(Preview) Microsoft’s Plan for Platform Survival, Meta and the Market’s Permission, A Lack of Situational Awareness

Summary

Sharp Tech with Ben Thompson

(Preview) Microsoft’s Plan for Platform Survival, Meta and the Market’s Permission, A Lack of Situational Awareness

Date: August 6, 2026
Hosts: Andrew Sharp & Ben Thompson

---

Episode Overview

In this episode, Ben Thompson and Andrew Sharp dive deep into the recent earnings of big tech companies, focusing primarily on Microsoft's strategic positioning in the rapidly evolving AI landscape. They explore how Microsoft's current approach is shaped by both necessity and opportunity, comparing it to its hyperscaler peers (Amazon, Google, Meta) and drawing historical analogies to IBM’s survival strategy in the 1990s. The discussion expands to address the existential threats facing digital incumbents, the market's reaction, and the meta-game of AI platform dominance, concluding with pointed observations on industry open letters and notable anxieties among tech giants.

---

Key Discussion Points and Insights

1. Setting the Stage: Big Tech Earnings & Market Context

• Microsoft’s recent success: Microsoft saw a record one-day market capitalization increase ($450 billion, up 15%), reflecting market confidence (01:19).

  • Quote (Ben): "The market is very happy with Microsoft. The company was up 15% and added $450 billion in value in a single day last week. Biggest single day jump in market history, according to Bloomberg." (01:19)

• Comparison of Hyperscalers: Each 'hyperscaler' (Amazon, Microsoft, Google, Meta) faces different levels of AI threat and has different core strengths:

  • Amazon: Least threatened, with robust e-commerce and cloud business; used to tight margins.
  • Microsoft: Similar to Amazon in cloud, but its software business is directly threatened by AI.
  • Google: Straddles the line with a strong cloud business but a core search product highly vulnerable to AI disruption.
  • Meta: Striving to stay on the AI frontier, though its core social platform may be less directly threatened (04:18–07:00).

• Digital vs. Physical Risk: The more digital the business, the greater the existential AI threat; physical businesses like Amazon’s logistics are relatively safe (06:05).

---

2. Microsoft’s “Middleware” Strategy: Lessons from IBM

• Middleware Aspiration: Microsoft aims to become the operating system or “middleware” layer between enterprise customers and the rapidly advancing AI models. This is reminiscent of IBM’s strategy in the 1990s.

  • Quote (Ben): "With Microsoft specifically, they want to be middleware, is that right? And to what extent is that a function of necessity? Because that's the only option that makes sense for them at this point." (07:02)
  • Quote (Tech Analyst/Ben): "What we're good at is being big. And actually being big is useful because you're kind of good at everything, but you're not really great at anything or you're acceptable at many things." (08:20)

• Historical Analogy: Comparison to Lou Gerstner’s IBM – where being “big” and offering acceptable solutions across the board bought IBM decades of survival in spite of a lack of deep product differentiation (08:20–09:57).

  • Quote (Tech Analyst): "Once you've been a monopoly, you've kind of lost the ability to be great, because you get fat and flabby and life's too easy [...] The analogy there was, look, this is the Microsoft path." (09:46–09:57)

• Nadella’s Leadership: There’s growing anxiety and urgency in Microsoft's executive communication, notably with CEO Satya Nadella's personal involvement and “meta messaging” about AI threats (11:22).

  • Quote (Tech Analyst): "[Nadella] felt the need to, I need to take a much firmer hand and much more control of this company because we're actually in bigger trouble than people realize." (12:35)
  • Quote (Ben): "He's clearly spooked by the moment and trying to counter the anthropic messaging." (11:22)

---

3. The Core Threat: AI’s Impact on the Software Business

• Direct Threat to Productivity Apps: Emerging AI agents (from OpenAI, Anthropic) could supplant core Microsoft offerings (e.g., Office, Teams) by taking over routine workflows and knowledge work (13:20).

  • Quote (Tech Analyst): "Anthropic and OpenAI are coming for Microsoft's business, like day to day productivity. They want your agent to do all that. They want your agent to do your email, your agent to do your doc. Why do we have documents? Like, why can't we just ask the AI what it is?" (12:52–13:20)

• Long-Arc Changes: Tech threats play out over long periods—what looks “meh” in the present can become existential over a decade or more (13:37–14:40).

  • Example: Intel’s decline was predicted years before it visibly manifested in financials.

• Metadata and Data Moats: A recurring theme is the value of usage metadata collected by leading AI providers—the more a company builds on OpenAI/Anthropic infrastructure, the more potentially strategic information is surrendered (15:14–16:58).

  • Quote (Tech Analyst): "The important part is all the meta around it, what tools are you using, what things worked, what things didn't. And if you're relying, if you're using their user facing software, you're giving away the crown jewels of your company." (16:13)

---

4. Microsoft’s Platform Gamble: Processors Not Monopolies

• The New OS Metaphor: Microsoft seeks to “reduce models to being processors”—AI models as commodity chips, with Microsoft’s middleware/harness as the orchestrator (17:02).

  • Quote (Tech Analyst): "They want to recreate the world where Microsoft Windows is the layer everyone uses. And then underneath that is an intel chip or an AMD chip or whatever it might be...They’re seeking to reduce the model companies to being processors." (17:02–18:18)

• Modular vs. Integrated Approach: Microsoft's strategy appeals to companies seeking flexibility (multi-model, model-agnostic approaches), but this could result in “lowest common denominator” offerings versus the bespoke acceleration available by fully harnessing a single frontier model (20:04–22:13).

  • Quote (Tech Analyst): "There's a thing where you can feel this with the models...as the models got better, they sort of internalized more and more of the functionality that came from the harness." (22:39)

• Harness Over-Engineering Risks: Rapid evolution in AI models means heavy middleware investment can quickly become obsolete or counterproductive (23:08).

  • Memorable moment: Discussion of “vibe coding” and how iterative model improvements can make sophisticated harnesses unnecessary or even problematic.

---

5. Industry Responses & Meta’s Position

• Open Letters and Industry Anxiety: The hosts joke about the surge in industry open letters, seeing it as a defensive, even fearful, signal from incumbents (24:38–25:42).

  • Quote (Tech Analyst): "The fact the whole industry is lining up to sign this letter is not a indicator of strength for the company signing the letter. You don't sign open letters because you're winning. You sign open letters cause you're freaking terrified." (25:24)

• Nvidia’s Fear Despite “Winning”: Even Nvidia, today’s chip king, feels the heat: in a future where only two companies (OpenAI, Anthropic) dominate model consumption, Nvidia risks being commoditized (26:54).

  • Quote (Tech Analyst): "Nvidia feels like they're in the catbird seat. Everyone uses Nvidia chips. Why are they writing an open letter? Because in a world where there's only two companies that buy chips...Those margins are going to get compressed a lot." (26:54 – 27:04)

• VC & Startup Implications: If AI models become universal utilities, and any successful startup is instantly replicated, the VC/startup ecosystem could be profoundly disrupted (27:21–27:44).

---

Notable Quotes & Memorable Moments

• Microsoft vs. the Model Frontier:

  • "What Microsoft is proposing to do is they want to reduce models to being processors. They want to recreate the world where Microsoft Windows is the layer everyone uses." (17:02)

• On Market Anxiety:

  • “You don't sign open letters because you're winning. [...] You sign open letters cause you're freaking terrified.” (25:24)

• On Incumbent Blindspots:

  • “What these models can do is so far beyond what I think most people can appreciate. But someone like Satya Nadella gets it.” (25:42 – 26:32)

---

Timestamps for Important Segments

• Microsoft’s historic earnings jump and contextual intro: 01:19
• Hyperscalers: Comparing Amazon, Microsoft, Google, Meta: 02:04–07:00
• IBM analogy & middleware strategy: 07:16–10:00
• Satya Nadella’s tone and urgency: 11:22–12:35
• Direct threat to Microsoft’s software productivity business: 12:52–13:20
• AI platform positioning—models as stateless processors: 17:02–18:18
• Integration vs. modularization narrative: 20:04–22:13
• Harness over-engineering and model advances: 23:08–24:12
• Industry open letters & signs of incumbent fear: 24:38–25:42
• Nvidia’s position and future risk: 26:54–27:04

---

Language & Tone

The episode’s tone is conversational, witty, analytic, and occasionally sardonic. Both hosts rely on zippy analogies (IBM elephants, vibe coding), deep strategic context, and unvarnished assessments of industry self-interest and anxiety.

---

This summary captures the core analysis, strategic insights, and color commentary of Ben Thompson and Andrew Sharp, providing a rich guide for listeners and non-listeners alike to Microsoft’s AI survival gambit and where it fits in the broader tech battleground of 2026.

Transcript

[Host]

Hello and welcome to a free preview of Sharp Tech.

[Ben]

Hello and welcome back to.

[Tech Analyst]

We are in Person. This is Sharp Tech.

[Ben]

Welcome back to SharpTech.

[Tech Analyst]

When I get the Zoom open or Riverside open, cracks me up every time.

[Ben]

Are you live and in person? I'm good. It's great to be in Madison. It's balmy outside relative to what it was in jail. January and we're here for some company meetings. There's an AI show and tell that we have planned that I'm already pretty stressed out about. So.

[Tech Analyst]

Because you have no AI in your life, are you gonna. Are you gonna demonstrate looking up NBA stats?

[Ben]

Yeah, demonstrate copy editing Sharp text posts and whatnot. You guys are on a whole other level. We're also going to a Brewers game later today, which I'm very excited about. Will be my first baseball game in probably five or six years. So here we are, you know.

[Tech Analyst]

Well, you just asked me are we going to tailgate? And I was just flabbergasted.

[Ben]

A total noob east coast elite question, of course.

[Tech Analyst]

We were just talking about tailgating on. We were talking about tailgating on Sharp or on dithering. And I was expressing how east skaters say they tailgate. And I find it very sort of underwhelming. Something in D.C. they don't even bother.

[Ben]

So don't even bother. Just show up at game time. Absolutely. Well, for now, before the tailgate, we are going to review big tech earnings which came down last week. And we're. We're going to start with two companies that are encountering very different reactions from the market. First up is Microsoft. The market is very happy with Microsoft. The company was up 15% and added $450 billion in value in a single day last week. Biggest single day jump in market history, according to Bloomberg. And Microsoft, they are not pushing toward the frontier whatsoever at this point. They haven't been since 2024. What do you think of what they're doing instead?

[Tech Analyst]

If you zoom out, I knew earnings would be last week. I would come back, have all these to sort of digest. And I think all three updates this week were actually interconnected. I argue we should have done like one big post trying to tie the omnibus. Okay, yeah. But I was trying to look at every different. Hyperscaler is slightly different in where they are in terms of do they have a cloud business? Like, you know, where are they at relative to the frontier? What's their focus there and what is their overall threat from AI? So the one I didn't get into as much as I wanted to is Amazon. I just. We're gonna get to them in a little bit. But Amazon's on one extreme. They have a total cloud business and I think their core business is not very threatened by AI. Yeah, like the AI is not delivering

[Ben]

my pack durable physical infrastructure for Amazon

[Tech Analyst]

is a good place to be. Yes. So I think Amazon's like their core business. Their, their E commerce business, I think is fine, arguably better than ever. And their cloud business, they're, you know, obviously doing very well. And Amazon is also sort of a company that is more accustomed to okay with tighter margins, which I think is one of the concern about hyperscalers in the long run. Once compute is finally everywhere, who's going to have the, you know, our margins are going to compress. They've always thought about having more. Right. We talked about that sort of last episode. So Amazon is sort of one extreme. Then you have Microsoft, who is taking a strategy that's similar to Amazon in some respects, particularly in the fact they're not on the frontier. They're building their own models, but they're not building models that are going to be as good as the models from OpenAI or from Anthropic. But they're serving those models.

[Ben]

And they got off that train a couple years ago.

[Tech Analyst]

I mean, they weren't really on that train because when they're on the train, to the extent that they were supporting OpenAI.

[Ben]

Yeah.

[Tech Analyst]

But then what happened? The way to think about what happened with Microsoft and OpenAI, Microsoft is like to. OpenAI will keep your inference business. So the business where it's actually customers using your models, your training business, you got to go find someone else.

[Ben]

Yeah.

[Tech Analyst]

And so that was the whole shift to talking to Oracle, to talking to SoftBank, building these mega sort of data centers. And, and yeah, that's the train that they sort of got off of. Their software business is, I think, very threatened by AI. I mean, the, the, you know, we're going to talk about vibe coding a little bit. This will be part of AI show and tell. But we are not representative of a large enterprise, to say the least. And there's an aspect of us experimenting with AI astrotechery is important to just understand the product. Yeah. But in a very tangible example, we were mostly still using teams for one specific use case, which was podcast workflows. We're totally off. Like we made our own software right now. Again, that's a very small example, but that is, that's what you could imagine that writ large, at least in the fullness of time. So they have A cloud business that in some respects is similar to Amazon, but they have a core business that is very much threatened. You move over. Now we're at Google. Google obviously has a big cloud business. I'm sure we're gonna talk about that in a little bit. They are also seeking to be on the frontier, and they do have a business that's threatened by AI, which is sort of search generally. And we've spent a lot of time talking about that. I think there's more to even say about it than I wrote about this week in that regard or I mentioned a couple things in passing. So we'll talk about Google a little bit. They're kind of in the middle. Then you get to Meta. Meta is. Meta is maybe between Amazon and Google in that they're seeking to be on the frontier. They don't have a cloud business. That's the difference. And they are threatened by AI, but I think less acutely than Microsoft, than Microsoft, or even than Google. You can see very clearly how AI is a threat to search. It's harder to see how AI is a threat to social networks or entertainment or whatever it is. But my sort of thesis is, number one, it takes up a lot of time, which is a detraction. So anything that takes up time is a detraction from meta in the long run. And just my overall thesis is anything digital is fundamentally threatened by AI. Like the opposite of the Amazon thing, where physical. The more physical you are, the safer you are, the more digital you are, the more threatened you are. And so that's sort of like the gradient of these. Of these entities. And so I thought it was interesting to talk about all four of them this week, kind of considering those aspects, even if I wasn't super explicit about it, and what the differences are between them, how the market has responded to them, all these bits and pieces. So that's the overall context of Microsoft. Yeah, Microsoft sort of, in this case.

[Ben]

Well. And so with Microsoft specifically, they want to be middleware, is that right? And to what extent is that a function of necessity? Because that's the only option that makes sense for them at this point.

[Tech Analyst]

Yeah. So the way to think about Microsoft is exactly that. And it's funny because I wrote an article a few years ago comparing Microsoft to IBM, and it's actually one of my favorite articles. I remember it was very. Some articles stand out because they were so painful to write. I remember I wrote this one at an Airbnb in Madison before we had a place here. And I remember it took me hours to get it Out. And I posted it at like three in the afternoon instead of like six in the morning when I wanted to. But I went through Lou Gerstner, who says elephants can't jump or something like that, like his autobiography. And I was talking about the issue IBM faced in the 90s, where they're this behemoth that seems way behind the times. Investors are demanding they break up. And he went the opposite direction. He's like, we should not break up. You don't understand. We're not actually good at anything. All these pieces you want independent are not gonna do well in the market.

[Ben]

They'll all fail independently.

[Tech Analyst]

What we're good at is being big. And actually being big is useful because you're kind of good at everything, but you're not really great at anything or you're acceptable at many things.

[Ben]

That certainly describes my Microsoft experience over the last 20 years.

[Tech Analyst]

Well, so this was IBM in the 90s. And so what Gerstner did was he. The Internet's coming along. You have all these big companies that are like, how do we handle the. What do we do with the Internet? And IBM's like, we'll take care of it. And what IBM did was they built up a huge consulting arm, which, by the way, Microsoft talked a lot about building up their. They actually had a name for it. Like, they've always had a big sales force, but, like, the whole Ford Deployed Engineer thing. I should look up what the name was. It came up on the earnings call.

[Ben]

I mean, that does make a lot of sense for the next 10 or 15 years here.

[Tech Analyst]

Well, I mean, everyone's doing it for a good reason, right? Palantir really sort of has led the way here. And then also we're gonna build a bit. They built a bunch of middleware, which basically was. They built all these layers for companies to have websites and to have E commerce and to have all these bits and pieces that we talked about last week. There's still companies. There's companies that are on IBM because they've been on there since the 60s. There's a lot of companies that have been IBM since the 90s because they just built in all these layers to get you online, to get you, you know, all the things that you needed to do that. And it was a very smart strategy that basically gave IBM an extra 20, 25 years of life.

[Ben]

Yeah.

[Tech Analyst]

And. And the problem, the framing of that article was once you've been a monopoly, you've kind of lost the ability to be great.

[Ben]

Okay.

[Tech Analyst]

Because you get fat and flabby and Life's too easy. And what Gerstner did was realize, look, we don't have the chops to succeed by being the best at any individual thing. What we're good at is kind of doing everything because that's what we've been doing for a long time. And the analogy there was, look, this is the Microsoft path. Like, you're not going to succeed by being the best asset, the leader in your can kind of do everything and it's good, actually being big is your biggest asset. And I put that in the context of I'd been very, at that point, a strong endorser of Satya Nadella's approach to deprioritizing Windows being a services layer sort of. Yes, they were late relative to Amazon, late to the cloud, but most everyone else was late to the cloud. So they and all their customers were sort of coming to the cloud together. And I wrote that in the context of a keynote where he spent a long time talking about like pen or like their digital ink, like their ability. I'm like, what are you talking about? Like, you should not be. You are not going to win on product differentiation. You're going to win by kind of being good at everything. And, and I think Microsoft's continued down that path by and large. And you fast forward AI and now that comparison I think is becoming very, very tangible. Okay, so cited Al has been posting on Twitter like all these articles, like it's coming from my.

[Ben]

I know you said a couple of weeks ago he's coming for your job. He really is. He's very, very active on Twitter with these essays that he's putting out and they're always these long tweets and I wish he would put them on a website somewhere because I just can't read that much on Twitter. But he's clearly spooked by the moment and trying to counter the anthropic messaging.

[Tech Analyst]

Yeah, I mean, I get the sense that spooked happened maybe like a year or so ago. And you do get a one thing that's interesting. I was at the Build keynote in June and usually with a Microsoft keynote, Satya Nadella would do the first 10, 15 minutes, do the framing of what's going on and then he'd hand it off and then a parade of executives would sort of do the rest of the keynote. This keynote was also inadella and to the extent maybe I over part of my take this week is that actually I'm right to index on the meta of these things, what executives say actually matters and what I took away from that keynote is he felt the need to. I need to take a much firmer hand and much more control of this company because we're actually in bigger trouble than people realize.

[Ben]

Yeah.

[Tech Analyst]

And I suspect a lot of it was they have this whole software business, Right. Like that is very tangibly. You can make the software that Microsoft does now. Are you gonna make it.

[Ben]

We just did.

[Tech Analyst]

Are you gonna make it like actually useful and dependable and are companies gonna want to waste time maintaining it, et cetera, et cetera. All these are open questions and we've litigated that and we'll continue to litigate probably for years to come. But Anthropic and OpenAI are coming for Microsoft's business, like day to day productivity. They want you, your agent to do all that. They want your agent to do your email, your agent to do your doc. Why do we have documents? Like, why can't we just ask the AI what it is? Right.

[Ben]

Yeah.

[Tech Analyst]

And again.

[Ben]

And they're both very, very expensive for enterprises as well. And as those costs continue to rise, you could envision a future where enterprises are choosing between spending on Microsoft or spending on OpenAI as a platform or Anthropic as a platform.

[Tech Analyst]

Right. And again, all this, this is the thing about tech in general. Even right now, when stuff seems to be moving so quickly. One of the fun things about analyzing tech is how many decisions actually take years to play out. Right. We've talked about this in the context of chips. Right. Like intel was doomed. I wrote in 2013, Intel's in big trouble.

[Ben]

Took a solid 10 years.

[Tech Analyst]

Yeah. It took for that to actually manifest, but it was totally right. And intel actually needed to make changes. I thought I was too late in 2013, and I was, I think, too late. But it would have been better to make changes in 2013 than to wait till 2021 or whatever sort of, you know, Gaussier ripped the band aid off again, like there's pluses and minus to his tenure, but like there was an extent of certainly ripping the band aid off and saying we have to like shift our model over time. But this applies to software, to building, like building out these big things. By the time your business is tangibly threatened or not tangibly or like it shows up in the results, it's way too late.

[Ben]

Yeah.

[Tech Analyst]

Like, and so you got that sense from Nadella and what he's been talking about in his blog posts. There's a bit of both an articulation of the threat to Microsoft and trying to convince everyone else this is a threat to you as well.

[Ben]

Totally.

[Tech Analyst]

Which is these companies are trying to understand your business. They're trying to figure out what you actually do with your computer and they will replace you. So they can replace you. Like this gets to the question.

[Ben]

I mean, it's explicit counter messaging, relative.

[Tech Analyst]

Oh, for sure. But like, so one of the big things with the AI models, right? And there's a remains an open question in my mind, will these models generalize? Right. Can you learn one thing and then thus know how to do other things? I'm on the. I don't see evidence of generalization yet. Like there's all these, like, there's all this amazement about like these math discoveries, right? Math is like coding. It's like, it's a knowable thing. Like you're discovering things that exist. It's a bounded space to an extent in a way that maybe some other things aren't. What you need to do is you need to get data, see what works and verify did it work or did it not work? So at the end of the day might be a moot question, like maybe you don't generalize, you just gather data on everything in the world, right? We'll get neuralinks in everyone's head and gather the thought processes and then we'll actually be able to do everything. But this is the threat, which is we know for sure, if you can get enough data and you can verify, then you can do it. And that's the threat he's putting his finger on. The more you use an OpenAI or an anthropic and you use their harness, you use quad code, you use cowork, you use Codex, they are gathering everything. And even if they're saying, oh, we're not keeping your data from inquiring the model, that's actually not the important part. The important part is all the meta around it, what tools are you using, what things worked, what things didn't. And if you're relying, if you're using their user facing software, you're giving away the crown jewels of your company, which is how people actually do work, what works, what doesn't.

[Ben]

All the metadata you're training, the disruption threat.

[Tech Analyst]

That's right, that's right. And so, and so what Microsoft is proposing to do is they want to reduce models to being processors. They want to recreate the world where Microsoft Windows is the layer everyone uses. And then underneath that is an intel chip or an AMD chip or whatever it might be. And what does that chip do? That chip gets sent stuff by the operating system, it does some processing, it sends something back, and it's the operating system as the orchestrator that is figuring everything out, that's passing things to developers. On top of that, Microsoft's saying, you don't use their harnesses. We're going to build a harness. We're going to give you the tools to build your own harness. Such that these models are basically stateless. They're just things you send something to and you get something back, and they have no other context. All they get sent is exactly what they need to know to do computation. And then it comes back and they're very explicit about it. Like that's what they're seeking. They're seeking to reduce the model companies to being processors. The model companies are trying to be everything. Everything.

[Ben]

Right.

[Tech Analyst]

And so that's the direct sort of confrontation that's going on.

[Ben]

Yes. And so when you look ahead, I mean, with Microsoft, is this the right strategy? And it may not work well.

[Tech Analyst]

So this is where it's interesting to put Microsoft in contrast to Meta. Okay, because Microsoft, like, how do you survive in a world where you're not on the frontier? And what they're saying is we will be the layer between the frontier.

[Ben]

You don't have to worry about the models. It's going to be sort of model agnostic. And you, you have your own harnesses with us and you don't have to fork over proprietary data. That puts your sense.

[Tech Analyst]

I think it makes sense. It's a very reasonable strategy. The question, and I think the challenge for Microsoft is most fraught right now because in any new technology, it doesn't work good enough. Right. The AI still isn't good enough for a lot of these workflows. How do you. And so the more you integrate and the more you do all the pieces around it, the better stuff works. And this is like sort of classic Clayton Christensen sort of theory about integration versus modularization. But the idea being early on in a technology's life cycle, it's not good enough. So the more you do, the better. The integrated solution wins at the beginning. In the long run, modularization takes over where you, you, you want different suppliers you care about, like competing on, care about costs. So there was a huge wave of, oh, these models are so expensive. That felt a little astroturfed over the last few months, to be totally honest. And that is a narrative that Microsoft absolutely wants to feed because they want you thinking, oh, we can't double down,

[Ben]

we can't burn through the budget in two months on these models.

[Tech Analyst]

And so you think about it, oh, a harness is good because we can use different models. You can do sort of XYZ and. And it's kind of funny because I'm not sure how much I buy the multimodal. Sort of like if you ask a really smart model a very easy question, it's going to give you a very quick and fast answer. Right. Like, a lot of the model capability is more of a ceiling than it is sort of a floor.

[Ben]

Good point.

[Host]

Yeah.

[Tech Analyst]

There's an aspect here where it feels like Microsoft, they love the expense, use lots of models, expensive models, cheap models narrative, because that fuels the. How do I do that? Well, you need a harness, right? You need a harness that is a router that chooses the right model for the job. And again, I'm not saying there's a company as well. There is cost differences, to be clear. I'm not trying to underplay that. But it also is a narrative that feeds what Microsoft wants people to think, which is you can't build directly on these. And on the flip side, a lot of companies are gonna find this appealing. They don't want to try to figure all this out. It's like the cloud. Microsoft's a good friend. What's going on?

[Ben]

Exactly. They've been working with Microsoft for 35 years at this point.

[Tech Analyst]

There's also a question, though. How much better is it actually if you're working directly with the frontier model companies?

[Host]

Right.

[Tech Analyst]

Can you actually build stuff that you can't build otherwise? And can that stuff actually accelerate your business such that your competitor is so worried about having multiple suppliers and not getting locked into the model companies that their products suck compared to yours because you gave in. You're like, that's right. I'm all in on OpenAI. I'm all in on Anthropic, and I'm so much faster and accelerating so much more quickly. And, oh, by the way, they have this feedback loop that is actually extending their lead. The idea that Microsoft is going to be making a harness that is a lowest common denominator functionality and is comparable

[Ben]

to what Anthropic and OpenAI are offering.

[Tech Analyst]

That's right now. We'll see. I don't know. Right. At the end of the day, if you. Does the harness actually matter that much? Yeah. Right. And the. There's a thing where you can feel this with the models. So when I started doing the vibe coding stuff, it was with 5.5 on extra high. And there was like this. There was this extra high. Okay, no, no. For, for 5.5, not 5.

[Host]

Okay, okay.

[Tech Analyst]

So for 5.5 and there was this like a superpowers, like a skill package someone put together that forced it to do this planning and this review. And someone recommended to me and I was already doing planning, reviewing, but it was still, I thought, kind of useful. You get to 5.6, which overdoes this stuff on its own. You layer on this stuff. It was out of control. That's part of the reason it was out of control is doing all this crazy stuff. And basically the point is as the models got better, they sort of internalized more and more of the functionality that came from the harness.

[Host]

Right.

[Tech Analyst]

And there's like a harness called PI. They actually had a really interesting blog post this week talking about our whole approach is to do the minimum amount necessary in the harness and to make it possible for you to just add on the specific tool tools that you need. Because everyone's over engineering this such that three months down the road a new model comes out and you're actually making the whole situation worse because like your harness is like competing with your model because you did all this work up front to overcome the model's limitation. But then the model got better. And like why it's all wasted effort. Right? You should just let the model do everything. It's funny because that's a world that's actually good for Microsoft. If they're like where, well, possibly good for Microsoft. Because if it's all about the model and it doesn't matter what harness you use, then yeah, use Microsoft's harness. Use your harness. There's also a world where Microsoft's harness is so over engineered and is so careful to try to do things and not let the models get the data they need. They're actually not even coming close to tapping into what the models can do.

[Ben]

Right.

[Tech Analyst]

Like the models would be better if you gave them way more stuff. Right? Yeah. So all this stuff is kind of, this was so interesting. It's all fairly unknown. I think what they're doing is logical. There's a lot of benefits that come from being the incumbent. There's a reason why you could look at as a technologist, look at why would any company use IBM solution to build a webpage in the 1990s. And yet it worked phenomenally well.

[Ben]

And it will be interesting. Obviously it's all sort of tbd, but you look at like the number of companies that sign the open letter. For instance, you look at the weekly,

[Tech Analyst]

oh, I didn't get a chance to write about the open letter. Hilarious. Oh, that thing cracked me up.

[Ben]

And it's clearly a reaction to massive uptake for anthropic and OpenAI and the concern that that inspires everywhere else.

[Tech Analyst]

Well, the funniest thing is OpenAI signing the letter. I didn't see that. OpenAI, the most two online company in history.

[Ben]

Like, don't worry, guys. Yeah, we got it.

[Tech Analyst]

Figures by figures up in the wind. Of course they're. They're the. They don't want these open bottles competitors. But.

[Ben]

But do you know what I mean? Like, that's sort of an indication of the real.

[Tech Analyst]

For sure. Like the. You. The fact the whole industry is lining up to sign this letter is not a indicator of strength for the company signing the letter. You don't sign open letters because you're winning. Mm, exactly. You sign open letters cause you're freaking terrified.

[Ben]

Yeah.

[Tech Analyst]

Because you see these, these entities swallowing everything. They can do stuff you never expected. Right. You see, you know, you talk about Apple and Apple not seeing the memory crunch sort of coming as we've talked about, and. Or you see even the open. This response, like, has Tim Cook vibe coded an app? I'm guessing probably not. And I'm not saying you have to vibe code an app to understand AI. But there is an extent to which the. What these models can do is so far beyond what I think most people can appreciate. But someone like Saita Nadella gets it right. And so when you see this Nvidia too, Jensen Huang, you know, is sort of the guy pushing this letter. Important enough for him to get on Twitter.

[Ben]

Exactly.

[Tech Analyst]

One of the worst decisions a human can make.

[Ben]

Well, to the extent that we lack perfect visibility into how enterprises are using these tools today, I think the reaction from the incumbents across tech, the non anthropic and OpenAI incumbents, says a lot about what companies are using these things for.

[Tech Analyst]

Right. So Nvidia feels like they're in the catbird seed. Right. Everyone uses Nvidia chips. Why are they writing an open letter? Because in a world where there's only two companies that buy chips.

[Ben]

Yeah.

[Tech Analyst]

Nvidia's ordered around. Yes. They will make like. They will have the scale, they will have the capability, they have the cash flow to make their own chips and like. And it's the same thing for all the hyperscalers. In a world where there's only two, and all you're doing is competing to serve OpenAI and serve anthropic, those margins are going to get compressed a lot. You're just gonna be a financing vehicle to help them build these sorts of things out. And so like the end the whole why are VCs on this thing? Is there even a startup ecosystem if you just use the models to Every

[Ben]

startup gets replaced within a year or two if it succeeds well, shifting gears, you wrote on Monday with respect to meta that you came away from the meta call a bit alar. So scale of 1 to 10, how alarmed are you by the enterprise pitch from meta? Or is annoyance a better way to put it with meta?

[Tech Analyst]

So the meta and Microsoft comparison I think is super interesting.

[Host]

All right, and that is the end of the free preview. If you'd like to hear more from Ben and I, there are links to subscribe in the Show Notes, or you can also go to SharpTech FM. Either option will get you access to a personalized feed that has all the shows we do every week, plus lots more great content from Strikeri and the Stratchri Bundle. Check it out and if you've got feedback, please email us at emailarptech fm.