Ep 732: The State of the AI Race. Who will win in 2026: OpenAI, Microsoft, Google Or Anthropic (Start Here Series Vol 12)

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life.

What company will win the AI race in 2026? Yeah, it's a lot of fun for AI geeks like you and me to argue this and to follow the benchmarks and literally every single week or every other week argue about the new next best model. But the answer obviously to that question is much more nuanced than a single company. And the answers obviously hold extreme value for today's business owners and decision makers. That's because you've probably already, and if not, you will soon, look at the Big Four in the face and say, hey, which company should our organization be using when it comes to AI? Microsoft, Anthropic, Google, or OpenAI? So on today's show, we're going to be giving you some of the answers. But let me just get straight to the big picture here. 3 years ago, one company controlled AI in OpenAI. But that era is over because now you have Microsoft, Anthropic, Google, and OpenAI all competing for the same thing. Two things mainly: the mindshare, right? That's the, the total number of consumers on the platform. And then the enterprise dollars, right? How quickly can you get those consumers over to start using it on the business side? But I think the 2026 winner of the AI race will actually come down to the workflow and how easy and how accurate it is to bring in, in an agentic manner, your company's data and for that system to deliver enterprise-ready outputs. And it's not necessarily going to be the best model. Whoever makes it is going to be the one that wins the AI race. So on today's show, we're going to go over what the Big Four are actually competing for in 2026, why the chatbot era ended, and what comes next. Where today's large language models from each company excel and where they struggle. And last but not least, I am going to give you my kind of hot take, but, you know, educated opinion, we'll say that, on who is going to win what AI race from the big 4 companies. All right. I hope you're excited. Well, I am. So if you're new here, welcome to Everyday AI in our Start Here series. This is the essential podcast series to both learn the AI basics and to double down on your knowledge. So whether you are the key decision maker in your company or you're brand new getting into AI, well, the Start Here series is for you. So if you haven't already, please go to starthere series.com. That is going to give you free access to our inner circle community. And in there, in the Start Here series space, you can go and listen to and read the entire Start Here series. I believe we're already on volume 12. All right. And if you missed our last episode, and yes, it does help to listen to them in order. They're not super long, right? Most of them are between 25 and 35 minutes. And if you listen on 2x, they're even faster. All right. But in our last episode, we covered measuring AI ROI, why you're doing it wrong, and the 7 steps to fix it. My gosh, listen to that episode, put it into practice. But today, On our Start Here series, volume 12, we're going over the state of the AI race. Who's gonna win? So let's get into it and talk about what's changed. And well, to put it simply, the chatbot era is very gone. It's done, right? Bless up. I'd say even toward the middle of 2025, a lot of enterprise decision makers were still looking at AI like a chatbot. It's not like that anymore. In my opinion, I think GPT-5.4, thinking from OpenAI, is kind of the first daily driver model that you can look at across the entire business spectrum and be like, yes, we now have probably, right, depending on how educated you are, right, in terms of large language models and knowing how they work, right? And yes, there's other great models, Gemini 3.1 Pro. Claude Opus 4.6, but I think we're probably now finally in March 2026 at the time where you can have a single model that is generally intelligent enough and transparent enough to be like, okay, we can move pretty much all of our day-to-day operations under here if we have the right experts kind of driving this thing. And it took a while to get the harnessing and the tool use around these chatbots to turn it into something more. Right? I've been calling it in one of our earlier, shows on the Start Here series was about the AI operating system. So make sure to go back and listen to that. But that's essentially where we've, kind of transitioned from away from AI chatbots, right to now these are agentic systems that do work, right? They deliver full outputs. And every major player of the Big Four has shipped a huge agentic product in the last 90 days. So let's go over some of the stats, some of the facts, some of the figures. Let's talk numbers. Okay. OpenAI, 900 million weekly active users. All right. Also, this is as of March 2026, right? So if you're listening to this, I don't know, in November or December of 2026, you know, obviously a lot of this, the models, the numbers have changed, but I'm going to guess, I have a pretty good thumb on the pulse on these things, but I'm going to guess my overall vibe and advice to you on each of these models will still probably hold true by the end of the year, if I'm being honest. But for the most part, OpenAI is the leader when it comes to the number of users. They have become synonymous. ChatGPT is AI, right? You ask someone that isn't using AI every single day, be like, hey, have you ever used AI? They're going to say, oh, you mean ChatGPT? It is synonymous with AI. And that's because they have 900 million weekly active users. And the last revenue estimate we got was $25 billion of annualized revenue. So this is the consumer default. But what most people don't realize, they're also the enterprise default as well, just because they've done a great job at converting those individual weekly active users on the consumer side, onto the business plans and the enterprise plans. Google is all over the place in maybe a good and bad way. Right. So the latest reports that we got were 750 million monthly active users. Okay. Different from the 900 million weekly active users that we got from OpenAI and 8 million paid enterprise seats. And I'd say top to bottom, they're the full stack leader. All right. That doesn't mean they're the best. Right. But I mean, they have Gemini everywhere, right, in AI mode. So there's a good chance that you're using Gemini all the time and you don't even know it. I believe it's 3.1 Flashlight that powers the AI overviews and the AI mode. It's coming out everywhere, right? Especially if you use Gmail, if you use Google Workspace. We'll get into more of that later. But from everywhere from Vertex using Gemini in their Google AI Studio, it's legit everywhere. All right, Microsoft. It's weird to say that Microsoft might be in third place when it comes to the enterprise, even though they had an unfair head start, right? They launched Microsoft Copilot in 2023. They were the first enterprise product for AI, right? And according to reports, you see about 15 million paid Copilot seats. And according to those reports, 90% of Fortune 500 penetration. But they are kind of the default enterprise control plane, right? Most companies I talk to, right, when they hire us for front-end AI strategy or to train their teams on ChatGPT or something like that, Most of all, or mostly, they're coming from Copilot, right? And what we've seen, I think, a lot more recently is companies that, well, they're still going to use Copilot because the company pays for it, but they're trying to move some of their operations in there. And I know Microsoft is doing a lot of work to counteract that in some of their more recent moves that we're going to talk about later. Anthropic, last but not least, reportedly $20 billion in annual recorded annual recurring revenue. So not too far behind OpenAI's reported $25 billion. So even though they're much, much smaller, they are doing a good job, at least on the revenue side. We haven't seen a lot in terms of total users from Anthropic. I think it's because, you know, OpenAI and Anthropic are both reportedly going public in 2026 with an IPO. I would assume Anthropic keeps their overall enterprise numbers under wraps because it is very small. And I'm just going to go ahead and say this now, right? People always accuse me of being anti-Anthropic. No, I'm not. I love Anthropic. I have a, you know, $200 a month Anthropic Max plan, right? There's obviously great applications for Anthropic, right? But there's a lot of information out there that's bad and wrong, right? So I'm just here to set the record straight. I currently have also Yeah, I currently have no, you know, no sponsorships or advertising with any of these four, although I have— we have advertised with Google and Microsoft in the past on this channel. Right. There's a lot of bad information floating out there that says Anthropic is winning the, you know, the enterprise race from RAMP and Menlo. They're not. It's not even close yet. They're in fourth place. But what they've done a great job. They've built great products, fantastic models. And I would assume, right, they probably have the highest revenue per user because they don't have a lot of users, right? They're, they're nowhere in the, you know, 900 million or 750 million range that OpenAI and Google are. But great models, great products. And like I said, they're probably doing the best in terms of revenue per user, which is an important metric to look at. So kind of the, for our livestream audience, you can kind of see the little graph here. If you're listening on the podcast, nothing overly visual on today's show, but you can always watch the video version on our website at your everyday ai.com. And essentially there's a lot of overlapping races, uh, because ultimately I don't think that there technically is one race because you can lose the, you know, the consumer mindshare, uh, like Anthropic is, and you can still win the overall race, right? But if I had to kind of, you know, peg each one, what is the lane that they're running in. So OpenAI, I'd say they're the consumer default, right? They are synonymous with AI. They are AI, right? Even though they didn't create it, right? You gotta tip your cap to Google there and the inventors of the transformer. But they are so far ahead of everyone else. I think personally, obviously, the numbers tell it on the consumer side. The numbers seem to tell it also on the enterprise side.. And when it comes to the models, the benchmarks, right, their newest models there as well. Full stack multimodal, that's Google. I think Google is in such a great position in the long run because they are the only one multimodal by default, accepts video input. They have probably the best source of training for the future of AI, like AGI, when it comes to YouTube. Right? And great on the robotics model side with their Gemini Robotics as well. So Google is, like I said, all over the place in a good and bad way. But I think they are primed to really excel in the long run. And I think they finally brought some updates to where people probably use Gemini a lot and had some bad experiences, which is in the workspace. And we're going to get to that in a little bit. Then Microsoft, they are, I'm just going to say, the enterprise control plane. Right? That's been historically dominated by Microsoft, right? Because they are the only ones. Well, they are the actual operating system of the computer until Apple ever decides they want to do AI, which they're not in the conversation, right? When you fire up your Windows PC, Copilot's there, right? ChatGPT is not living on your desktop when you fire it up by default. Google Gemini isn't, Anthropic Claude isn't. It is Microsoft Copilot. However, their models are powered by OpenAI and Anthropic, but that's kind of their lane. It is embedded workflow orchestration with the tools and the software that your company already uses. Huge on the identity and security governance side, obviously. And then Anthropic, I'll say they are the premium intelligence layer. So I don't think in the long run they're going to have the best models, although right now their current models, Opus 4.6 and Sonnet 4.6, are some of the best that there are. I don't think in the long run. I do think that's going to be a Google OpenAI conversation, but they have done a better job than probably everyone else at just kind of the premium intelligence integration. Right. So everything from coding agents and how you can use Claude code in different places to their recent, you know, plugins that have literally shook the stock market, right? Skills, you know, popularizing skills and MCP. So they are kind of the company that's inserting, you know, kind of their intelligence layer in more places in the high-value financial workflows and model-agnostic cloud deployments. So a little bit more in some of the strengths for each, right? OpenAI, let's get to them. They're the default. They're the, you know, no matter who you're comparing against, whether you're currently a Copilot, you're saying Copilot versus ChatGPT. If your company is a Google Workspace company, you're saying, okay, are we Gemini or are we ChatGPT? Or maybe one of your teams really took off with Claude and you were using it early on. It's Claude versus ChatGPT, right? ChatGPT and OpenAI, they're not going to be out of the conversation for anything, I don't think, in 2026 at least. Today, best model, GPT-5.4. Thinking, GPT-5.4 Pro. Scientific benchmark side, they're winning. Everything from the traditional kind of, you know, MMLU type, right? The ACTs for the AIs to now the coding, the agentic harnessing, the computer use. So some of these areas from a benchmark perspective that used to belong to Anthropic, yeah, it's OpenAI's, right? I could obviously change next week, right? Because anytime one of the big three come out with a step-up model, a decimal model, right? So going from a 3.1 to a 3.2 or a 4.6 to a 4.7 or a 5.4 to a 5.5, right? Anytime that happens, the benchmark and the sentiment changes, but OpenAI is never going to be out of it., right? They've got some huge CapEx deals, right? More than $1 trillion between SoftBank, NVIDIA, and Amazon, signing huge deals all the time. And when it comes to getting work done, you have to talk about the GDP bow, right? 83% win-tie rate. So in blind taste tests, so to speak, producing work, yeah, GPT-5.4 is better or ties expert humans 83% of the time. Best in the business. All right. Google, like I said, they're the multimodal full stack. All right. And they are all over the place. But according to Google, more than 120,000 enterprises are using their technology across the stack. More than 8 million paid Gemini Enterprise seats. And their year-over-year generative AI revenue is up 400%. So yeah, it was the end of 2024 that Google just woke up and chose violence. So at least the last like 15 months, I think Google has been the winner. Right. And then I think in terms of, you know, OpenAI versus Anthropic, they've been competing. But, you know, obviously Google got off to a very bad start when it came to their Bard.

AI moves too fast to follow, but you're expected to keep up. Otherwise, your career or company might lag behind while AI-native competitors leap ahead.

But you don't have 10 hours a day to understand it all.

That's what I do for you. But after 700+ episodes of Everyday AI, the most common questions I get is, where do I start? That's why we created the Start Here series, an ongoing podcast series of more than a dozen episodes you can listen to in order. It covers the AI basics for beginners and sharpens the skills of AI champions pushing their companies forward. In the ongoing series, we explain complex trends in simple language that you can turn into action. There's 3 ways to jump in. Number 1, go scroll back to the first one in episode 691. Number 2, tap the link in your show notes at any time for the Start Here series, or you can just go to starthereseries.com, which also gives you free access to our inner circle community where you can connect with other business business leaders doing the same. The Start Here series will slow down the pace of AI so you can get ahead.

Right. The Bard model wasn't good. They came out with some marketing that kind of showed some capabilities of Google Bard that weren't exactly there. And I think that really set them back by probably more than a year. But what they're doing on the multimodal and creative side is absolutely bonkers. I mean, when you talk about v0.3.1, again, these models might have changed by the time you listen, y'all. I understand that. But when you talk about their video models in v0, their Nano Bananas, you know, image generation model, what it can do with infographics and slide generation, right, on the Gemini robotics side, I mean, their footprint is larger and wider than even Microsoft's, even though Microsoft has that huge head start of just living in the operating system. All right, speaking of Microsoft, like I said, they are the enterprise control plane. All right, so we've seen reports more than 15 million paid Microsoft 365 Copilot seats, and they did show a 160% year-over-year growth. And they did just recently launch a new offering. So we'll see what that does on the revenue side. It's called the M365 E7 bundle. It's kind of a higher tier with some more inclusions than your basic Microsoft 365 365 Copilot seat, as well as the Copilot Cowork, which we're going to talk here in a little bit. But last but not least, you have Anthropic. And I do think that they've come to dominate that premium execution layer. So essentially inserting their intelligence in different ways, because I think that they've been developer friendly and developer focused from the beginning. So when it comes to large enterprises, a lot of their teams that maybe were helping make the decisions were maybe more comfortable with some of the earlier Opus and Sonnet models. Obviously, they were kind of first to the game with Claude Code, even though, yes, Microsoft, you know, Microsoft GitHub Copilot has been around for a very long time. But when it came from a perspective of an agentic command line interface coding tool, that came out to a desktop program, even though I think that was still Microsoft GitHub Copilot should have, could have been that tool. But for whatever reason, it was Anthropic, and that's helped them get extremely, uh, great traction. Yeah, so reports we've seen anywhere from $14 to $20 billion on the annualized run rate, uh, run rate, same thing, investing a lot on the CapEx, building their own actually, um, and the product footprint, everything from Cloud Code which is on itself a unicorn, right? Reportedly $2.5 billion in revenue just from Cloud Code. They have the new Code— sorry, Cowork, which we'll talk about on the Microsoft side as well, which is, you know, allowing you to do your work on your desktop. So it's kind of like Cloud Code for non-developers. The new Cloud Marketplace, the Office add-ins, the plugins, right? Anthropic over the past probably 3 months has been their best 3-month run to date. All right. And then we have to talk about equity investments because that's where this race gets not tricky, but the answer is nuanced because some of these companies might not be competing in areas where they could well, because they already have their money and their equity in someone that's winning. So as an example, as of October, when OpenAI converted its operating structure from a nonprofit to a PBC, or a public benefits corporation. Well, Microsoft is the single largest shareholder of the OpenAI Group PBC with a 27% stake in the company. They, yeah, they are the technically the largest single shareholder in the PBC. Also, Microsoft has a total investment commitment of $5 billion in Anthropic. So Microsoft has a lot to gain even if they are losing consumers, right? If they have, you know, if they're shedding Copilot licenses and those people are ultimately going to ChatGPT or to Anthropic, well, maybe it's a dime out of the left hand, but a dollar in the right hand or vice versa, right? But ultimately, as long as they're not losing to Google, Microsoft is winning. I think a lot of people are overlooking that. Google also big investments. So in reportedly about a 14% to 15% ownership stake in Anthropic. So where Microsoft has a very large stake in OpenAI, Google parent company Alphabet has a pretty big slice of that Anthropic pie, which from a revenue-only side is growing faster, uh, according to reports than anyone else's revenue pie. All right. So let's get back to OpenAI and tell the story a little bit that way. So ChatGPT went from 0 to 900 million weekly active users in just over 3 years, which going from truly 0 to nearly a billion, right? Probably by the time most people are listening to this, it's going to be a billion weekly active users. It's never been done. It is just straight up hockey stick to the moon growth. Also, revenue finished with $2 billion in 2023. Now very quickly up to that reported $25 billion annualized revenue rate. And it's not just consumers. Yes, the overwhelming majority of those $900 million are free users using ChatGPT, but they do have $50 million consumer subscribers and 9 million different businesses that now depend and use and use the platform. And I'd say kind of my take here, ChatGPT is the easiest to use. It is. All right. I kind of wish they had their Codex, their new coding platform. I kind of wish it worked inside of ChatGPT, whether you're using it on the web or the app. That's the thing, right? I think Anthropic has done a great job. With their desktop app for Mac as an example, you know, having the chat, co-work, and code all under one roof. So I do think that was maybe a miss there from OpenAI, not just kind of integrating Codex a little bit more tightly into ChatGPT, specifically when it comes to the desktop app. But it is still the easiest to use by far, right? Anthropic, yes, it's a little— it can be a little confusing., but I'd say Anthropic is probably the second easiest to use. But the UI, UX, ease of use, right? If, if your grandparent or parents are using AI, they're probably using ChatGPT because it is the easiest. And like I said, it's become synonymous. Uh, and the new model I think is, uh, really, really, really good. Right? I did a recent show on it, so you can go check that out if you want to. But I mean, it introduced native computer use. Spreadsheet modeling, and direct financial data integrations. The trade-off, well, the cost can get high, right? Especially on the enterprise side. You know, they don't publicly disclose those, but, you know, we hear anything from, you know, $50 to $70 per enterprise seat, right? So a little more than Microsoft's baseline enterprise seat licensing. Obviously, their new $99 OpenAI Enterprise one will come under that., but, you know, the, the trade-off there, and I've been saying this, OpenAI, I don't think cares about being profitable, right? They've updated their profitability projections to say they might not be profitable until 2030. So essentially you've had, you know, Anthropic and OpenAI, the two kind of quote unquote AI startups, take a very different approach. Uh, OpenAI has said we care about users and they're winning that game by far. Um, and I think Anthropic, has cared probably more about profitability. Again, we don't know the price per user, but my strong assumption is Anthropic is crushing everyone else when it comes to revenue per user because I don't think they have a lot of users, but I do know they have a ton of revenue. All right. So their biggest strength might also be their biggest weakness, right? OpenAI kind of had their infamous code red where they kind of realized that they maybe they were lagging behind a little bit on the model side. So I do think over the last month or two, they've really corrected course on that. But yeah, a lot of people argue that maybe OpenAI is a little too distracted, right? So they have their video product. You know, it seems like they're trying to compete maybe with Google and Microsoft and Apple all at the same time. Could they do it? Maybe., right? Whereas Anthropic, its closest competitor, is not doing any of that. They're not doing images like OpenAI is. They're not doing video, right? OpenAI reportedly going into consumer hardware with devices maybe, you know, early 2027. But the consumer focus maybe has slowed down some of their model progress, some of their research, because they're having to devote a lot of their compute to things like this Sora app, right? Things like that. They are spending reportedly more than anyone else. But at least they are winning on the overall numbers game. All right. Anthropic, I think, is the best for high stakes, high value technical work. Emphasis on technical, because I do think if you're looking at overall general work productivity, I still do think that OpenAI and Google are probably the best for that. But Claude is a leader on complex reasoning, long documents, and economically valuable knowledge tasks. And they do say that 8 of the Fortune 10 are Claude customers, and there's 500+ companies that are spending over $1 million annually. So the product surface is very narrow. Anthropic, you can make the argument, is a company that is extremely focused on doing one thing very, very well, and that's having very capable models that have a product-market fit in usually niche areas, but now they are starting to broaden out. I think toward the end of 2025, um, again, Anthropic has shocked me because they were really just geared toward the software development, uh, the, the devs, uh, right, highly technical financial tasks, and they have really branched out. I think Claude Cowork, um, you know, was one of those things., but really pushing skills, uh, pushing the plugins. Uh, I do think that they're broadening out from maybe, you know, the, the, the technical, uh, model of choice to now they're probably, uh, you know, starting to win consumers over, uh, as they've branched out a little bit. And I think Claude Code is one of the reasons why. Claude Code, really good. Uh, personally, um, I do like Codex, uh, better, uh, but Claude Code has been a massive hit. You could make the argument it could be a top 5, you know, AI product of all time. I mean, just Claude Code in, you know, less than 6 months came into $2.5 billion in annual recurring revenue. And that's right now 4% of all GitHub commits, which is a pretty big amount for a product that is this new. Also, they just launched their Claude Marketplace. With 6 different partners where you can kind of, uh, companies using that can commit their spend, uh, to some of their partners, um, cloud partners in their, uh, kind of plug-in ecosystem. All right, Microsoft. Microsoft, I'd hate to say it, they are the safe, uh, kind of, you could say, old school AI, right? They're the safe AI, the safe bet. And, and one of the reasons is, well, it's the one that's approved by most enterprises out of the box because it is literally built into the ecosystem. It is built into the operating system. So it's already embedded. According to reports, more than 90% of Fortune 500 companies are using Microsoft 365 Copilot daily. A big new shift. Well, Microsoft announced Copilot Cowork powered by Anthropic's Cowork technology. And it is going to be probably a slow rollout over the next few quarters until the wider business world has their hands on it. But this does bring that multi-step task, you know, where essentially it's just doing your work, right? If you've used Claude Cowork, it has access to all your files on your computer. It can produce documents. It can use a terminal. It can power a power and control a browser, right? So it is very much like an intern or a junior researcher. It can access all your files, create files, use a computer, use a terminal, use control a browser, right? All these same things that a human can. So I do think that this co-work kind of movement now that Microsoft is starting to productize as well is one worth paying attention to. All right. Microsoft, their biggest advantage is one that no one else can really replicate. That's AI by default, right? Like I said, even if people maybe don't want it, they're going to see Copilot inside of Word, Excel, Teams, and Outlook. Although I do think it's gotten much, much better. I think Microsoft probably had some of the same early problems as Google when, you know, they were slapping their Copilot in all of their, you know, all of their Office products and it wouldn't always work very well. I think they've gotten that tied up in the last probably 6 months. And now that they have kind of the the agentic options inside of their Office products even more so. And we'll see what happens with this new $99 per seat per, for the E7 Frontier Suite that kind of bundles AI security and governance and as well as Copilot Cowork. But the biggest setback for Microsoft, it is the learning permissions and governance curve. It is way higher than the other three. That's because, well, it's there by default. So it is, it's kind of like a running joke that Microsoft Copilot could be the best AI out there if anyone could figure out how to use it. I do know that that's something Microsoft is tackling. We've seen reports that their CEO, Satya Nadella, is essentially, you know, putting in some work as a product manager on Copilot, which is very rare for the CEO of a multi-trillion dollar market cap company to start working on product. But that's what we've seen. So I am actually rather bullish long term on Microsoft Copilot for that very reason. Right. When you have the CEO rolling up their sleeves to say, hey, this isn't working, we've got to change this. I think that's the way to go. All right. And then last but not least, Google, the most powerful multimodal AI ecosystem on Earth. So like we said, 750 million monthly Gemini users., with 8 million of those reportedly paid seats and huge year-over-year growth. And they are the only major provider with native video input, which is big. I still— people still aren't using this, and it's such a cheat code, alongside text, image, and audio handling. And there's deep integration literally everywhere. So across Search, Android, Chrome, YouTube, Drive, Docs, Sheet, right? So if you are a, uh, Gemini— I'm sorry, a Google, Workspace organization. Well, it's gotten a lot better. Again, early, you know, mid-2024 when you started to see, you know, Gemini rollout in beta across Workspace, it wasn't good. There's actually some instances where it was downright bad, right? I remember doing a couple of videos on our YouTube channel just being like, yo, this doesn't work. Now it's the best, right? An update they actually had this week. So in March 2026,, right? Like certain benchmarks, like, you know, spreadsheet editing. Now they're the best. But it's all, it is all over the place. So again, kind of a good and bad thing, but you have your Gemini. There's a different version of Gemini for business and enterprise, NotebookLM, Gemini and Vertex AI Studio. It's everywhere. But one of the newer things is the new workspace overhaul. So this fills, you know, it puts Gemini in everywhere. In Gmail, in Drive, in Docs, right? So you can essentially talk to Gemini now anywhere inside Google Workspace, whereas before it wasn't quite as easy. So now it can just, you know, synthesize and understand your files, emails from anywhere, right? So you can essentially leverage and tap into your entire Google Workspace ecosystem no matter where you are within Workspace. And now Google Drive is also being kind of repositioned as like a RAG database, right? It's an active, citable AI knowledge base across your entire history inside Google Workspace. Now, shifting away a little bit here, I think what we have to pay attention to is the agent coding war. I think that is maybe where this thing starts to get decided. Because coding is the first use case where AI created a concretely measurable ROI at enterprise scale. And gains in coding and computer use, I think, do spill over into non-technical applications in large language models. Because, I mean, we've heard from both Anthropic, we've heard now from OpenAI as well, that their models are writing their future models, right? Which is also kind of scary, right? We have this self-learning and recursive AI that's writing itself., right? But once that happens, that means that then those— yeah, I know this sounds weird, but then those models can write hundreds or thousands of specialized models that are then just better at all these other tasks, maybe on the non-technical side. So you do have to pay attention to the agent race, and the computer use race, because those are things that are ultimately going to impact the non-technical everyday business work that most knowledge workers are doing. I'd say right now Claude Code is leading in adoption., and it's trending. But Codex, I think, is probably the hottest on the market, along with the new GPT-5.4 models inside Codex. I do think it's— yes, it's way slower, but my personal opinion of burning billions of tokens in Codex and in Claude Code, I'm maxed out on all those. I am way more bullish, and I do think that it has the higher bar right now. Codex does. And then Obviously, you have Microsoft GitHub Copilot that is kind of like one of the OGs when it comes to, you know, AI coding assistants. And Google, again, both great and confusing. Google's kind of all over the place here. They have everything from Firebase to Jules to Code Assist to the Gemini CLI to Antigravity. Right? So Gemini, obviously, on the coding side, has a lot of offering. But I do think the coding winner could set the template for every agentic category that follows. And also the infrastructure side, right? You have to see where these companies are investing, uh, on CapEx. So OpenAI, uh, reportedly, yeah, $110 billion, uh, funding round targeting $600 billion in total compute spend through 2030. Yeah, they're saying we're spending a ton of money on compute, and that's one of the reasons they're saying they're not gonna be profitable, uh, you know, or not as profitable as quickly as maybe some of their other competitors. Google spending up to $185 billion in CapEx, Microsoft $37 billion in Q4 2025 alone. Anthropic has launched a massive $50 billion investment initiative to own and construct their own AI data centers in the US. So that's technically kind of separate, right? But the faster that these companies can get up these CapEx, you know, AI data centers, right? That's going to help them get ahead in the kind of the day-to-day, in the week-to-week AI race. All right. So as we wrap here, how do you navigate this? All right. I gave you a very quick— I know it's 37 minutes in here. I gave you a very quick recap of where we've been over the last 3 years and the pros and the cons of where these Big 4 companies are. And yeah, it's the big 4 companies. Maybe Meta will reenter the discussion. Perplexity is not really its own model maker, right? Because I'm sure people are going to be like, oh, what about, you know, Amazon AWS, right? They're second tier right now. That could change, right? But right now it is essentially the 3 frontier AI providers and Microsoft that leverages both of them and Microsoft making their own models. But for the most part, stop buying one vendor for everything. Uh, you know, leadership changes by workload and by the quarter. Even if you have a great workflow right now with one of the Big Four, you need to be working on your 1B plan because I can't say this enough. Oftentimes these companies are going to change something in their big model and they might not even announce it. It might not even be a decimal point change. It just might be an under the hood change and things could go off the rails for your company. You always need to be building modularly. You need to redesign processes for agents first that take action, not just chatbots waiting to be asked. So that's a big, kind of change management piece for enterprise leaders out there. And you need to match the platform to the problem. Right? So as an example, maybe right now it's OpenAI for breadth, Claude for depth, Google for scale, Microsoft for governance, whatever it is. There's no problem in having, you know, two different, although I do think it is best to move as many of your day-to-day knowledge work processes into an AI operating system. But I think for specialized tasks and for certain departments, I think it's okay to have a secondary, you know, tool or a secondary company. So let's end it here. Who's going to win the AI race? All right. I'm going to break it down because like I said, there is no one race. It's many different races. But I think there's a couple, there's 3 big races I think that are being run. Consumer, enterprise in models. Consumer, OpenAI. No one else is going to touch them. Google is starting to close the gap. OpenAI is so far ahead. I don't see them losing that lead anytime soon. Enterprise, I do think, as strange as this sounds, this has always been Microsoft's battle to lose. And I think that from 2023 to 2025, they were losing. So I think right now it's Microsoft versus OpenAI. I do think Google is trending. Google has a fantastic Gemini for Business and Gemini for Enterprise product, but I don't think anyone knows about it, right? People think it's just the Gemini.google.com. No, that's the consumer version. I think Google could start to dominate in that category, you know, at least when you talk about non-technical front-end users. Obviously they have everything on the Vertex side, the AI studio. But yeah, Anthropic, not in the picture. Sorry. They're not an enterprise player when it comes to enterprise, total enterprise users. All right. And then model capabilities, that's a toss-up. Right. Obviously Anthropic's models, top tier, top of the class. OpenAI's new releases, you know, in certain areas, great benchmarks as well. And then Google. I think I've said it before, Google at any time, because of the mass training data that they have, I think anytime they can come out with a new Gemini 3.2, Gemini 3.3, and be the best model in the world. So I don't think, although we always are watching that race, the model race, I think it's the least important of all three because it's becoming more and more important, the harnessing, the tool, the tool use. And like I said, it's not about who has the best model. It's about who helps you bring your data and all of your day-to-day workflows agentically into the, into the picture. That's what it's ultimately about. All right. I hope this one was helpful as we went over the state of the AI race. Again, this is part of our Start Here series. So whether you are brand new to AI, or you just want to double down on your knowledge, please go to starthearseries.com and go sign up and go listen to the entire series there. So thank you for tuning in. Hope to see you back tomorrow and every day for more Everyday AI.

Thanks, y'all. And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit youreverydayai.com and sign up to our daily newsletter so you don't get left behind.

Go break some barriers, and we'll see you next time!