7 Tech Predictions for 2026 (271)

Welcome, welcome, everybody. My name is Jeff Towson, and this is the Tech Strategy Podcast from Tecmo Consulting. And the topic for today, seven tech predictions for 2026. This is basically my own shortlist of what I'm going to focus on and what I'm trying to get a lot smarter about. you know, as things emerge, which I think they will try to kind of get there first and be a bit ahead of the curve. So this is sort of my own watch list.

I also felt kind of bad about last week's podcast. I was sick. I think the thinking wasn't terribly clear. It's strange that even like 10 days later, it still bugs me that I don't think it was very good. So hopefully today we'll maybe... I don't know, reverse the tide on that one at a minimum.

So anyways, that's what I'm going to go through today. Let me give you my standard disclaimer here. Nothing in this podcast or my writing or website is investment advice. The numbers and information for me and any guests may be incorrect. The views and opinions expressed may no longer be relevant or accurate. Overall, investing is risky.

This is not investment legal or tax advice. Do your own research. And with that, let's get into the topic. Okay, no real concepts for today. We'll just sort of go through the list. These are not in order in any way. These are just kind of...

Literally, it's just from the notes I keep on my phone. So like number one on the list, and this is kind of 2025, but I think it's going to continue into 2026, which is look, Google Gemini is absolutely moving. They are like, I don't know, I don't have a good phrase for it, on fire. It is just the amount of... Generative AI tools, products, services, tech that they have been deploying, launching in the last year is crazy. You know, they were kind of surprised, I think, when OpenAI landed.

And boy, have they really caught up. I would argue they're number one in generative AI across the board. So here I got some numbers. In 2025, they released 60, six zero, basically products. Now, a lot of those are sort of updates and, you know, you go back to March, that's Gemini 2.5. You know, by November, it's Gemini 3. They got AI.

Search Mode now, which they didn't have before. They got Flow, sort of an AI filmmaking tool. They got Pixel 10. Now, the three that I pay attention to, the three I use literally every day. are Notebook LM, Nano Banana, and Deep Research. Deep Research is pretty amazing. I mean, Deep Research is like...

I use it for business thinking. I wouldn't trust it for business thinking because I think it lacks judgment. But if you want something a little bit more like the kind of stuff I would give to an analyst just out of college, like, hey, work on this. You know, take apart the refrigerator market in Australia. You know, kind of well-known, easy to understand. You can give that to deep research.

Go away for 15-20 minutes, come back, it'll write you a full report on all the players, the market share, the trend, and it'll be 75% of the way there. So I use it for a lot of sort of what I would call basic research, stuff I used to give to analysts a lot. Now I just use this for that. I wouldn't use it for business judgment, but you can use it for other stuff. But deep research, that's great.

Notebook LM, I use that not every day, every other day. And that's, you know, that's just an LLM that only studies and only writes about what you upload. So if I want to write about network effects... I'll go into that and I'll upload all my podcasts and all my articles about network effects because I don't want it mixing my thinking with whatever it finds on the web.

No, I wanted to literally just study what I give it. And I've got, you know, so much content I've developed over the last 15 years. I can basically just use that as a tool for my own research. Fantastic. I do it all the time. I use it all the time. Nano Banana, which is just image generation. It's pretty great. I kind of use Grok Imagine a lot for this stuff or Quinn. Alibaba has really pretty great.

Image generation now. You can make little videos as well. I'm using Nano Banana a lot now. So yeah, those are kind of my go-to three. But if you just look at what Google's been up to, it's crazy this past year. See if I missed it. I'm sure I've missed a lot of stuff. Anyways, I'm going to keep watching them. I follow what they're doing pretty good. That's number one. I think Gemini's just kind of, I would put them as the leader, like number one everywhere. Okay.

Number two, data architecture within enterprises, businesses. This is like maybe the most important thing a business should be working on today. I think it's going to be the key. It's going to be sort of the key that unlocks all the doors. And people are really starting to talk about data architectures and how you deal with... vastly increased amounts of data that is multimodal, unstructured, and really has to be put into sort of a processing. I view it as like a river.

coming into the firm from various sources that then goes through various processing plans to get it AI ready and usable. It's a pretty big endeavor to build this. So that's kind of what I'm thinking about a lot is this new data architecture we're seeing emerge. And what you really want at the end of the river, when it goes, the rivers come together, they all coalesce, they go through the processing plants. What you want to come out the other side.

is just a constant flow of data that is AI ready, that you can plug into any foundation model or tool you want without having to do any more. cleaning or tagging or any of that stuff so now the interesting part is the data level multimodal unstructured data It is everywhere within companies. That's every conversation. It's everything cameras can see. It's memos. It's reports. It's meeting notes. It's all of that.

Plus, you're really going to start bringing in data outside of your firm as well, from your industry, from your ecosystem, from your partners, from your supply chain. So basically the idea, I heard this said the other day, which I thought was cool. Someone said data is exponential, but insight is linear. So data is growing exponentially. And you have to sort of build an architecture that can handle exponential growth. And then the insight is linear because to a large degree.

processing the data, getting the insights, getting all of that out of that, getting it ready for AI has been very manual. It's a lot of humans looking at things, tagging things, making sure it's right, embedding it, distilling it, putting it into your vector database, a lot of mechanical sort of manual labor stuff going on.

So if you can build an architecture that can do that, that's pretty amazing. A lot of really smart people are working on this. There was a quote, I think, I don't remember where I saw it. I saw it on LinkedIn or somewhere. by a venture capitalist named Jennifer Lee. There's a lot of Jennifer Lees in venture capital, but that's all I got. Basically, she was asked, what's the most important thing for next year? And she basically said this point. I'll read a quote from her.

Her little blurb, quote, unstructured multimodal data has been enterprise's biggest bottleneck and their biggest untapped treasure. Every company is drowning in PDFs, screenshots, videos, logs, email, and semi-structured sludge. Models keep getting smarter, but the inputs keep getting messier, which causes systems... paraphrasing, which causes systems to hallucinate, agents to break in, lots of expensive workflows that rely on humans to do Q&A, and so on. So, I'll give you a more quote.

Okay, this is another direct quote. That's why untangling unstructured data becomes a generational opportunity. Enterprises need a continuous way to clean, structure, validate, and govern their multimodal data. So downstream AI... Workloads actually work. Okay, that's the end of the quote. Yeah, that's pretty spot on. The one I've been thinking about a lot because I was up at the Huawei cloud and actually Tencent cloud a month or two ago.

Both of them talked about the same thing related to this. They talked about putting GPUs into storage. So typically you have your GPUs, your CPUs, your processing, and then you have separate, you have storage. which is not really exciting. People don't talk about it. Used to be a lot of spinning drives. Now it's mostly flash, some solid state, things like that. KV caching is kind of a big deal. And there's this idea that...

We have to start putting GPUs and CPUs into storage itself so that the processing is done there. There's a couple of reasons for this. One. The data coming out, which then feeds into your compute, your LLM, your foundation models, it's too messy. It needs to come out of storage cleaner. Fine. So you can start to do that processing there. The other bit is... the amount of data is just exploding in volume. So it's very inefficient to sort of...

flow these huge volumes of data out of your storage into your CPUs to be processed and then flow them back to be saved and so on, you're better off doing the processing in the storage device. It also, it's important for security.

It's all done there. It's important to have one sort of version of the truth that has been processed and already within storage. So this sort of combination of smart storage, GPUs being put into storage, that's kind of what you hear a lot of people talking about right now.

That's a pretty big deal. So that's number two. Data architecture and enterprises, big deal. Number three, AI service business models are really kind of... complementing, sometimes transforming, sometimes destroying platform business models. Now, I've been doing this sort of off-the-cuff analogy that like, look.

Platform business models were the lines of the Savannah. They were the biggest predator. They were the most powerful business model we've seen. Digital platforms, right? Marketplaces, all of this stuff, Google, WeChat, all of it. Okay, AI service business models, like ChatGPT, they're not really platforms. They're a new business model. They're usually single player. It's not two players.

you know, merchants and consumers connecting in a platform. These are basically single player services. You go on ChatGPT just yourself. You're not connecting with other people for the moment, really. You know, I kind of said that's the tiger. We got the tiger. Now we got the lion. We don't really kind of know which is the strongest. And we're kind of figuring that out. And there's kind of one of three ways this is playing out. The simplest one is, look, sometimes the tiger just kills the lion.

That's what I look at when I see RoboTaxis taking on Uber. RoboTaxi is an AI service business. It's kind of like a vending machine with a little knowledge flywheel accelerating its development. You know, it's not a platform business model, a marketplace with drivers and riders. You don't really need Uber anymore if you have robotaxis. That business is just done. It's too expensive. Robotaxis are just much cheaper. They're much more efficient.

In that case, okay, AI service business model, yeah, it's disrupting, if not destroying, a very powerful business model. Okay, on the other end, we have something that it's like, okay. What if it just complements or augments a platform business model? Well, that would be Alibaba, Taobao. They are putting AI services into their marketplace platform.

And they're just making it more robust. But it's not really changing the core business of we're connecting merchants and consumers. We're just giving AI tools to both sides, which is what they did in Singlestick. So I'd call that an AI-first marketplace. That's still connecting humans and businesses for the most part. Okay, so that would be a complement, an augmentation. In between those two, you kind of have a transformation. What do you do with...

WeChat. I'm sorry, not WeChat. ChatGPT. If you're Google search. Is that an augmentation? Not really. A lot of people have stopped using search engines. I don't use search engines very much anymore. I usually start with something like ChatGPT or start with Gemini or something like that. Now, it may go on to the search engine and hunt for things, but I wouldn't call that an augmentation. I would call that a...

Pretty major digital transformation. I don't think Google search is going to look the same in five years as it is today. It's going to have to change. It might be a lot less powerful. I think that's probably likely. Yeah, it's going to get taken down. And, you know, we'll call that sort of a standoff between the lion and the tiger in that case. So I'm just kind of watching these two powerful business models compete.

Some of the things I've been noticing are, look, it doesn't look like a lot of these generative AI tools have very powerful competitive advantages. We don't see network effects in most of these business models. I mean, the most powerful competitive advantage, the reason why the lion is so powerful is because it's a platform business model with network effects.

Okay, we don't see network effects in most of these AI services. They're just not there. People will point to this sort of flywheel you get within intelligence and knowledge. It's not a network effect for the most part. We may be seeing one of the most powerful business models in many cases get replaced by something that is a superior service to customers, but doesn't have the power.

for the business anymore. The owners aren't going to make the profits they used to make. Now, we've seen this before. Newspapers print daily newspapers, 1970s, 1980s, and used to be one of, if not the most powerful business model, tremendous competitive advantages. Digital newspapers, digital blogs, all of these things came along, and in truth is, it's a better service for users. It is. It's just much better.

Daily blogs, these online cunt, they don't have any competitive advantages. In fact, they struggle to even break even most of the time. So the user experience is better. It's a better product or service, but the powerful business model that everyone got rich on is gone. We may see that with a lot of these platform business models. They're just going to get replaced. Robo-taxis may not make very much money. So, now, Uber didn't make a lot of money either, but you get my point. We may see...

a powerful business model replaced by a better service without all these sort of wealth-creating attributes like network effects. Now, ChatGP, now this may change. ChatGPT is a single-player, no-network-effects type business. In fact, it doesn't have many competitive advantages, really. They're just spending big in their first. It has some, but they ain't great. They may add an open AI app store.

which they've talked about. They tried a year ago. I wrote about it. It didn't happen. They're trying again. If ChatGPT adds an app store, well, that becomes an innovation platform. That has network effects. So we may see this emerge over time that maybe we'll see some more powerful competitive advantages come back. And maybe we'll even see network effects revive.

You know, among these two powerful sort of models, it's kind of a big fight and I'm watching to see how it plays out. So anyways, that's kind of point number three. It may augment it. It may transform it. It may disrupt it.

All right, number four, AI agents and specifically agent operating systems, agent operating basics, I think are probably going to be the biggest thing in 2026. That to me is number one in terms of what businesses need to start doing. Well, it's a tie between that and the data architecture question. You know, AI has been generative AI, of course.

It's a lot of a human sitting at a desk or a phone typing in something and waiting for a response, call and response. But the AI doesn't have any sort of agency within itself. AI agents start to be able to make decisions and act. You give them tools, you give them data, you give them access to things like payment, they can start to do that. Okay, that's, you know, that's a big deal. Now, you're not going to let it...

People sort of jump three steps ahead and they say, oh, this is going to be an agent running around the street shopping for me, doing all my stuff. And I've outlined three types of sort of pretty autonomous agents, which I called the ABCs. Advisor. broker, concierge. Okay. That's probably not what we're going to see. What we're going to see is a version of sort of agent operating basics. You know, I talked about digital operating basics for a long time.

Now it's sort of digital and AI operating basics. I would sort of put agent in there as well. Digital AI agent operating basics. This is stuff we're all going to start building into our companies. And most of it's going to be just taking care of highly repetitive, boring tasks and letting it do that on its own. And so this idea of an agentic operating system, I've done a couple podcasts on this. I think that's a really big deal.

I think that's going to, it's going to be slower because anything you do on the enterprise side, the B2B side, usually takes much longer to get sort of adopted than people hope. But yeah, I think that's a pretty good. That and the data architecture are my sort of too big focus. That's top of the list for me. Now, if you want to play with this stuff, what I like to play with is Comet, which is the perplexity browser.

And basically any tabs you choose to open in this browser have agents and you can, you know, well, it's mostly generative AI at this point, but the agents are sort of emerging. So it's sort of a generative AI browser, but the agents are starting to show up. I think in the news today, I think Meta just bought Manus, which is agent-focused as well, a Chinese company. So anyways, okay, that's number four. Number five is related.

Agentic A-commerce. This is the one that scares me the most. The idea that agents are going to start changing how we do commerce. ABCs. You'll have agents that represent buyers, their advisors. You'll have agents that are brokers between buyers and sellers. And then you'll have agents that represent sellers, brands. Those are your concierges. Agents, I'm sorry, advisor, broker, concierge, ABC. This is a complete sea change in how we do e-commerce.

It could be a major problem for brands who find themselves no longer able to deal directly with their customers and have to deal with their agent instead. That is a major sort of intermediation which cuts you off from... Pretty much like most of the tools that brands use to build their business require a direct relationship. I need to have a lot of engagement and touch points with my customers.

How do you do that if you're dealing with their agent? That's going to get me a lot of data on their preferences, on their intent, what they like. How do you get data if you're dealing with the agent and not the human? Then I can start to personalize and customize. Well, who are you customizing for? Are you customizing for the customer? Because you're not talking to the customer. Maybe you're talking to the agent. Do you want to personalize for the agent and hope it gets to the customer?

I want to play on emotional factors. That's what a lot of brands are based on, habit, emotion, feeling. That's a lot of brand equity. Well, agents don't have any of that. It's just a straight... sort of B2B-like transaction. So you can see it kind of hits everything. How do you do loyalty programs? How do you do membership programs? Now, for some products, like in fashion,

Okay, customers like to go and shop for themselves. If I have a good personal agent, an advisor agent, I'm not shopping in the supermarket ever again. It can buy ketchup and mustard and all the things. I will literally never engage with the supermarket again. I don't care. I don't like it. I can see whole areas of my life. I will just hand it off and forget about it. So that's a major...

transformation of all of commerce. Well, not all of commerce, a lot of commerce. That could be a big, big deal. Like if you do strategy for brands, which I do. We talk a lot about omni-channel and being omnipresent in the omni-channel and building engagement and building touch points and gathering data and personalizing and cross-selling and getting loyalty and moving the retention numbers. Okay, agents change literally all of that to some degree. That's a big one.

That's number five. So if you want to know what to do about that, I'm watching Alibaba very closely because they're one of the biggest AI companies in the world and a big e-commerce company, and they're putting those things together right now. So I'm watching them to learn how to do this. All right, last two of these will be short. Number six, I think AI cloud companies are going to continue to be the leaders. I set out at the start of 2025 with the priority of like, I want to be one of...

maybe not want to, but I want to be on the short list of people who really understand the AI cloud companies of Asia, because I think that's sort of the center of the action. So I actually, you know, proactively went out and spent... Decent amount of time with Huawei, Tencent Cloud, Baidu Cloud, Alibaba Cloud, quite a few times during the year and tried to really, I think I'm pretty good on all of those companies now.

Not all the way, but I'm a good 70% of the way there. I'm sort of becoming the expert on those four companies. So I'm going to keep watching the AI cloud companies. Within those, Alibaba cloud is international. They're the most international of that list. Baidu is sort of number one domestically within China. They're rocking and rolling, but they're not really operating out.

10 cents a bit in the middle. Huawei is actually very, very important because it's building the hardware for most of this. It is kind of the biggest player in hardware that's supporting generative AI. Definitely within China, but increasingly in Asia and other countries. So I'm looking at them much more as a hardware perspective. Those are the four. Outside of China, okay, then everyone knows who they are. It's Google Cloud. It's AWS.

It's Microsoft, Azure. Everybody knows those companies. So yeah, follow those. So I'm focusing on the Asia 4, but I'm also keeping an eye on the other three. So anyways, I'm following that. That's number six. Keep watching that. Last one, number seven. I think robotics and physical AI out of China is going to shock people. A lot of people say, you know, the standard thing is, look, it's...

It's generative AI, and then it's agentic AI, and then it's physical AI. That's Jensen Huang's slide when he talks. Okay, so physical AI is next. What does that mean? Well, it's taking these generative AI tools. whether they're agents or whether they're just generative AI sort of foundation models and taking them out of the browser and taking them out of online and putting them out in the real world. Now, when you hear this talked about out of Silicon Valley,

the subject that immediately comes up is what kind of foundation models do you need for this? And how are you going to do the data? You know, when you take a foundation, when you take a... a generative AI, an agent, or just generative AI, and you put it in the real world, you know, that's a very strange thing. It could be walking around the world. It could be moving itself. It could be grabbing things. It's dealing with real-world input.

That's a real problem in terms of the foundation model. So what people talk about is the need for 3D world models. That you can't just, you can't navigate the world as a generative AI just... predicting words you know with standard llms no you need 3d world models that let you sort of anticipate and predict how things will happen in the real world whether it's something falling off a table

picking up a glass, driving down the street, all of that. So you got to incorporate the physics, all of that. And then you have to get a lot of data. Well, the problem with that is, how do you get data? you're going to deploy now tesla sort of got around that problem because they put their cars on the streets and people were driving them and they sort of learned by people driving them around but for most physical ai

That's not going to be the case. Are we going to train a humanoid robot in 100,000 different homes so that it can move things around the living room and kitchen? Now, the data actually is very, very sparse coming in from the real world versus the Internet. So what people talk a lot about is doing simulations, and you use these world models, both the sort of...

guide it and be its foundation model, but also where you can run simulations. Hundreds of thousands of different living rooms and kitchens can be built this way, and the AI can basically be trained in simulation. So you hear about those two subjects a lot. simulation, the need for new foundation models, things like that. Okay, when you go to Asia, China, that's not what you hear about. You do a little bit, but not nearly as much. The number one subject that you always hear about...

is the supply chain. You know, these robots, the humanoids, they have hundreds and hundreds of different highly specialized parts. Well, you don't build a supply chain quickly. I mean, this supply chains take 5, 10, 15, 20 years to build, especially if you want something that's really, you know, sort of an ecosystem around a certain product type. It takes a long time.

China already has the manufacturing ecosystem. They have it for pretty much everything. But specifically, they have it for electric vehicles. And it turns out there's a whole lot of overlap between building robots and building EVs. Obviously, the batteries are a big part of it. All the sort of rare earth elements, things like that. Yeah, so China's basically, that's why you see these companies, including Tesla.

going right from building EVs to building humanoid robots because they can sort of lever over, leverage their supply chain to a significant degree. And we see that in... China as well, like Xpeng and these other companies are now releasing robots. Now, I don't understand how... Now, the West kind of has, let's say, better technology to some degree, but...

You know, the vastness of the manufacturing ecosystem of China, if you haven't been on the ground and seen it in action, it's pretty amazing when you actually see the scope of it and you realize, you know, there's no way to replicate this quickly.

Now, I think the U.S. hopefully will, but 10 years, five years minimum. So I think you're going to see China very rapidly start putting out all sorts of robots with varying degrees of sort of... you know, agentic AI or, you know, physical AI that can power them on their own as opposed to ones you, you know, you control with a controller.

But yeah, it's moving pretty fast. And, you know, I've been to Unitary and Hangzhou twice this year. The difference between when I went there in March and when I went there in November was stunning. When I went, I've told this story before, when I went there in March, they had just released their first humanoid. And it was small and it didn't do much. It could sort of walk. You had to control it with a controller. You basically train it.

in a simulation and then you download the training and you just trigger various trainings like press the button for shake your hand and it would reach out and shake your hand but it was kind of awkward and slow and the walking was a bit awkward and slow November, same robot was doing backflips. It was doing kung fu. It was running around and it was smooth. Like it was super fast, super agile, and very, very smooth.

six months. It was really quite stunning to see the advancement. So we'll see what they have in November. But yeah, it's going to keep moving pretty fast. Anyways, that's number seven. Robotics from China, I think, are going to sort of stun everyone. That's it. I'll read them back real quick. But yeah, so number one, Google Gemini rocking and rolling. I expect them to keep surprising everybody and stay out front. 2026, the data architecture, number two.

That seems to be sort of the key next missing piece for enterprise, building out this sort of architecture that processes and cleans and makes AI ready. Sorry, AI-ready data in real time. Big deal. That's what I'm looking at Huawei a lot for. Number three, AI service business models. It's going to be interesting to see how those play out against platform business models. Lions versus tigers. Number four, AI agents and agentic operating systems, agentic operating basics.

That's a big, big deal for most companies to start taking that apart. Number five, agentic e-commerce, potentially a huge disruption to a lot of companies. Number six, AI cloud. That's still where I kind of focus half my attention. And number seven, robots and physical AI coming out of China will probably surprise people. Anyways, that's my list. We'll see how I do.

I'll keep this and I'll look at it again in six months, see if I was on target. And that is it for the content for today. A bit shorter, so I wasn't completely wrong on that, about 30 minutes. Yeah, I hope that's helpful. It's end of the year. Tomorrow is January 1st, so I guess I should go out tonight and do something on New Year's. I never really do New Year's Eve. Well, I used to when I was younger. Kind of overrated. I've done like the classic things like New York City.

You know, downtown San Francisco, go into the city for New Year's on bar. I've done that a lot of years of my life. I always kind of find New Year's Eve to be not disappointing, but underwhelming. So I'm more of a stay-at-home on New Year's Eve. Restaurants are not bad. I like going to restaurants with friends. That's a pretty good New Year's Eve.

Or maybe private homes. But yeah, the whole going out into the city and all the craziness. One, I've done it. And two, I always thought it was a bit underwhelming. So I've sort of given up on New Year's a little bit. I really like Christmas. I mean, I really get into Christmas. I really like Thanksgiving. But New Year's, eh, I don't know. We'll do a little something, but probably not much. Anyways, that is it for me. I hope that's helpful, and I will talk to you next week. Bye-bye.