FIR #467: Mary Meeker’s 2025 AI Trends Report

Jun 5, 2025

Ask episode

Chapters

Transcript

Episode notes

Pick a superlative, and it probably applies to Mary Meeker, the venture capitalist and former Wall Street securities analyst best known for her annual Internet Trends Reports. These reports, released in the form of presentation decks, were the culmination of deep research Meeker conducted. Her last report was published June 12, 2019 at Recode’s Code Conference, but she just released a new one dedicated entirely to AI. With credibility as strong as hers, it’s likely that this report will become the defining source of truth about the state of AI. In this short midweek FIR episode, Neville and Shel break down the report to focus on elements that are meaningful to professional communicators.

Links from this episode:

The next monthly, long-form episode of FIR will drop on Monday, June 23.

We host a Communicators Zoom Chat most Thursdays at 1 p.m. ET. To obtain the credentials needed to participate, contact Shel or Neville directly, request them in our Facebook group, or email fircomments@gmail.com.

Special thanks to Jay Moonah for the opening and closing music.

You can find the stories from which Shel’s FIR content is selected at Shel’s Link Blog. Shel has started a metaverse-focused Flipboard magazine. You can catch up with both co-hosts on Neville’s blog and Shel’s blog.

Disclaimer: The opinions expressed in this podcast are Shel’s and Neville’s and do not reflect the views of their employers and/or clients.

Raw Transcript:

Shel Holtz (00:10)
Hi everybody and welcome to episode number 467 of For Immediate Release. I’m Shel Holtz.

@nevillehobson (00:17)
And I’m Neville Hobson. It’s been six years since Mary Meeker last published one of her legendary internet trends reports. And now she’s back with a 340 slide deep dive focused entirely on artificial intelligence. It’s a massive document, rich in data, charts and insight, but very much aimed at a Silicon Valley audience, investors, platform builders and technologists. And while it’s fascinating, it’s also dense.

If you’re not steeped in the language of model training, deployment, and what it actually costs to run AI at scale, it can feel like stepping into someone else’s board meeting. So I decided to reinterpret it, not to dumb it down, but to make it readable and useful for business communicators, strategists, and leaders who need to understand what’s coming and why it matters. In today’s short form episode, we’re going to unpack what’s in the MECA report, some of the big themes, why they’re significant.

and how communicators in particular might think about sharing this kind of insight within their organizations in plain English, right after this message. Let me start with one insight that really sets the scene for everything else. AI adoption is moving faster than the internet or mobile did, and there’s no waiting this one out. For business leaders, this means constant adaptation.

You can’t rely on annual strategy cycles when AI tools are evolving weekly. Businesses need to become more agile, not just in tech, but in how they communicate, train and respond to change internally. So the question is, how do organizations keep up with something that doesn’t just disrupt industries? It rewrites the rule book as it goes. Shell, when you saw the scale and scope of this report, what stood out to you? And how do you think people outside the tech investor bubble should engage with this sea change?

Shel Holtz (02:06)
When I first started looking through this report, the first thing that struck me was that the word unprecedented seemed to be on every slide. In fact, an analysis I did using AI said that unprecedented appears more than 50 times in the report. And Mary Meeker does not tend to be hyperbolic. So when she says something is unprecedented, you can take it on faith that it…

@nevillehobson (02:26)
Hmm.

Shel Holtz (02:32)
probably is unprecedented. She talked about ChatGPT growing to 800 million weekly users in 17 months. That’s the fastest user ramp up in history. She called that a milestone. It took the internet 23 years to achieve global distribution. ChatGPT amassed about 365 billion annual searches. It took Google 11 years to do that. So what struck me from this whole unprecedented

adoption and scale section of her report is the need for communicators to understand that the people in your audience are using this. It doesn’t matter whether it’s internal or it’s external. So that was one of the big things that jumped out at me. This is not something, regardless of where Gartner

might place it on their hype cycle. It is not something that is going through slow adoption. There was a report, we talked about this a couple of weeks ago, that said, you know, it really is just another technology. I mean, it may be phenomenal and allow you to do wonderful things, but it still has to spread like a technology. ⁓ And according to Mary Meeker’s report, it’s already done that. ⁓ It’s not going to take as long as other technologies usually do. It’s here.

and people are using it and you have to consider that. There was a lot more ⁓ that jumped out at me, particularly around ⁓ the reshaping of the workforce, but I know you’re planning on talking about that.

@nevillehobson (04:09)
Yeah, it’s interesting. You’re right. The word unprecedented did did appear a lot, particularly in some of the kind of superlative metrics that that pepper this document. What struck me, I suppose, thinking about listeners who may not have read this, yes, thinking so what’s the big deal about this? you’ve mentioned a key point that much of this has already happened and it started already. Hence that point about you can’t think, well, we’ll kind of.

you know, factor in our planning as we go along. If you’ve got a five year plan, you need to do it weekly as the report suggests. And that might surprise a lot of people. You surely can’t be that important. It is. I like the way it’s the report is set up and kind of in eight sections that cover the breadth of what she discusses. But if you look at the kind of headline statement for each of the sections, you do get a sense of the almost a dramatic aspect of

what is actually happening according to Mary Meeker’s report. Change is accelerating. We’ve talked about that. User growth, usage and investment are surging, she says. Performance is improving while costs drop. Usage costs and losses are all growing. And that’s what Mary Meeker describes as part of the huge land grab where you sacrifice immediate profits for simply scaling and getting more people on board. Monetization is complex, she says, open source China and competition.

are rising. AI is merging with the physical world and talks about this now moving out into factories and areas that don’t necessarily make big headlines, but this has been going on already for quite a while. AI-driven growth in internet usage is historic, is driving a new wave of internet activity, says, large-scale adoption across emerging markets. And this is interesting. Many people are coming online primarily to use AI tools. Something I started doing recently was

I installed on my desktop computer, my primary computer, the extension in Chrome to use chat GPT as the primary search tool, not Google, and it replaces Google. It didn’t take me long to suddenly realize that actually I’m getting more value out of my searching this way than I was using Google. Now that just works for me. I know others do agree with that. But this indicates to me, according to another report I saw,

that the younger you are, the more likely you are doing this and not traditional search engines. So this is again, all adds to that picture. And then that final one, number eight, work is evolving rapidly, as you mentioned, and that’s definitely one to focus on. But this gives a sense of the, I guess, the dramatic nature of this that ⁓ is hard to look at it in its totality without

examining each of these areas that are in extreme detail in the report. One of the reasons I did this, which was a I started this for myself, I let me do something in simple English, because this is peppered, absolutely crammed with tech talk. And if you’re not involved in this, you will struggle with this, I’m sure, or maybe not even get the correct synthesis of what you’re reading to understand it fully.

So I think my report did it that did it for me that’s effect. So I decided to publish it, which I did as a PDF on a blog post that you could download. So it’s 340 slides. I’ve done it all down to 18 pages of a PDF. Now, so I’m not saying hey, that’s all you need to read. On the contrary, you need to read this in in tandem with Mary Meeker’s report so you can get the detail because those charts are quite important. I don’t include any of those. But

Let’s talk about the workplace shell because this is something you and I have talked about a number of times on the other episodes of the podcast, particularly one or two of the monthly long form episodes. And I found this one something that if we think people are up to speed with this, I’m not so sure that’s totally true. The awareness that this is happening, we’re going to have co-workers who are AIs and that we definitely talked about.

And the fact that she mentioned in her report, many businesses are lagging behind in supporting the shift that is already underway. So the report underscores the urgency of re-skilling, rethinking performance and supporting AI literacy. And I guess the question arises as it arose in our previous discussions, are we doing enough to prepare teams for AI native workflows or are we still thinking of it as a tool, not a co-worker? That’s the conversation we had recently in fact.

Shel Holtz (08:46)
It is, and no, we’re not doing enough to prepare.

to manage teams in that environment. The other thing that I think we’re still living in denial is that this is going to affect employment. She makes the point that AI-related job postings surged 448 % between January of 2018 and April of this year. Traditional IT job postings have fallen 9%. So this is a pivot or perish moment in

@nevillehobson (08:53)
you

Shel Holtz (09:20)
the world of technology, and you know that’s coming for the rest of us as well. Another one she points out is that 73 % of outputs from a prototype GPT-4.5 were mistaken for human writing. There was a great post from Chris Penn a week or two ago in which he found five posts, I guess he had seen them in his newsfeed on LinkedIn.

from people who said, ⁓ AI writing can never be as good as human writing. It doesn’t have lived experience. It doesn’t have empathy. It doesn’t have soul. So Chris went and found work that all five of those people had done, things they had written. And he said, you know what? It wasn’t very good. So to claim that we’re as humans producing this Pulitzer quality

⁓ literary level of work as organizational communicators, I think is a conceit. As I’ve said many times, got this from Martin Waxman, most of what we write just has to be good enough and AI can do that. I was listening to the On with Kara Swisher podcast just yesterday and she had on two journalists, each of them have written a book about Sam Altman and OpenAI.

And one of them had interviewed the CEO of EY worldwide. used to be Ernst & Young, one of the big accounting firms. And she said, we will not be laying anybody off because of AI. But we also think that we can grow the business to twice its size without having to increase headcount. you know, where did those new hires come from? Job openings plummet.

I think if we continue to live in this dream world of no layoffs, no job consequences from AI, it’s gonna turn around and bite us in the ass. So we have to start living in reality there. But in terms of the workplace, she also makes the point that this is going to weave its way into everything we do. She says that the biggest changes aren’t coming from Silicon Valley.

They’re showing up where software meets the physical world. And that could be on a farm in Iowa.

@nevillehobson (11:42)
Yeah, it is interesting. There are so many areas this is already touching that you don’t quite realize. I guess the sense of this is like a speed increasing wave, not exactly a tsunami, but a wave that’s approaching at rapid speed. It could be a tsunami in that case. We’re not aware of it, it seems to me. One of the things that I noticed as a consequence of this report is how many people I’ve seen, notably via LinkedIn,

who have also put together some kind of documentation summarizing this port. All of it that I’ve seen, and I’m only talking literally of a dozen or so posts, have used an AI to produce it, which got me wondering on one or two I’ve looked at, one in particular, the first thing I spotted was a date error. Did anyone check the AI’s output before they wrote their LinkedIn post saying, hey, here’s a report about this is my interpretation, et cetera. I used chat GPT to summarize.

the report for me, which I then myself wrote the insights from. So I made sure that I did that took me twice as long otherwise as it would have. And I’m fairly confident that it is as accurate as the report itself. It’s because that’s what I’ve been referencing. I noticed David Amano did something quite cool. I must admit he created the GPT that as it’s described, it summarizes only the uploaded Mika AI report with no outside data.

So I played around with it and it’s got actually when you when load it up in your in your chat GPT client, it’s got four already populated potential questions you might want to ask. So I just clicked one. What does report say about chat GPT use? And boom, off it reeled off a whole load of bullet points and so forth. Hey, I’ve got all that in my report, too. So at least we’re kind of consistent in how we accept that was all generated by AI. The point to me about that, though, is the argument.

That’s ongoing still. I find it’s such a waste of time. Is all this stuff about, the dashes and that indicates as AI can’t be any good, blah, blah, blah. No, look, watch out. This is now what’s happening every single day by people up and down every country and every business are using AI to help them be productive, not to con others in that, ⁓ aren’t I clever, but it’s actually the AI who wrote it. Not at all. Maybe some would do that, but that’s not important.

So pay attention to this tool that others scoff at sometimes that this is going to help you because it’s an inevitability, right? And whether you like it or not, this is here. It’s it’s coming. It’s here now. And others are already making good use of it. As we’ve reported, we mentioned again today that there are others further still who are not doing what they need to do to prepare the people for this wave that’s coming. So it is coming.

I think one other area or a number, but one in particular comes to mind as I’m talking this point is the competitive nature of this from a geopolitical point of view. So we’ve heard Trump going on about this, that and the other. China is catching up fast with these tools. Other countries are doing the same. What does that all mean for us in the context of what’s going on in the world right now, which is pretty alarming everywhere you look. So now there’s this. ⁓

You can’t control the things you can’t control. What you can control is your own organization probably. So I’d focus on that and don’t worry about all the other stuff for the time being until the rockets start coming over, right? So ⁓ it is interesting. One other point to mention, I want to just get across this point here, Shell, ⁓ about how ⁓ this isn’t just a tech race, which is related. It’s a global economic and geopolitical competition, says Mary Meesher. The report highlights three threats to current AI leaders.

powerful open source alternatives, intense U.S. competition, and China’s state-backed AI ambitions. For the global businesses, it isn’t abstract. It affects vendor choices, data handling, and regulatory risk. So should businesses be preparing for an AI environment that’s fragmented, politically sensitive, and constantly shifting? Yes, they should.

Shel Holtz (15:55)
Yes, they should. And I think you’re going to see the companies that weave AI into their products are going to make them like Swiss Army knives. Basically, if a model improves over another model, you’ll be able to take the API and just switch it to the new model rather than be married forever to a model that may deteriorate over time or the company may fall into, you know, rack and ruin.

⁓ So I think the software industry is going to catch up with the fragmentation of ⁓ the LLMs, the frontier models after a while. ⁓ You were talking about David Armano’s GPT. I wasn’t aware of that until you put it in our Slack channel. I’m definitely going to go take a look at that. ⁓ I just created a GPT. I got a new camera ⁓ and

It has all kinds of settings I don’t understand. And it’s been a long time since I’ve really paid attention to photography. And I was looking at all the pitches that are on Facebook and other channels for photography lessons and guides. And I was watching some YouTube videos. And finally, I said, the heck with this. And I went to chat GPT and I created a GPT. said, ⁓ create a tutorial that will teach me the settings of, and I gave it the model name and number of the camera I got.

@nevillehobson (17:20)
Yeah.

Shel Holtz (17:21)
while concurrently teaching me the fundamentals and principles of good photography. So these lessons say, okay, today you’re going to learn about aperture and we’re going to teach you this. So, you know, it works better than anything that I had found online to create what I wanted. Now for this report, since I only saw last night that this was going to be our topic today, and I hadn’t known about the report for more than a day or two.

I hadn’t had a chance to look at it at all or read a single piece about it. And with limited time, I just dumped all of the links that you shared, the report itself, and one or two other things I found into Notebook LM. created a notebook and I had it create a briefing paper. I had it create the 15 minute podcast that has the man and the woman talking to each other about the substance of the material that’s in the folder. In fact, the briefing report,

divided it into two parts. It was Mary Meeker’s report itself. And part two, it says critical analysis and nuance. And this is some of the other reports that didn’t just summarize Mary Meeker’s report, but shared their own views about it, some of which may have been contradictory. For example, ⁓ one of them made the point that just because people have adopted AI doesn’t mean they’re using it for…

work and productivity purposes. A lot of people have adopted this as just to have somebody to talk to or to ask the occasional question or to play games with or what have you. So you really have to start to think about the depth of use among the people in your organization and not just how many active AI users you have. They could just be, you know, making meme images with them for all you know. you know, so.

@nevillehobson (19:13)
Nope.

Shel Holtz (19:13)
I thought

that was one interesting observation ⁓ that came out of some of the reports.

@nevillehobson (19:21)
No, that’s a very good example, I think. A reality of how you can’t narrowly define use of tools like this because people will decide themselves how they’re going to use it, such as you’ve just set out. I do similar. It’s actually spookier. You mentioned the camera. I’ve been using Chatt GPT to research a camera I’m thinking of buying or a camera type I’m thinking of buying instead of the traditional route, which is Googling it, looking at photography journals and what the journalists say. No.

Chat GPT has done a great job. My need is different to yours, however, because what I’m looking for is something that will let me not use my phone when Laura, my wife and I are out looking at interesting places or walking around stately homes in the countryside. Instead, having a camera that I can use just for that, and it’s not going to be connected to the internet industry. That’s not a prime need. Chat GPT came out with five recommendations. And here we’re not talking about single lens reflex cameras.

not talking about mini compact cameras, talking about something slightly beyond that, that suits that use. I don’t want to have to figure out aperture and all that. I don’t want to know about any of that stuff. So probably a compact camera. It’s come up with one I’ve now been doing further research on, but it started with that. And that’s a change. And it’s a simple one. people are doing that kind of thing all the time, I’m sure. yeah, exactly. So

Shel Holtz (20:42)
absolutely.

@nevillehobson (20:45)
The bit about talking. Yep, I use chat GPT a lot as a sounding board. I often think to myself, my God, if I don’t delete some of my old chats, is that going to crash the whole of chat GPT? I must have thousands of chats in there, thousands that I’ve done since 2022 or 2023. Do I want them for any reason? Not really, but yes, I do. I don’t want to delete them because…

the way chat GPT works, it knows everything about me now from all those chats. And it surprises me sometimes when I ask it something. And it reminds me that we discussed this very topic six months ago, stuff like that. So it is part of our landscape. This is whether we like it or not. Merrymakers report, highly technical though it is, but look at some of the ⁓ business friendly versions or interpretations or

aligned commentaries that are popping up here and then including the one I wrote and published in my blog today, that it will give you the means to better understand all this in your own particular context. But, you know, don’t don’t look at it then come back to in five months, it’ll be out of date by then I would imagine so it’s as fast moving without any question.

Shel Holtz (21:58)
A couple of other notes that I made as I was doing the research on this. First of all, think communicators need to, if somebody hasn’t already taken on this role in the organization, lead the conversation about job evolution. Because if all you’re thinking about AI is that it’s going to ⁓ be in tools that your employees will use, so they’re typing up something in Word and it now has some AI features, you’re thinking about it wrong. That’s not the…

thing that’s going to drive change to the way people do their jobs. That’s just a new tool. So we need to understand job evolution as it relates to AI, ⁓ things like working with AI coworkers and lead that conversation and help prepare the employee population because it’s going to be a dramatic change or may I say unprecedented. I also think that

@nevillehobson (22:53)
Hmm

Shel Holtz (22:54)
There’s going to be a lot of hype. People are going to chase shiny objects. So think it’s important to anchor our AI messaging in concrete demonstrable ROI. We have to look at what are we trying to accomplish and how is the AI helping us accomplish that. ⁓ Somebody made the point that having good governance is going to be a market differentiator. So again, leading the conversation on AI governance. If somebody hasn’t already taken up that.

mantle and run with it. ⁓ Also, it’s important to keep everybody grounded in the role of humans here. One line I read that I liked, and I think this is in one of the reports about Mary Meeker’s report, was that the models ⁓ can predict a perfectly logical next action that falls flat the moment it meets unwritten office culture or office policy.

⁓ You have to keep the company grounded in the fact that people need to be involved in these things. ⁓ Finally, Mary Meeker made the point that this is ⁓ the biggest change in knowledge distribution since the printing press. So bigger than the internet, ⁓ because the internet was after the printing press, right? ⁓ Knowledge is now suddenly more available.

@nevillehobson (23:55)
Yeah.

Shel Holtz (24:18)
to more people, knowledge has become cheap, right? Because, I mean, this is basic economics. A big supply means that the thing that you have the big supply of is cheaper. What’s not cheap is judgment. And again, I think that’s a human quality ⁓ and not an AI quality. And finally, there’s also long-term thinking that AI really doesn’t do it. If you’re talking about having a…

an AI CEO, it’ll probably do pretty well at navigating the issues and challenges you’re facing right now. But looking 10 years ahead, my understanding is that it’s not very good at that yet. So keep everybody grounded in the need ⁓ for the human in the loop.

@nevillehobson (25:06)
There was one quote I did like in the report that says in 1999, Vince surf, the father of the internet described the internet era as moving in dog years, seven times faster than regular life. The AI era says Mary Meek is now running at cheetah speed. What’s that at least 10 times faster, if not lots more. So that that kind of thing helps you think, ⁓ this is this is really, really moving. The other thing just to

for my final point to mention is something I mentioned in the blog post where I kind of summarized why does this matter to communicate is more than any other, but not just HR operations, marketing, etc. Your employees are using AI with or without policy. That’s the thing to bear in mind. That’s not like someone they’re using it no matter what your customers expect faster, smarter, more tailored interactions. Your competitors are building new capabilities you might not see coming. And there are about a dozen more, but those three struck me as something that

keep in mind when you’re looking at these developments and trends that Mary Mikos pointing out and others too, and all the interpreted wisdom that’s emerging what others are saying to navigate all of that. And you realize if you are skeptical, that this is definitely real and this is happening and you need to be the change agent as a communicator. So that’s your job.

Shel Holtz (26:25)
And we have a link to the report and to many ⁓ news write-ups and summaries and analyses of the report in the show notes. And that’ll be a 30 for this episode of For Immediate Release.

The post FIR #467: Mary Meeker’s 2025 AI Trends Report appeared first on FIR Podcast Network.