Sharp Tech with Ben Thompson - Understanding Intel’s Decline, Searching for Structural CHIPS Solutions, The Future of Section 230 Protections

Episode Date: September 5, 2024

Tracing the history that led to the decline of Intel, why Ben is conflicted about the potential solutions to the company's woes, and a Third Circuit verdict on Section 230 that could upend three decades of precedent across the tech ecosystem.

Hello and welcome back to another episode of Sharp Tech. I'm Andrew Sharp and on the other line, Ben Thompson. Ben, how you doing? Doing okay, Andrew. How are you? I'm doing all right. Can I give you some quick parenting news from this afternoon? Yes, please. So my son Charles is 17 months old and he's talking now. So he's adding a few new words every day. and earlier today we were in our living room and he saw a shopping bag and he pointed right at the logo on the bag and he said the word apple. Was it an apple bag? It was an apple bag.

My wife had been to the Apple store and thanks to my 17 month old son, I've now fully internalized the Steve Jobs branding brilliance. Yeah. It's a great logo. the fact it's a solid object, I think is very underrated. Like, you know, you look at an Apple store and just the way it sort of can stand out very, very distinctly. Obviously, everyone knows what an Apple is. My great disappointment, though, is that it was not an Amazon box. Oh, my God.

He said Amazon. That would have been great. Well, I mean, I had never heard him say the word Apple before. So for a second, I thought that was the first time he'd ever said it. My wife informed me that he had, in fact, said the word Apple in the point. past. I don't think I could have handled it if my son learned how to say Apple by looking at Apple products and not actual apples. Would have been a touchgrass moment for the entire family. But he has no idea how to say Microsoft, no idea how to say Amazon, but he can recognize Apple from across the room.

So there you go. And we're not going to be talking about Apple today, but we are going to be coming back next week with the big iPhone event. Get excited for that. But we're going to start with another great American success story. That would be Intel. Here's an email we got from Mark over the summer. He writes, what is the progress report for Pat Gelsinger at Intel? He had lofty ambitions, but is he actually achieving them? So Ben, that email came in, I think in early August, and it made me laugh because knowing you and the Intel history, I was like, man, that's a loaded question. Mark, how much time do you have. And since then, there have been four

strategy posts about Intel over the last three weeks. No, I mean, it's been interesting to watch the evolution. And I'll add links to each one in the show notes. For the sake of context, I'll list bullet points from Tuesday's article on Intel where you wrote,

Intel doesn't have the best manufacturing, Intel doesn't design the best chips, Intel is out of the game in AI. And you also noted, the future does not look bright. Is that a fair summary of the state of Intel in broad strokes? Yeah, I do feel bad about writing about it so frequently. You're right. I've been sort of thinking about it nonstop for like the last month, ever since they're sort of really poor earnings.

And to some extent, nonstop for the entire run of trajectory. Like it's right. And I think the Galsinger question is interesting. I've talked to Pat a few times. I've met him a couple times in addition to interviews. Great guy, you know, very enthusiastic, very sort of persuasive about where they are, what they're at. And the challenge with Intel is always they do, they've always talked a good game. And are they actually sort of delivering on what they need to deliver on? And I think the concerns that I have, there are concerns about Gelsinger's tenure.

I think that number one, they over in, like, COVID came along. They started selling tons of chips, both PCs and sort of server chips. They didn't see, you know, and that was when Gelsinger first started. That was the time where like, okay, we're going to embark on this massive sort of capital expenditure. We're going to really invest to sort of catch up in farther manufacturing, where they set up a separate foundry, all these sorts of things. And there should have been a recognition probably immediately that we're going to need to cut a lot of costs. we should probably cut our dividend, like things are really at risk.

And I have a lot of sympathy for the difficulty in doing that because this is a company that I wrote over a decade ago needed to move to this sort of found remodel and then over that they were in trouble. Like I wrote this in 2013. They are in long term trouble. And what happened over the following decade? Their stock went up and up and up. I don't think it quite reached their all-time highs, which is the dot-com era, but it was

approaching it. And so who am I to sit here and say they have these issues and you fast forward a decade. I was definitely right. But it's really tough when you have a real fundamental misalignment with your shareholders. And Intel was consistently being rewarded and I think was managed to optimize things for shareholders. And there's always an ongoing conversation about whether this is a problem, is Wall Street 2 short-term oriented, all these sorts of things. And I think the issue with making chips is you're talking about five to 10 to even 15-year decisions. Like you're making really long-term decisions. And most people, the responses to your quarterly earnings aren't really considering that.

Like there is, I think, a real mismatch for these particular industries in particular between sort of the long-term and the short-term. The issue in 2013 is it was clear every single time you went to a new sort of process, or particularly a new technology within that process. So you go back and like Intel led the way to FinFET and like 17 or 18 or so like that. Like every time you make this change or 5 nanometer when TSM went to EUV, it's extremely high risk and it's super expensive. And so when I started writing trajectory back then, fabs cost three to five billion dollars, you know, or maybe maybe even less.

Now they cost $20 billion. Like to make that money up, you need to have volume. And you need to sort of be. And the problem with Intel is they had missed mobile. And this is sort of the core thing. And there's lots of reasons why that might have been. Intel claims they turn down Apple, Apple or at least Tony Fidel, who I've talked to previously, says their chips were never competitive. they were too focused on sort of performance,

which has always been the case with Intel and not focused on power efficiency. And they just weren't even competitive. And so Apple initially went with Samsung and then, you know, started designing their own chips with TSM. And there was sort of a shift that happened between Samsung and TSM, which is actually an interesting point that maybe we can circle back to. Right.

My understanding is that Intel was optimized for basically a different era of computing. And they weren't able to anticipate the massive shift that was underway with mobile and now the continued shift that is happening with AI. Yeah, well, I mean, Intel was just always optimized for speed, frankly, like performance. And they ran into trouble in the early 2000s where they had a couple of things that happened, or maybe this was the late 90s where they were pushing a completely new architecture for 64-bit computing. And AMD came out with like basically an extension of 32-bit, which is a lot easier for

software developers to manage, but it wasn't sort of like, Intel was like, oh, this is our chance for a clean sheet. Let's sort of redo all this. We're carrying Cruff from the late 70s. And the problem was it's hard to make clean sheet changes in like massive ecosystems. And so AMD comes with this sort of like hack almost, which Intel was forced to adopt because Microsoft adopted it. And then it was just a, it was a much easier with for software developers. Lots of reasons why sort of the the AMD 64-bit extension became sort of the thing. So that was one. Another one was they got to laptops.

And so Intel hit this wall with their chips where they were just making them faster and faster and faster. And it got to the point where they were just using too much power and they're running way too hot. And they couldn't advance anymore. And so the shift there was to shift to sort of multi-core parts where instead of having one core that ran super fast, you would have multiple cores in a chip. So you could do multiple jobs at the same time.

But the long. and short of it is Intel has, you know, they were dragged repeatedly into, okay, we can't just solve this with brute force with sort of like, like being sort of fast. And part of the brute force mentality was we have the best manufacturing. We can make the fastest, smallest chips. Well, and for any normies in the audience, the reason focusing strictly on performance and speed didn't work in mobile is because efficiency is really important because battery life is right battery is right out of cell phones and what people care about and so if you couldn't deliver that to apple then that was a problem and they weren't going to go with intel chips right and because intel the key thing here

the reason i got stuck in this history is intel ran into that with laptops like where their chips just weren't efficient enough and they did have to rejigger and they it was sort of probably the last peak of the company where they came out with this core processor i believe that believe it was actually developed in Israel. But basically it was this new architecture that assumed multiple cores in your processor. That was the chip that won the Mac business back. I think the Macs launched on the Core 2, Core 2 duo or something like that. But the point being, they had to sort of be dragged into that, right?

And so Mobile was that on steroids. The other problem with mobile is the key thing that Apple did with mobile was you had, number one, Blackberry is sort of assuming a sort of super low power world. They were built mobile from the beginning, but they were sort of embrace the constraints to a certain extent. Microsoft was trying to cram down windows into mobile. Apple started with a clean sheet. Now, it was the Mac OS kernel, but everything was stripped down. And none of the software that ran on a Mac would sort of run on an iPhone without sort of a lot of reworking.

Yeah, some of the core stuff would function. But in then they completely rejiggered how stuff to work. Stuff couldn't stay in, you know, could get flush from memory at any time. It wasn't the head program. You had super severe memory constraints. But it was very ambitious. It was a small computer. It was still based on macOS.

So they had more ambition for performance than BlackBerry. But that meant that the whole software stack kind of started afresh. Right. And suddenly Intel didn't have this built-in advantage of, look, you've been building software on our processors for 30. years or 40 years, whatever it might be, we just have this inherent advantage in software, because you're not going to go rewrite all that stuff, right? And Apple's like, yeah, we are going to go rewrite all that stuff. And you're starting from sort of a clean sheet, and Android obviously followed in sort of

Apple's leadership in this regard. And so Intel still had a better process, but number one, their chips weren't necessarily competitive. And number two, they didn't have any sort of software moat in the way they did in PCs that let them, you know, put AMD back down and sort of make sure they captured the laptop market. And the reason this was already a problem by circa 2013, which by the way, was frankly a few years too late already because they, you know, they clearly missed. You know, their thought was, we can make a mobile processor by virtue of our manufacturing. Yes, it's not going to be as

efficient, but it's going to be so much smaller and better that we will win anyway. But, but they didn't have the software. It's only there at a software disadvantage. because all the software was being written for ARM. And, you know, and yes, there's most software is fairly abstracted, but there's still some low-level stuff that sort of the architecture matters. And so Intel didn't have a software advantage. Their manufacturing advantage was not sufficient to truly deliver a compelling sort of mobile, mobile efficiency sort of story.

And then in the meantime, they hit all these manufacturing walls and TSMs are past them. And suddenly TSM was sort of better and faster. But even before then, you could see mobile is this massive category. It's driving massive amounts of volume. There's going to be fewer PCs or PCs are going to flatline. Yes, there is the server story. And Intel made a ton of money on mobile, not by selling mobile chips, but by selling server chips because this huge buildout to because mobile is intrinsically connected with the cloud. Every app you use has sort of a cloud component that's running out of a server.

And Intel was building all those. But you could see, look, this is only going to get more expensive. You're not going to have enough volume just selling your chips. You need more volume. And design is not a differentiator for you anymore because the world is not as anchored on CISC, sort of as it once was. And so you need to open a foundry. That was my take back then. You fast forward to 2021.

when Gelsinger took over, I wrote, look, this company needs to split. And the reason they need to split is the chip business is going to get dragged down by the manufacturing. It's behind. There's losing to AMD in the data center. They're losing in all these places because their chips are just slower because they're built on sort of a process that isn't as good. And so their chip team arguably needs to go to TSM. on the other hand for the manufacturing arm to succeed it needs to not be catering to intel it needs to credibly claim to be an independent third party it needs to standardize on chip design process from cadence or whatever it might be and so galsier comes in and he's like oh wait we're building a third party chip party business we're going to set up two divisions but we're not splitting the company and i supported it and And it was, you know, reasonable arguments.

Like, AMD used to have its own thing. They split. And their split was a big mess. Global Foundries used to be AMD's manufacturing division. Global Foundries couldn't keep up. They didn't make it to EUV. They sort of had to stop. Their performance wasn't there.

AMD had to go through all these sort of lawsuits and negotiations or whatever to get out of global founders. They had a purchase agreement to sell the chips. And yes, AMD sort of caught up, but it took a decade when they got to TSMC. and by the way, global foundry is sort of founded and never sort of progressed on the line. And Galsinger's argument is like, look, like we've seen this path. Splits are messy. It'll take at least a decade.

And like the best thing we can do now is build up this division, have it separate, but Intel can be the core customer that gives the volume necessary to sort of make this stuff work. Right. But then it becomes harder to entice third party customers and build the business out that way. What I came back to last week is they've made progress. The other thing is they said we're going to do this five processes in four years sort of thing. And they've done that technically, like they've made Intel 4 and Intel 3 and Intel 28A. They haven't done any of them truly at scale.

And the trick to move down the process node tree. It's not just about making the process. It's doing it at scale. That's the magic. That's what makes TSM special is can you actually do this at scale with minimal defects, have acceptable yield, sort of make money. And we don't know if Intel can do that. And there's a lot of years of learning going down the stack that they are trying to skip. And I feel like that does I should have given that a little bit more skepticism. So they have Intel three. Their argument is we had Intel 3 working in Oregon. We pushed it to Ireland early so that we didn't have to sort of build it twice. But it blew up their margins. Like they say it's fine.

We'll get the margins up. I hope so. We didn't see it last quarter. What we saw was what sounds like a node that doesn't have acceptable yields and is losing you a bunch of money. Right. And so you fast forward to their big. Everything is focused on this 18A node. and can they deliver?

Where's they don't have an announced major sort of third party partner? Do they deserve the benefit of the doubt at this point? And I guess what the reason I wrote Intel honesty is this is kind of an admonition to myself, which is anything happened in the last three years that changed my prescription from three years ago? And the answer is no. I want Intel to succeed. I want Pat Galsinger to succeed. But they have lots of other operational issues.

Like I said, they didn't cut soon enough. They overproduced during COVID. Then they had a huge overhang. Then they didn't sort of see the Qualcomm armed ship coming. So they're ramping up production of a chip that's made by TSM that is competitive, but it's going to have terrible margins because they have to pay TSM to make it. And what I come back to is what do we want in 2030? 35. What do we want to have happening? What I think we want happening is I don't think Intel chips matter that much. It's a viable business. But the question is, do we want another foundry on the leading edge? And by the way, there's a lot like Intel is important. Intel pushes, Intel is traditionally pushed manufacturing forward and TSM has followed them. TSM has not yet truly demonstrated that they are able to lead and push.

push the window. They got to the lead and part, you know, there's a whole lots of stories and reasons why Intel fell behind. But when it comes to actually pushing not just a two nanometer or 1.4 nanometer, but to one nanometer and beyond. Intel does have 30 to 40 years of incredible breakthroughs, a lot of aggressiveness and Morris Law originated at Intel, right? It does. And having Intel pushing the envelope, they still have great R&D. They still have great sort of innovations. And beyond the having sort of an American or Western company that that is a foundry, having competition is good. Having a company with with with a 40 year record of genuine innovation is good. And so if we want an intel to succeed, it feels like we, you know, we want that

to be sort of an independent entity that earns its own way. Right. And I mean, that's my concern. It's like the point of industry subsidies is to jumpstart domestic. businesses and make them competitive in the global marketplace. And with Intel, reading your analysis over the last couple of weeks, it seems like there's a broken culture, there's problems with the current structure. And there was a report that Intel's talking to investment bankers and considering potentially splitting the company. The stock, by the way, has lost 59% of its value this year.

So clearly you're not the only one who's feeling a little bearish these days. And honestly, my instinct here, is I have a little bit of Andrew Sharpen me. I kind of want to zag when everyone is zgging. And I wrote that originally. It's like, Intel needs some time here. Let them see if they can deliver on 18A. That's what I wrote a couple weeks ago.

I enjoyed it. And the reason I sort of circled around is, it's sort of that point. It's like, look, this was my heavily considered position in 2021. If I'm really, really honest with myself, has anything happened to change that prescription? And I don't think it has. And I don't think Intel has sort of, yes, they've separated the divisions and the different P&Ls.

You know, they've done things around the edges, but is there fundamental any sort of assurance that the cultural issues and the ability to earn third party business and all that sort of stuff that was very much in question? And, you know, you go back to the 2010s. They did do a bit of outside foundry stuff like Nokia partner with them to build a chip. and Intel basically effectively bankrupted Nokia because their costs got so out of control and they couldn't get it working appropriately and all these sorts of things. And if you're a third party,

if you're a broadcom or much less an Apple or an AMD who at this, you know, and the question I have is if I ran one of these third parties, why would I go to Intel? Like what assurances do I have? Yes, in theory, I can get leading edge capacity, which at TSMC, Intel or Apple takes it all,

NVIDIA wants to take it all, AMD is always a little behind because they don't get the fastest stuff the first year it's available. They have to wait a year or two until they can sort of get capacity. But TSM has also sort of made them into what they are. They don't want to even breathe a hint that they would go anywhere else. Like TSM has the power here. They get it aside who gets what. And so no one wants to make them upset.

So why are you going to roll the dice on this entity? And it's just a real, how do we get from here or their problem? And I don't like my solutions here. Like more government money doesn't sound like a great idea. Well, I mean, that's my question is the Chips Act is sort of looming over all of this. And back in March, Intel and the Biden administration announced that Commerce plan to give Intel $8.5 billion in government grants and another $11 billion in loans to advance Intel's projects in Arizona and New Mexico, Ohio, and Oregon. and the Commerce Department also said that that package would support $100 billion of investments from Intel. But just reading about all the challenges, like, are we sure it's a good idea for the government to bet on Intel as a cornerstone of the domestic chip industry?

Just cultural challenges, leadership challenges, all of it. Like, I'm not filled with confidence for where Intel's going to be in 2035. I'm not filled with confidence about any of this. Like, it's really hard to construct. the proper incentives here, particularly if money's there for free. Intel needs money. Like, that's the core issue here. And so in that regard...

More than $20 billion, right? Yes. That looks like that's going to be the case. Particularly to get through the next couple nodes through 14A, call it, you know, $15 billion a node, maybe more. But, you know, and ideally their product business, if it came back to it, will provide maybe half that. They need someone else.

but how do you get someone else? How do you benchmark this sort of process? To ask, you know, in an ideal world, you would have, say, an Apple, do some processes at TSM, some at Intel, and you could compare once and for all who has a better process. And Intel can say, hey, we have a, ours is actually better or faster and the price is good. But the amount of complexity and cost that would impose on, say, an Apple is just not worth it. They already have access to TSM in the leading edge. Why should they, why should they even, even go there?

And, and part of the other issue is to become good, to become competitive, you need volume. Like, like, you have to, it's a real chicken and egg sort of question. You have to figure it out and get stuff working so that you can be competitive. And that's why Gelsinger's argument they needed to stay together because Intel needs to be the, the proving ground for this is a fair argument. They need to serve somebody. Right. But then you get into the problem if, oh, hey, we have to make our chips work to prove us out what's going to happen. The manufacturing is going to cater to their chips. It's going to not become suited to serving third parties. It's just going to fall back into this deeply integrated model that has been the Intel model for years and years and years. So I'd say two things. Number one, if we are serious about having American lead and managed manufacturing.

And the reality is, TSM is building fabs in like Arizona. Those fabs are not going to be worth anything if something happens in Taiwan. Like, like just the, that's just sort of the, the reality of it. Why is that?

Because all the sort of talent and management, whether it be the actual management via the internet or whether it be all, all the chip designers and developers and lead engineers that make stuff work. Like, set and forget process. You're constantly managing stuff and dealing with stuff. And yes, there's going to be people on the ground in Arizona doing that, but it's going to be deeply connected to and wired into TSM core, which is, which is in Taiwan. And so number one, you can start to make nice with China and figure this out. And then it'll be fitted. Number two,

Intel is going to need more money, I think. But the number three, if you're going to, if you're you're giving them more money, can we structure that money to get us to a better structure? And that better structure is sort of an independent fab that breaks these cultural connections that has to sort of fend for itself once and for all. I think the route here is some sort of, you know, demand subsidization where you're paying for results as opposed to money going in. They do need money going in to build the stuff to a certain extent. But it's not a really satisfying answer. There's a lot of messiness complexity.

And honestly, I wrestled with this article all weekend. I published on Tuesday. I've been wrestling with it like for 48 hours ever since. I don't feel great about it. But I do feel like there's a bit where like we have to be brutally honest about what's going on here and what's necessary. Right. Well, and when you say demand subsidy, that's the government committing to purchase guarantees

of these chips that Intel eventually produces. Is that right? Yeah. It's like, it's basically like giving Intel that third partner. Right. Like an Apple that won't buy the chips at the outside. Right. I mean, I think ideally, I don't even know how you would do this, but you would subsidize like Apple

and AMD and Broadcom to make more chips than they need and make some on Intel and some on TSM and then you have an apples to apples comparison. It's like real stuff and does it work or does it not? I don't know how you would structure that. And that's asking a lot of those companies. I mean, the problem is if you're a company like, again, why wouldn't you just go to TSMC?

It's a better process, at least for now. it's a guaranteed sort of price. You get this level of customer service and familiarity and all these sorts of things that has been honed over many decades of TSM being a standalone foundry. And so there's just this huge hole in the market. Again, the intel, the play here is,

the AI stuff does help Intel from a foundry perspective because people really want to be on the leading edge and there's a finite capacity at TSM to be on the leading edge. More demand than can be served at this point. Yeah, they just need to, Intel needs to prove they are on the leading edge and that they can be a consistent and reliable partner. Like that proof point has to come from somewhere. And, and the worry is if that's, if it's Intel, even if Intel gets its chips on there and they're good, how can you get a third party to believe I will be able to do that as well? Or is that just because Intel's taking care of itself, you know, and, you know, because that that was part of what

made the whole thing work previously. Are there any alternatives that the government has to betting on Intel and investing in Intel if they want to have a domestic manufacturer that could actually credibly produce on the leading edge? Well, so Intel is already separating internally, right, the foundry from the chip business. So that Intel's assets and capabilities in R&D is what's necessary going forward. what I would like to see is that as its own entity. That's what the U.S. needs to be concerned about from national security perspective.

That's more important than the Intel chip business. The Intel chip business is the part that makes money. That's a business that can go fab with TSM and they have, you know, they're not in great shape, but it's a functional business. They have a lot of walk in in lots of places. They have huge networks with OEMs and ODMs as far as building out servers for for companies, for the government, building PCs, like most PCs are still Intel. There's a lot of work that goes into that and they have a lot of structure.

That is one reason they do have a lot of employees because they support all these sorts of different areas and sort of use cases. And so an Intel chip business on its own is a viable business. It's also not critical to national security. Like there is competition for basically every line that they have. That's part of their problem is their chip business is diminishing in its viability or diminishing and it's profitability, but it is a viable business. The part we care about broadly is the foundry business.

And Intel has all the pieces much more than like some getting a startup or something or figuring out someone else to pick it up. You can't overnight invent all that infrastructure that is required to produce. Or get all the tacit knowledge and the ability. So what I think what we need is an independent foundry that has all the right incentives is not catering to the Intel chip division. It is really heavily incentivized to be a reliable and accessible customer service oriented third partner. And so basically my takeaways, Intel's going to need a lot of money.

If they're going to need a lot of money, let's not put bits and pieces out. Let's give them a whole bunch of money, ideally through a purchase guarantee, and shoot for 2035, that we have a standalone foundry that stands on its own two feet, that legitimately competes with TSM for leading edge business. And that's, I think, the best, like ripping the Band-Aid now, embracing the reality of what's this going to cost and what it's going to take, and sort of going from there. That was my summary. Okay.

Well, and along those lines, there was also news from Bloomberg this week. Intel is supposed to receive $8.5 billion in grants and $11 billion in loans from the 2022 Chips and Science Act, but only if the chipmaker meets key miles. and after significant due diligence, Intel, like other potential recipients, hasn't received any money yet. In ongoing negotiations, Intel has grown frustrated with what it sees as the government dragging its feet and has urged officials to release funding faster, according to people familiar with the matter. Still, the company has resisted sharing certain information requested by U.S. officials who are seeking to vet the viability of Intel's manufacturing roadmap,

said the people who asked not to be identified because the talks are private. it. I read that only to note that this strikes me as a mess on both sides at the moment. And it's an interesting snapshot of the bureaucracy that tends to hamstring government-led projects in the modern era. And just zooming out, one of the first podcasts you and I ever recorded was exactly two years ago. And it was about Intel and the Chipsack passing. And it's been two years since that bill passed. And it's still not really clear to me how much the ball has. has been advanced, but at least you're outlining a way forward that seems to make sense, but it's going to require a lot more commitment from the U.S.

government. Yeah, a lot more commitment and a lot more risk. Like, I think the, the, the, if I were to write the alternative argument to what I did, it's like, look, what, and this is what I wrote originally. Look, is a bad quarter. Let's not overreact. They're, they're overcoming problems that are decades or 10, 15 years in the making. Right.

they're trying to turn the Titanic right now. That's right. And honestly, maybe I should have written that. Like I really sort of torn. And there is a market for a second foundry in the long run. You know, the problem is there isn't one right now because they haven't shown they can,

they can deliver. And so maybe that maybe, maybe that's what I should have written. I honestly, I keep going back and forth. Like, like give them what, give them a chance.

It's, it's a mess. to split this up. The point about AMD is a fair one. It took them a decade. And by the way, we lost global founders as a leading edge manufacturer along the way. Like maybe that's not what we want is to split them up.

Like give them more time, give them more money, let them figure it out. And I think from an analyst perspective, that's actually the better take, which is that makes more sense. One other thing I do come back to is I know a lot of people in chips part of being in Taiwan

and for a long time I was the one over the last several years saying no like let's give them time this is the right path and there is and I might be biased on this point by this sort of feedback there's pretty universal sense amongst anyone that works in chips that don't trust Intel. Nope. Don't think they're going to do it.

Like when you go back to like the origin of techery and this idea that understanding culture and understanding these sorts of things is really important. And I do worry there's a certain level. No one in the industry trust Intel. Everyone has felt bullied by them at one point or the other. It's hard to articulate, but I've always felt uncomfortable by the fact that I like to be contrary to a certain extent, but there's a lot of people deep in the industry that just they disagree. They didn't disagree because they thought my analysis was wrong. They disagreed because they just didn't think Intel could do it.

They didn't think Intel could do it not because they weren't technically capable, but because they weren't culturally capable. And how do you reset and break that culture? You make it a new company. You recapitalize it. You set it on a new path. And you give it no choice but to succeed. There's no leaning on the chip division.

There's no sort of. taking care of yourself first. You just have to make it work. Well, it's interesting too because Intel is a textbook example of the dangers of short-term thinking and prioritizing shareholders and quarterly reports. And at the same time, it seems like it would be a textbook example of your maxim. It's okay for companies to die, except that in this case, it isn't okay for national security recent. So we have to figure out a way to navigate this. But in any event, I'm sure there will be

more to discuss in the years to come. Any final thoughts? Or should we keep it moving to section 230? We can keep it moving. Honestly, I've kind of been, I felt unsettled all week about this article. I guess that's a testament that I still really care. I don't know that it was the, you know, at the it or whatever. I'm just writing a blog. Who cares? But I care. I don't know if it's, I don't know if it's the right take. This is why I laughed when I got Mark's question back in early August because I knew if we discussed Pat Gelsinger and Intel, it would be like a four-hour podcast if we let the tape keep running. And so yeah, it's definitely near and dear to your heart. And the solutions are not obvious to anyone at this point. Yeah. And if it ends up that Intel

gets more money and say, hey, you know, give it a shot. I'm okay. I'm okay with it. It's a bit of a 51.49 sort of thing. But yeah, it's. We'll see. It hasn't seen a dime yet. So that's step number one.

But in any event, to keep it moving, we mentioned this news in passing last week, but there was a potentially massive ruling in the third circuit concerning section 230 liability. The case was Anderson v. TikTok, and it centers on a 10-year-old girl who accidentally hanged herself trying to attempt the conduct depicted in a TikTok video. The TikTok video was called the Blackout Challenge, which encouraged viewers to record themselves engaging in acts of self-asphyxiation. That is obviously a horrifying tragedy, number one, and the Third Circuit ruled in her family's favor last week, allowing a lawsuit against TikTok to proceed to trial. But as far as the legal implications, the way this has evolved is actually pretty interesting.

So I'm just going to provide a bit of background for anybody who hasn't been following. The Anderson lawsuit was dismissed at the district court level in New Jersey in 2022 because that court held that TikTok was shielded from liability by Section 230, which, says that for purposes of tort liability, quote, no provider of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider and that, quote, no cause of action may be brought and no liability may be imposed under state or local law that is inconsistent with this section. So that's where precedent has lived. Meanwhile, just just to sort of drill

down on this. Section 230 is downstream from, I think it was a case called, there's a copy surf case and a prodigy case. There's two cases that undergirded this. But one case was that the person was, it was held that they weren't liable or the service wasn't liable. The next one was that they were liable. And the distinction between the two is that the second service engaged in moderation.

And so they were, you know, this was like the forum era of the internet in like the 90s. And this is, right. was like, what's the, what's the movie in Wolf of Wall Street? It is. Straton Oakmont, I believe. Yeah, they were, they were actually like someone like like, you know, posted some stuff that was, and they sued because it was like for whatever might be and held prodigy liable. I believe it was prodigy or was that copy serve? No.

So copiouserve was found to be not liable because they were just hosting content and weren't moderating whatsoever. And then Prodigy, the selling point for Prodigy was that they would moderate. and try to protect people from porn and different things that could disturb children and whatnot. And because they were doing that, they were found to be liable downstream of the CompuServe precedent where they decided that if you're just content neutral, then you're not held liable as a publisher. And then in the wake of Prodigy being held liable because they were moderating and because they were trying to clean up the internet, Congress freaked out and passed Section 230.

man it's great having a lawyer on the podcast this is great you have to explain this stuff um the great irony of section 230 is it was in the communications act of i think 1996 the vast majority of which was struck down except for 230 um so yeah it's basically what two 30 said was just by virtue of engaging in moderation does not make you liable for the content that that's sort of on there and it's it is what what is interesting is all if you go back to the congressional record and you read through all of it, which I have at some point. I wrote about this, you know, I've written about 230 a few different times. It was definitely all about protecting kids and like, you know, an adult content and, you know, all the sorts of things that you can imagine motivating Congress in this sort of regard. In 1996, particularly. Right. What's, what is notable is pretty quickly and then consistently what 230 means has been pretty dramatic. radically expanded beyond just limiting adult content.

It's basically come to mean that a platform is not responsible for any user-generated content, period. And so the big question is, you know, what was the intent of Congress here? Anyhow, let you continue with the details because we will come back to that point. Yeah. So meanwhile, what happened over the past few years after the district court dismissed this Anderson v. TikTok law, lawsuit was that both Texas and Florida passed laws that attempted to restrict the ability of social media platforms to use algorithms to ban or shadow ban content based on political viewpoints. And those laws were then challenged on a constitutional basis in a case that was

captioned Net Choice v. Moody and was decided by the Supreme Court this summer. So at the Supreme Court level, in a 6-3 verdict, the Supreme Court found that a platform's algorithm, them reflects editorial judgments about compiling the third party speech it wants in the way it wants and is the platform's own expressive product and is therefore protected by the First Amendment. So this panel of judges in the Third Circuit hearing Anderson v. TikTok then wrote that if the Supreme Court has held that platforms engaged in moderation are exercising their First Amendment rights and engaged in a form of speech, by algorithm particularly, then they should be treated as speakers for purposes of liability.

And they are not subject to the protections of Section 230, which protects interactive computer service providers, but not speakers. So that's where we are right now. This was the Third Circuit interpreting the Supreme Court's ruling in Net Choice v. Moody. And it's pretty interesting. I mean, does that explanation make sense to you? Feel free to add anything I left out. What do you think of where things stand at the moment? Well, it's definitely the case that this decision is in total rejection of basically all precedent in this area, which again, I mentioned that it's sort of expanded over time to basically say that platforms are not liable for these generic content. And that's a pretty sort of expansive protection. And that is.

is what makes back of the day forums viable. So it makes social media viable. Like the, you know, whether it be, you know, Facebook or TikTok or sort of whatever, whatever it might be. And the difference, the distinction from. And so this gets into like, well, what was the congressional intent? How much does precedent matter? All these sorts of pieces.

Because you go back to the forum era when, when 230 came around, there, the, there was no algorithm. The algorithm, there was an algorithm. There's always an algorithm. The algorithm was time. Like stuff was put in order in a forum, whether the list of threads or within the threads, the messages, it was like, you know, whatever it was posted in order of when it was posted. So that time is an algorithm, to be clear, there is always an algorithm.

But the idea was, oh, wait, so there's no. So no, the original section 230 did not imagine a word. where stuff was sorted by the service provider, it came about in a world where there was no algorithm, unless you count time as an algorithm, which technically is, but it's not how we're thinking about algorithms sort of in this case. It didn't account for platforms deliberately amplifying content

based on who's engaging with what. That's right. And so there is, if you, so kind of the complexity in this case, is there, if you look at it very logically, and sort of legalistically, yeah, there's a real problem here.

If the editorial decisions, in writing those algorithms, which aren't a human going out and picking stuff out, it's, again, based on things like engagement or all those sorts of pieces, if that is protected by the First Amendment, like, that's always been, like, part of the reason why 230 is sort of important is 230 basically saves on a lot of litigation costs. right? Yeah. If you tried to sue Facebook for some content, at the end of the day, the vast majority of those lawsuits would be dismissed on First Amendment grounds.

Right? Like, like, you know, and so what 230 does, it kind of shortcuts that process. Instead of litigating every single message that you want to sue about, well, it's like, hey, look, they're not responsible. You can sue the user, but you can't sue Facebook. And because, because that's sort of the. I mean, that literally happened with this TikTok lawsuit. It was dismissed.

Then you also have stare decisis. How much does it matter that courts have consistently interpreted 230 in this way for, believe it or not, almost 30 years? Like that's how I like it already, it's 28 years. All pretty old. Right. It's been a while. Literally the entire internet is built on this assumption. So where's the line between the, this is sort of an old school constitutional law argument behind sort of an originalist sort of argument versus.

Staridicis, and like, what is the actual expectation that businesses were, massive businesses were built on sort of that this is how it operates. You know, it's a real sort of tension in here. No, the original Section 230 did not, but the algorithms didn't exist, right? So, I don't know. It's, my sense is, is that sort of precedent should, rule the day here. Yes, the sort of congressional record and the intent of Congress did not talk about algorithms, but that's a function of algorithms not existing at the time. And at the end of the day,

you do need sort of the rule of law and the basis of law. And part of that is you go back historically common law. Like the U.S. legal system is built on not a written constitution, but English common law going back, you know, hundreds and hundreds of years on top of which was laid the Constitution. And that's just sort of, and so my sense is this is a mistaken decision because precedent does matter. But from a strict legalistic textual reading, you can see how the court could get to get to this point.

What's your take? You're the one that's qualified on this matter. Well, for anybody who's not a lawyer, Stari Decisysus is a concept in law that basically boils down to this is the law because this is the way it's always been. And we're just going to keep that precedent rolling. And I think it's interesting. I mean, I really liked Judge Paul Mady and his concurrence in the Third Circuit because he traced the history of 230 and the ways it's been broadened and sort of distorted over the years. and he said, today, Section 230 rides in to rescue corporations from virtually any claim loosely related to content posted by a third party, no matter the cause of action and whatever the provider's actions.

The result is a Section 230 that immunizes platforms from the consequences of their own conduct and permits platforms to ignore the ordinary obligation that most businesses have to take reasonable steps to prevent their services from causing devastating harm. I thought he nailed where we are. And then from a legal standpoint, Clarence Thomas actually put his finger on the issue with his dissent in Net Choice, where he wrote, in the platforms world, they are fully responsible for their websites when it results in constitutional protections. But the moment that responsibility could lead to liability, they can disclaim any obligations and enjoy greater protections from suit than nearly any industry. So I don't think that this sort of inconsistency is going to be viable long term. And I think that a lot of the largest social media platforms just enjoy too much protection. But what I would like to see happen is a law passed by Congress, perhaps, if Congress could be bothered to address this issue for the first time in 30 years, where some of the biggest information platforms, on the internet are subject to a heightened duty of care, but you don't have to subject every, you know, blog or website on the internet to the same duty of care. Because I do think that the

Section 230 protections have been pretty elemental to a lot of the growth that we've seen on the internet. But I don't think it's viable to continue on this path where platforms like TikTok and Facebook or even Reddit are just not liable for anything that's posted on there, even when they know that it's causing harm and choose not to take steps to deal with it. Well, but that's the challenge, right? Like, did anyone at TikTok know about this blackout challenge going around? Should they have? What steps could they have taken?

It's kind of inverting section 230 where it used to be that section 230 was a good Samaritan provision. In that by virtue of you trying to do a good thing, that won't make you viable. What you're arguing for is actually, no, it's not just so if you don't do sufficient work, you will be liable. If you're not, it's not, we don't want good Samaritans. We want an active police force. It is a big deal. It's because this arguably makes the platform's not viable.

Or at a minimum, what they do need to do is build like Facebook has. thousands and thousands and thousands of moderators that are that are really in super aggressively that are going to err on the side of censoring more, not less, because they can't risk anything that is potentially, you know, that leads to some sort of some sort of liability. Now, you could zoom out and say, look, I think social media was kind of a mistake. Forums were fine. Let's put the toothpaste back in the tube. Yes, that's exactly my take.

which is fine, but you're, like, there is a bit here where it's a pretty severe judicial decision that absolutely does undo 30 years of, of, of sort of decisions that dramatically reduces the market value and the viability of these businesses. Now, again, maybe you think it's worthwhile, but that, that's a pretty rad. There's a reason we have the concept of stare decisis. Like Zardisicis is part of the concept of the rule of law, which is you understand what the rules are and you can operate according to them. And again, you can construct a logical legal argument why there is this conflict and it needs to be resolved. It's going to be resolved in this direction.

But it's a pretty severe, like it's a pretty fundamental shift. Now, again, you might say screw Facebook, screw TikTok, screw everyone. it's still worth it. We should go back to a world of nothing but blogs and forums and sort of direct responsibility. And again, maybe you're like the internet was a mistake. That's where we should be. But it's pretty radical. Well, that's why I wanted to mention it.

And it's going to be appealed by TikTok and could very well force the Supreme Court to rule on the scope of Section 230. So for the time being, this is just one interpretation of the. net choice decision. And so the third circuit by itself didn't just nuke section 230. Other circuits, I'm sure will weigh in. I mean, it's interesting because, you know, it's not really what the net choice case was about. It's sort of like there's like, it's sort of hung on this one provision. To your point, it is closer to the, the Clarence Thomas dissent. That more clearly weighs out this point. It was also dissent. It wasn't sort of the, the sort of majority decision. But I do think, think this does need to go to the Supreme Court without question. It is a conflict. There is a,

there is a real tension between stating platforms at First Amendment rights that that applies to their moderation and this saying they don't have any liability for, for their moderation. Like, the way it's supposed to work is that who's liable for posting this? It's the person that actually posted it. And their defense for posting it is they have a First Amendment defense, which has carve outs. Like there are carve outs around libel. There are carve outs around, you know, the clear and present danger.

Like, like, you know, Fire and a crowd of theater, yes. No, fire in a theater is not a real thing. Don't fall into that myth. It has to be a realistic, immediate sort of risk factor. I can't remember all the mythos around fire to a crowded theater,

but I think that was actually an example of what's not covered. But it's always misinterpreted. We have to find someone a link in here. But I promise you, it's a... Email at Sharp Tech.

But they're very narrow. It's very hard to find someone liable for speech. Like that's part, that's just sort of how the First Amendment is sort of interpreted, you know, particularly today.

And, but the idea is you go to the individual. You don't go to, you don't go to the platform. Right. But there's attention here. Yeah, they want to claim the First Amendment for the algorithmic decisions. But if you're claiming the First Amendment, aren't you then implicitly stating that you are making an editorial judgment for which you could, at least in theory, be held liable?

Well, and just to articulate my vision for the future, if all this nets out to a place where courts begin imposing liability in like extreme circumstances where a massive social media platform is knowingly amplifying dangerous content, that will then result in just much more boring versions of those massive platforms because they'll have to aggressively moderate to mitigate the liability risk. So just to be clear, I don't think anyone's saying that TikTok knowingly promoted this video. Right. I don't know. No, I may have lots of videos about TikTok. I don't think they're knowingly promoting asphyxiation videos. Okay. Negligently, let's say. They should have known. Yes, exactly.

Yes, okay. But I think my vision for those platforms would be that they would be more sanitized. And instead of being a bunch of cable channels, they become more like traditional broadcast channels where no one can curse in sitcoms and the content is all very vanilla. and people just gradually shift to more niche content sources, like with our group chat addiction, I think that might be healthier for all of society in the long run. And that's why I don't fear it as like a doomsday scenario. Right.

It's really bad if you're a Facebook or a meta shareholder. Right. It's not, you know, I'm sympathetic to that point. The world of forums and blogs was pretty great. That was sort of peak internet. You know, And this world where everyone can interact with each other at all hours of the day, it hasn't been going so well. Everyone's just pissed off all the time.

So one of the ironies, one of the ironies of the whole Section 230 debate is it's really been latched onto, I think, by Republicans and conservatives as, but it's latched onto in the context of they feel the platforms are biased against them, right? And they're promoting left wing content and suppressing right wing content and sort of whatever it might be. the reality is that if Section 230 went away because this decision is carried, and that Section 230 still exists, but it de facto doesn't exist for social networks that are algorithmically driven, it's going to, the censorship's going to get a gazillion times worse because they can't run the risk of something going up that is now. Now, again, it may be sort of a double bank shot in that the platforms get way more censorious, but they also diminish an importance because they're just so boring in

vanilla. But it's not clear that second part is going to happen, right? They will still be the easiest to use. But there's not, there hasn't been a real, uh, clear think through of the implications of section 230 ending, right? You, you like sort of Elon Musk's Twitter, uh, you should be pro, a maximal interpretation of section 230. Uh, and so that thought occurred to me.

Twitter would definitely be on the chopping block if they're suddenly liable. Yeah. So no 230, you get way more censorship, not less. And it's, it's always been so bizarre to me

that it's been latched on to by conservatives for this point. Now again, if you just want to kill social media, by all means, Section 230 is... And to be clear, this is why,

in addition to the Supreme Court weighing in, I think this is why I would want Congress to weigh in as well, because I don't want 230 protections to go away for the entire internet because if you're subjecting medium-sized platforms

or small platforms to the same standard of liability, then it doesn't become viable to stay in business for a lot of different entities. But if there's a way to tailor it to the largest platforms, that at least is a coherent vision

for the path. Yeah, well, so 230 would still exist for, say, a substack or sort of basically you'd have a choice, either use an algorithm. If you use, yeah, if you use or I mean, substack might get in trouble because they now have their Twitter sort of thing,

whatever might be. They might not be able to or they might have to take real care in recommending substacks, which is like sort of, you know, where you click on one, it's like, oh, you might also be interested in sort of X, Y, Z. That's an algorithm. And so you would then be downstream liable for that. Like the issue is it's really easy. It's really easy to sort of imagine a better world.

And getting to there, there's a lot of messiness and complexity in the meantime. You might kill a whole lot more than you realize. This all applies to the Intel discussion, by the way. I was going to say it's a nice theme for today's podcast. Yeah. Getting from here to there is very messy. But yes, a world in which this court decision holds will look very, very different. A lot of upheaval.

Yes. Well, we shall see. Frankly, it's going to look. There's going to be a lot more censorship out of self-preservation. And yeah, just be careful what you wish for. Well, one promise for the audience. The world is going to look more similar next week. We didn't record a mailbag Monday this week because of Labor Day. But we are coming back with the mailbag next week. Back to the regular schedule. Can't wait to get into it. We've gotten a lot of great emails. Email at Sharp tech.fm. Ben, until then, enjoy your weekend, and I will talk to you on Monday. I'm not going to enjoy my weekend. I'm still sort of an agony. Racked with this Intel situation. Yes. Good luck working through that over the next 72 hours, but we'll have mail to look forward

to at the end of that emotional journey this weekend, and I can't wait. Talk to you later.