we connect enterprises with agent services providers at Superintelligent. Many of these compliance frameworks are simply not negotiable for enterprises.

The problem is that navigating security and compliance is time-consuming and complicated. It can take months of work and use up valuable time and resources. Vanta makes it easy and faster by automating compliance across 35+ frameworks. It gets you audit-ready in weeks instead of months and saves you up to 85% of associated costs. In fact, a recent IDC whitepaper found that Vanta customers achieved $535,000 per year in benefits, and the platform pays for itself in just three months.

The proof is in the numbers. More than 10,000 global companies trust Vanta, including Atlassian, Quora, and more. For a limited time, listeners get $1,000 off at vanta.com slash nlw. That's v-a-n-t-a dot com slash nlw for $1,000 off. Today's episode is brought to you by Plum. If you're building agentic workflows for clients or colleagues, it's time to take another look at Plum. Plum is where AI experts create, deploy, manage, and monetize complex automations. With

With features like one-click updates that reach all your subscribers, user-level variables for personalization, and the ability to protect your prompts and workflow IP, it's the best place to grow your AI automation practice. Serve twice the clients in half the time with Plum. Sign up today at useplum.com. That's U-S-E-P-L-U-M-B dot com forward slash N-L-W.

Today's episode is brought to you by Agency, an open source collective for interagent collaboration. Agents are, of course, the most important theme of the moment right now, not only on this show, but I think for businesses everywhere. And part of that is the expanded scope of what agents are starting to be able to do. While single agents can handle specific tasks, the real power comes when specialized agents collaborate to solve complex problems. However,

Right now, there is no standardized infrastructure for these agents to discover, communicate with, and work alongside one another. That's where Agency, spelled A-G-N-T-C-Y, comes in. Agency is an open-source collective building the Internet of Agents, a global collaboration layer where AI agents can work together. It will connect systems across vendors and frameworks, solving the biggest problems of discovery, interoperability, and scalability for enterprises.

With contributors like Cisco, Crew.ai, Langchain, and MongoDB, Agency is breaking down silos and building the future of interoperable AI. Shape the future of enterprise innovation. Visit agency.org to explore use cases now. That's A-G-N-T-C-Y dot org. Now here it's worth noting that Meeker is still writing a bit for what feels like a Wall Street-style analyst audience.

What this report lacks for immediacy, again, we'll talk about the conspicuous lack of agent discussion in just a minute, it makes up for incomprehensiveness for the type of investor audience that wants to understand the big story and how it all comes together. So for example, in the slide after pointing out that 21 and 28% growth in CapEx and data spend, she also points out that that's been mirrored by 37% annual cloud revenue growth over the last 10 years for the hyperscalers.

Now, at the same time, there is clearly an even more aggressive shift now with a 63% jump in that spend between 23 and 24. And it is worth noting that it doesn't feel like Meeker is convinced that this is going to work out all that well for AI companies. On the one hand, she points out that NVIDIA's revenue is having no issues, growing at an average of 78% over the past five years, and that even the other competitor chips put out by companies like Google and Amazon have seen incredible sales growth in the last couple of years as well.

The same trend is observable for infrastructure suppliers and data companies, but is less clear for the foundation model companies themselves, who are pouring an incredible amount of CapEx into building models that rapidly become obsolete. Three of the four big tech firms producing models have seen a significant drop in free cash flow as they spend big on AI, Meta being the outlier managing to hold their margins steady, although notably they are growing CapEx at a slower pace to rivals.

For Meeker, the equation is simple, writing, so we have high revenue growth, plus high cash burn, plus high valuations, plus high investment levels. She says, good news for consumers, others TBD.

Now, on the point of this being good for consumers, Meeker points to the trend of cost deflation. She writes, we've never seen so many founder-driven companies with market capitalizations in excess of a trillion, most with gross margins of 50% plus plus free cash flow, attacking the same opportunity at the same time in a relatively transparent world.

In other words, while the internet trend was largely about scrappy startups carving out their niche in an emerging digital world, and especially in the early days, seeing intense growth but nobody with scale, the AI trend is the clash of the titans, with numerous trillion-dollar companies deploying basically all of their resources into pursuing their opportunity. Part of the outcome of this is cost deflation, and this is coming not only from competition but also from infrastructure.

Meeker's research shows that a billion-dollar-scale data center capacity using NVIDIA's H100 chips can generate 58 trillion inference tokens annually.

This in itself was a 10x increase from the previous generation of chips. And the next generation Blackwells are set to produce 24 times more tokens for a billion-dollar scale facility. Meeker projects that this scale of inference could generate $7 billion in annual revenue at current costs, or more likely will result in massive cost deflation for AI use. She also points out that inference capacity is going exponential. The global stock of NVIDIA-powered compute is currently growing at 130% per year. All told, inference

Inference costs are down 99% over the past two years, and that trend shows no signs of slowing down. Summing up, she writes, inference represents a new cost curve, and unlike trading costs, it's arcing down, not up. As inference becomes cheaper and more efficient, the competitive pressure amongst LLM providers increases, not on accuracy alone, but also on latency, uptime, and cost per token. What used to cost dollars can now cost pennies, and what cost pennies may soon cost fractions of a cent.

The implications are still unfolding. For users and developers, this shift is a gift. Dramatically lower unit costs to access powerful AI. And as end-user costs decline, creation of new products and services is flourishing, and user and usage adoption is rising.

Now, a third way that AI is different than previous tech trends is that this is not a US-only phenomenon. In fact, it's happening everywhere all at once. In a slide on the internet versus ChatGPT, she points out that it took 23 years for 90% of internet users to be outside of North America, but just three years for that to be the case with ChatGPT users. She said, we've not seen the likes of this around the world spread before.

In fact, she has an entire section about this, Section 7, Global Internet User Ramps Powered by AI from the Get-Go, Growth We've Not Seen the Likes of Before. She writes, Thanks to the rise in low-cost satellite-driven internet connectivity and access, the potential for the 2.6 billion that is not online to come online is increasing. These new users will start from scratch with AI functionality. When these new users come online, they likely won't be met by browsers and search bars. They'll start with AI and in their native language.

Imagine a first experience of the internet that doesn't involve typing a query into a search engine, but instead talking to a machine that talks back.

Imagine skipping the traditional application layer entirely, with an agent-driven interface managing disparate tech platforms from one place while understanding users' local context, language, and intent. The rest of this section shows how much global internet penetration has increased around the world over the last decade, and once again points to global growth in ChatGPT as something that is simply unlike anything we've seen before. The top country using ChatGPT absolutely isn't the US right now. India represents 13.5% of use, while the USA represents 8.9%.

After that, it's Indonesia and Brazil, both with over 5%, Egypt with nearly 4%, and so on and so forth. Meanwhile, DeepSeek is quickly infiltrating the rest of the world that ChatGPT doesn't have access to. And speaking of places that ChatGPT doesn't have access to, the fourth way in which this trend is different is that there are geopolitical implications right from the get-go, with, of course, the US-China competition being right at the center of it.

All the way back in the intro, Meeker writes, "...two hefty forces, technological and geopolitical, are intertwining." Andrew Bosworth, the CTO of Meta Platforms on a recent podcast, described the current state of AI as our space race. "...the people we're discussing, especially China, are highly capable. There's very few secrets, there's just progress. And you want to make sure that you're never behind." The reality, Meeker writes, is AI leadership could beget geopolitical leadership and not vice versa.

Now it is very clear that the US-China battle is defining much in AI. The total number of large-scale AI systems in the US and China absolutely dwarfs the rest of the world combined.

Meeker points out that China's AI is gaining increasing relevance and doing so rapidly, pointing of course to DeepSeek, but also to advances in models from companies like Alibaba and Baidu as well. She points out that LLM performance from China AI is catching up to US models, and that they're achieving this performance parity with lower training costs. Another point, which is a little bit more forward-looking, is that one advantage that China has comes in the form of embodied AI.

As we get into AI that is embedded in robots, China is way out ahead, not only of the US, but the rest of the world combined. China has more industrial robots installed right now than the rest of the world does in total. The USA is embarrassingly far behind when it comes to this particular dimension.

Now, Meeker points out that there are lots of implications of the competition with China. It's not just geopolitical, it also is going to have impact on the business models that win. Effectively, she's saying that China and open source are putting intense downward price pressure on these services, which is shaping some of how things play out.

What's more, and this is one that I do pay attention to, China's citizens are materially more optimistic regarding the benefits of AI than U.S. citizens are. Between 2022 and 2024, the percentage of Chinese citizens that agreed with the statement that products and services using AI have more benefits than drawbacks was in the 70s in China and down in the 30s and 40s in the U.S.,

Those sort of attitude challenges in the U.S. I think are going to create even more significant headwinds than they're creating right now, and we're already seeing some of that as it is. I think if I had to sum up, the fifth way in which Meeker seems to think that AI is different than previous tech trends is just the breadth or depth of disruption. You get the sense reading this report that what she's really trying to say is that yes,

Things are changing. Yes, it's happening at an accelerated rate. And yes, when I say things are changing, I mean everything is changing. The depth and magnitude of the disruption are just really unlike things that we've seen before. And like I said, this is all even with there almost being a bit of an outdated feel to this. There's this quaint little section where they ask ChatGPT the top 10 things it'll be able to do in five years. And many of them feel like they are virtually here now.

For example, number one, generate human-level text, code, and logic. That's not five years from now. That's arguably now and certainly within the next year. I also mentioned that there's almost nothing on agents. Not that Meeker and her team don't get that this is coming, but it's more like they have so much to catch up with just with the assistant era of AI that they can barely spend the slides to focus on this emergent trend, which is agents.

They do point out that interest on Google searches has increased about 1100% over the past 16 months, and they point to some of the early agents that have been deployed, but that's pretty much it for agents in this entire presentation.

Overall, this serves as yet another reminder, and one that I think is likely to resonate with an audience who might not be as attuned to things like this podcast and other sources of information that have been telling similar stories for the last couple of years. Despite having not published her Internet Trends report for six years, Meeker is still one of the most respected investment analysts and trend watchers out there, and I think that this report is likely to be spread around quite a few desks in Wall Street and other halls of power.

For now, though, that is going to do it for this extended edition of the AI Daily Brief. Appreciate you listening as always. And until next time, peace.