Latest / Deep Values: Value Investing Deep Dives / Why Intel Loses Billions While Revenue Surges
Transcript
- 0:00Imagine you're walking down the street and you see this
- 0:03restaurant and it is just absolutely packed.
- 0:06I mean, there is a line around the block.
- 0:08People are fighting for tables. The kitchen is churning out
- 0:10meals faster than ever. And revenue is literally the
- 0:14highest it's been in 15 years. Sounds like a great problem to
- 0:17have. You would think so, but then you
- 0:19peek into the back office and the owners just sitting there
- 0:22staring at a spreadsheet, realizing they are somehow
- 0:24losing thousands of dollars every single night.
- 0:28Oh wow, yeah, that is that's a bad situation.
- 0:31It's insane. And that is the exact paradox we
- 0:34are unpacking. Today, we're doing a deep dive
- 0:36into a massive 39 metric deep value financial analysis report
- 0:42on Intel Corporation. Right, updated through their Q2
- 0:452026 earnings. Exactly alongside their fiscal
- 0:48year 202510K, our mission today is to figure out how a company
- 0:52that used to essentially, you know, print money has become one
- 0:56of the most capital intensive turn around projects in modern
- 0:59industrial history. It really is a historic pivot.
- 1:01Yeah. But before we jump into the
- 1:04numbers to access more advanced researches, applications and our
- 1:07latest personal deep value agents, welcome to explore and
- 1:10join Deep Values APP to track total market temperature.
- 1:14Welcome to Search and download the iOS App Buffett Indicator
- 1:17from the Apple App Store or Bubble Watch from Google Play.
- 1:20And as a quick reminder, all analysis today is based on
- 1:23publicly available sources and may be subject to revisions or
- 1:26different interpretations. All contents from this channel
- 1:29are for research and entertainment purposes only and
- 1:31do not constitute any investment or financial advice or service.
- 1:35Please conduct your own due diligence before making
- 1:38investment or financial decisions.
- 1:40OK, let's unpack this because starting with this Q2 2026
- 1:44report, the July 2026 data, it's really the only way to grasp the
- 1:49absolute whiplash happening. Here, whiplash is the perfect
- 1:51word for it, honestly. Right, because the top top line
- 1:54numbers are just staggering. Yeah.
- 1:55Total net revenue hit $16.1 billion for the quarter.
- 1:58Which is huge. Yeah, it's a 25% year over year
- 2:01increase. Yeah.
- 2:01That is their strongest top line growth in over 15 years.
- 2:04I mean, that is an incredible top line acceleration for a for
- 2:08a legacy heavy industrial tech company.
- 2:11You know, turning a ship of that size takes immense momentum.
- 2:15Exactly. And the internal segments are on
- 2:17fire, like the data center and AI group DCAI that surged 59%
- 2:22year over year to 6.3 billion. Right, because their new Zeon 6
- 2:25processors are actually capturing those massive
- 2:28hyperscaler cloud budgets now. Yeah.
- 2:30And on top of that, the client computing group, so you know the
- 2:33PC and laptop side that expanded 13% to 8.9 billion.
- 2:37So demand is surging across the board.
- 2:39It really is, but. And here's the big question.
- 2:42If demand is that high, why is our metaphorical restaurant
- 2:45owner crying in the back office? Right?
- 2:48Because despite all that revenue, Intel posted a massive
- 2:51$11 billion gap net loss for the quarter.
- 2:54I mean, how do you grow revenue by 25% and still lose $11
- 2:57billion in three months? Well, what's fascinating here is
- 3:00the massive gulf between accounting reality and ongoing
- 3:04operational reality. OK, what does that mean?
- 3:06So that $11 billion gap loss generally accepted accounting
- 3:11principles, it's heavily distorted by management
- 3:15essentially clearing the decks. Clearing the decks.
- 3:17Yeah. The report shows this number is
- 3:19just packed with restructuring actions and these massive asset
- 3:23impairments. Meaning they're just what,
- 3:25throwing away old equipment? In a financial sense, yeah,
- 3:27pretty much an impairment basically means you're admitting
- 3:30that an asset on your balance sheet is no longer worth what
- 3:33you were originally paid for it. OK.
- 3:35So they are looking at older factories, equipment built for
- 3:38legacy 14 nanometer or 10 nanometer nodes, and
- 3:41acknowledging that hey, those machines just can't manufacture
- 3:44the chips of the future. So it was a one time hit.
- 3:46Exactly. They take a massive one time
- 3:49financial hit to write down the value of those assets.
- 3:52But when you strip away those transformative one time charges
- 3:55and you just look at the non GAAP basis, the actual ongoing
- 3:58business operations, they actually posted earnings per
- 4:01share of $0.42. Oh wow.
- 4:04Yeah, which soundly beat Wall Street estimates.
- 4:07So the core engine of making and selling current chips is
- 4:10actually profitable. Very much so.
- 4:12Margins are demonstrating these really critical signs of
- 4:15recovery. Jab gross margins actually
- 4:17expanded by almost 13 percentage points year over year.
- 4:20Wait, really? Yeah, up to 40.4%.
- 4:23I mean a 13 point jump in a single year for physical
- 4:27manufacturing, that seems unusually high.
- 4:29Like what caused that sudden snapback?
- 4:32It's the easing of factory underutilization penalties.
- 4:35Underutilization penalties. Right.
- 4:36So in semiconductor manufacturing, the cost
- 4:39structure is just brutal. Your factories cost billions of
- 4:42dollars to build and run. Sure.
- 4:44You've got the chemicals, the power, the clean room
- 4:46maintenance, the depreciation, and you pay for all of that,
- 4:50whether you were printing one single microchip or a million of
- 4:52them. I see.
- 4:54So back in 2024 and 2025, Intel didn't have the volume.
- 4:58Their factories were sitting partially idle.
- 4:59And when a factory runs at like 50% capacity, the fixed cost per
- 5:04chip just skyrockets. Exactly.
- 5:06It destroys your gross margin. But now that volume is picking
- 5:09up, that immense fixed cost friction is subsiding.
- 5:12The factories are running closer to full capacity, which spreads
- 5:16those costs over millions of chips.
- 5:18That makes total sense. And that dynamic is really the
- 5:20perfect bridge to the core of this whole deep dive for you
- 5:23guys listening. To understand why a tech giant
- 5:25burns billions of dollars during a growth phase, we have to look
- 5:29under the hood at their specific business model.
- 5:32Right, what they call the IDM 2 point O strategy, yeah.
- 5:35So explain IDM 2 point O for us. Well, to understand it, we have
- 5:39to contrast Intel with the rest of the industry.
- 5:42So most of the household names we talked about, you know,
- 5:45NVIDIA, AMD, Apple, Qualcomm, they are what the industry calls
- 5:50fabulous. Meaning they just designed the
- 5:52blueprints right, but they outsource the insanely expensive
- 5:56physical fabrication to pureplay foundries, mostly TSMC in
- 6:00Taiwan. Precisely.
- 6:01They don't own the factories. Intel, however, is an integrated
- 6:04device manufacturer. In IBM.
- 6:07They designed the logic architecture and they physically
- 6:09fabricate the silicon in their own proprietary foundry, right?
- 6:13And under CEO Lipbutan, they've logically split the company into
- 6:17two entities. You have Intel products, which
- 6:19designs the chips, and Intel Foundry, which physically makes
- 6:23them. And IDM 2 point O means Intel
- 6:25Foundry is now actually competing with TSMC to
- 6:28manufacture chips for other companies, not just Intel.
- 6:31That's the plan, yeah. Which sounds great on paper, but
- 6:34reading through their fiscal year 202510K, the financial
- 6:38reality of running those foundries is just terrifying.
- 6:41It really is staggering. Because their operating cash
- 6:43flow actually recovered to a healthy $9.7 billion.
- 6:47Which is a lot of cash coming in the door from selling products.
- 6:50Right. But here's where it gets really
- 6:51interesting. They generated nearly 10 billion
- 6:54in operating cash, but their capital expenditures, CapEx
- 6:58consumed a jaw-dropping $14.6 billion.
- 7:03They spend nearly $15 billion just on equipment and
- 7:06facilities. That leaves them with a negative
- 7:08free cash flow of roughly $4.9 billion, right?
- 7:12I mean, how are they spending that much money on equipment in
- 7:14a single year? Well, a massive portion of that
- 7:17is going toward extreme ultraviolet or EUV lithography
- 7:20machines. Right, EUV machines.
- 7:22Yeah, and to understand the cost, you really have to
- 7:24understand the mechanics. EUV machines are arguably the
- 7:27most complex machines ever mass produced by humanity.
- 7:30Wow. They use highly engineered light
- 7:33with a wavelength of just 13.5 nanometers to carve micro
- 7:37circuits into silicon wafers. Right, because the transistors
- 7:40are getting so incredibly tiny that normal light waves are
- 7:43literally too fat to draw the lines, Yes.
- 7:46Exactly, and a single EUV machine from ASML costs hundreds
- 7:50of millions of dollars. Hundreds of millions just for
- 7:53one. Machine just one, and you don't
- 7:55just need one, you need entire fleets of them to run a modern
- 7:59high volume fab. You know, I picture it like a
- 8:02multibillion dollar treadmill that just speeds up every single
- 8:05year. That's a great way to look at
- 8:06it. Like you buy a $200 million EUV
- 8:08machine just to span the game today, but the technology moves
- 8:12so fast that in three years you need the next generation
- 8:16machine. And if you don't run that
- 8:17machine at 100% capacity day and night, the depreciation cost
- 8:21just eat you alive. The operating leverage is
- 8:24extreme. I mean, to put a number on your
- 8:26treadmill analogy, the report shows Intel's depreciation alone
- 8:30consistently exceeds $10 billion annually. 10 billion?
- 8:34Yeah, that is 10 billion dollars in value vanishing from your
- 8:39equipment every year through wear, tear and technological
- 8:43obsolescence. That's hard to even wrap your
- 8:45head around. And then you add to that $16
- 8:47billion annually in R&D just to design the chips and you start
- 8:51to really see the scale here. So you have over $25 billion
- 8:55evaporating before you've even sold a single chip just to
- 8:59maintain the capability to participate in the industry.
- 9:01Exactly. So how do they bridge a massive
- 9:03liquidity gap when free cash flow is -5 billion dollars?
- 9:07Well, management has had to execute some pretty extreme
- 9:10financial maneuvers to generate cash.
- 9:12Like what? For instance, they sold off a
- 9:1451% stake in Altera to Silver Lake Partners.
- 9:17Cool. Why now?
- 9:18Altera was a highly valuable programmable chip business they
- 9:22acquired years ago. Selling a majority stick
- 9:25generated a pre tax gain of $5.6 billion.
- 9:29Which provides immediate liquidity.
- 9:31Right. But it sacrifices the future
- 9:33cash flows and strategic optionality that business would
- 9:36have provided. It's like selling your car to
- 9:38pay your mortgage. I mean they did something
- 9:40similar with Mobileye, right? Doing a secondary offering of
- 9:43shares to raise almost another billion dollars and reading the
- 9:47shareholder return section, this really stuck out to me.
- 9:50They completely halted their stock buyback program.
- 9:53Completely. In 2020, Intel returned nearly
- 9:56$20 billion to shareholders, mostly through buybacks.
- 10:00But since Q 4/20/22, they've spent $0.00 on buybacks.
- 10:04Well, they cannot afford to enrich shareholders when every
- 10:07spare dollar is required just to fund the factories.
- 10:10Which brings us to the most critical debate in the report.
- 10:13Since Intel is betting the entire farm on these physical
- 10:16factories, we have to figure out if this infrastructure is an
- 10:21impenetrable economic Moat or a ticking time bomb.
- 10:24The report mentions by 86 architecture switching costs as
- 10:28a massive mode, but I have to ask, if I'm a Chief technology
- 10:31officer at a big enterprise and another company offers me a
- 10:34cheaper server, why is it so hard to just switch?
- 10:37It comes down to the software layer.
- 10:39Decades of enterprise software, legacy applications, server
- 10:42operating systems. They've all been written
- 10:45specifically to compile and execute on the BI 86 instruction
- 10:49set. I see.
- 10:51So to switch to a different architecture, an IT department
- 10:54would have to rewrite, recompile, or emulate millions
- 10:57of lines of code. That sounds like a nightmare.
- 10:59It is the friction, the risk of breaking critical systems, and
- 11:03the labor cost of doing that usually vastly outweigh whatever
- 11:07you're saving on the hardware. It creates an incredibly sticky
- 11:10customer base. OK, so that's the software mode.
- 11:12The report also leans heavily on the geopolitical mode.
- 11:15Yes, very heavily. Because Intel 18A is the most
- 11:19advanced semiconductor process technology developed and
- 11:22manufactured entirely within the United States, right?
- 11:24And with rising global tensions, Western governments and defense
- 11:28contractors desperately want a silicon supply chain that
- 11:31doesn't rely entirely on East Asia.
- 11:33Exactly. Replicating Intel's domestic
- 11:36footprint would require hundreds of billions of dollars and
- 11:39decades of specialized metallurgical and chemical
- 11:41engineering knowledge. So that physical footprint is a
- 11:44formidable barrier to entry. But this raises an important
- 11:48question. What happens if the Moat is
- 11:50bypassed entirely? That's the real threat.
- 11:53We have to look at the long term structural threats, right.
- 11:55And the most pressing 1 is the relentless rise of ARM based
- 11:59processors. Yeah, ARM is a different
- 12:02architecture. It's a reduced instruction set
- 12:04computer or RISC companies like Apple, Qualcomm and hyperscalers
- 12:09building their own custom silicon like AWS, Graviton.
- 12:13They all use ARM designs. And the main advantage of ARM is
- 12:17power efficiency, right? Like an ARM chip uses
- 12:20significantly less electricity and generates way less heat than
- 12:23a traditional by 86. Chip much less.
- 12:25Which, if you were running a data center the size of a
- 12:27football field, saving 20% of your power and cooling bill is
- 12:30worth millions. 10s of millions easily.
- 12:32But wait, what about that software stickiness we just
- 12:35talked about? If cloud providers eventually
- 12:37completely abstract the hardware layer, meaning the software
- 12:41developer just writes an app and the cloud translates it
- 12:44seamlessly in the background regardless of whether it's
- 12:47running on by 86 or ARM, doesn't Intel's deepest Moat just
- 12:52evacuate overnight? If we connect this to the bigger
- 12:54picture, yeah, yes it does. Wow.
- 12:57That is the permanent existential threat to the BI 86
- 13:00franchise. If the instruction set is no
- 13:02longer a hard requirement for enterprise software, Intel loses
- 13:06its captive audience. And then they're just competing
- 13:08on price. They are forced to compete
- 13:10purely on manufacturing cost and power efficiency against ARM
- 13:13based competitors, which is a brutally difficult fight given
- 13:17ARM's inherent architectural advantages and power
- 13:19consumption. Man, then there is the
- 13:22geopolitical angle. We talked about the US
- 13:24manufacturing footprint being a Moat, but the report points out
- 13:27a massive global oversupply risk.
- 13:30How does building more domestic factories create a risk?
- 13:33Well, we have to look at the global supply dynamics
- 13:35impartially here. Intel is a primary beneficiary
- 13:38of the USCHIPS Act, right? They're receiving significant
- 13:43federal funding to build domestic foundries, right?
- 13:46However, they are not operating in a vacuum.
- 13:49Governments in Europe, Japan and China are concurrently
- 13:52subsidizing their own domestic semiconductor manufacturing at
- 13:56massive scales to secure their own tech sovereignty.
- 13:59So multiple major governments are essentially printing money
- 14:02to build giant ship factories in their own backyards at the exact
- 14:05same time. Exactly, and from a purely
- 14:08economic standpoint, this state sponsored capital deployment
- 14:12could permanently alter global supply and demand.
- 14:15Because there's just too much capacity being built.
- 14:17Right. If everyone builds highly
- 14:19subsidized factories, you risk creating a chronic global
- 14:22oversupply, particularly in mature semiconductor nodes.
- 14:26And in any commodity market, oversupply destroys pricing
- 14:29power across the entire industry.
- 14:31That is a massive macroeconomic risk and even if they navigate
- 14:34that, the report details brutal short term self-inflicted
- 14:38bottlenecks. Yeah, the execution issues.
- 14:40Right, It specifically mentions shortages in advanced packaging
- 14:44like their Foveros technology and the friction of tuning
- 14:47yields on their new AT and A node.
- 14:51What does that actually mean on the factory floor?
- 14:53Well, let's start with Foveros. Historically, a ship was a flat,
- 14:582 dimensional piece of silicon. Advanced packaging like Foveros
- 15:02is like building a skyscraper instead of a sprawling ranch
- 15:05house. OK, I like that analogy.
- 15:07It allows Intel to stack different shiplets, logic,
- 15:11memory, power vertically in three dimensions.
- 15:14It drastically improves performance, but the
- 15:16manufacturing complexity is exponentially higher.
- 15:19So there's more room for error. Exactly.
- 15:21If you have a bottleneck in that specific packaging step, you
- 15:25can't ship the final product no matter how fast you manufacture
- 15:28the base wafers. Got it.
- 15:29And what about tuning yields? OK, so a silicon wafer is about
- 15:32the size of a large pizza and it contains hundreds of individual
- 15:36chips. Sure.
- 15:36The yield is the percentage of those chips on the pizza that
- 15:39actually work perfectly without defects.
- 15:41When you introduce a brand new process node like 18 A, the
- 15:45initial yields are usually terrible.
- 15:46Maybe only 30% of the chips work.
- 15:49Meaning you were throwing 70% of that incredibly expensive
- 15:52silicon into the trash. Literally.
- 15:54So. Tuning yields is the agonizing
- 15:57microscopic process of tweaking the chemistry and lithography to
- 16:01get that yield up to 80 or 90% every single day.
- 16:05A factory isn't operating at high yield efficiency, it is
- 16:07burning cash. So given all of this, the
- 16:10negative cash flow, the ARM thread, the yield tuning, how is
- 16:13the market, specifically the deep value investors reacting to
- 16:17this turn around attempt? The reaction from institutional
- 16:20shareholders is a mix of high right?
- 16:22They don't want the C-Suite getting their full bonuses while
- 16:25the foundational factories burn billions in cash.
- 16:29Which is a very traditional value investor mindset.
- 16:33But if you run Intel through classic deep value investing
- 16:36frameworks, think Benjamin Graham, David Dodd or Peter
- 16:40Lynn, it currently fails almost every single test of a great
- 16:44business. Yeah.
- 16:45If we look at this through a traditional value lens like
- 16:48Warren Buffett, looking for companies that generate massive,
- 16:51predictable cash with low capital requirements, Intel
- 16:55looks terrifying right now, doesn't it?
- 16:56Yeah, let's look at valuation discipline, the price to
- 16:59earnings ratio. The PE is fundamentally broken
- 17:01here. It is mathematically meaningless
- 17:03right now, right, because they've generated net losses in
- 17:06recent trailing periods. You literally cannot calculate a
- 17:09multiple on negative earnings. And cash generation fails
- 17:12spectacularly because of the violently negative free cash
- 17:15flow we already discussed. Exactly.
- 17:16So if you're listening to this and looking at your portfolio,
- 17:19you might see that the only standard value metric Intel
- 17:22actually passes is the price to book ratio, the PB.
- 17:25The PB sits at a deeply compressed 1.11 X total market
- 17:30cap is around $115 billion and their shareholders equity, the
- 17:35literal accounting value of their factories and assets is
- 17:38about $103 billion, right? You might think, hey, a stock
- 17:42trading at just 11% over its liquidation value is on
- 17:45clearance. But that is a classic value
- 17:48trap. A value trap.
- 17:49Yeah, a low price to book ratio doesn't just automatically mean
- 17:52it's a screaming bargain. In this case, it reflects
- 17:55extreme market pessimism regarding the terminal return
- 17:58profile of those physical assets.
- 18:00Because a factory is only valuable if it can produce goods
- 18:02profitably. Exactly.
- 18:04If it can't, it's just a very expensive, rapidly depreciating
- 18:07building that constantly requires more capital.
- 18:09And we see that exact fear reflected in the actions of
- 18:12famous investors outlined in the report, like Seth Klarman's Bow
- 18:15Post Group, legendary for value investing.
- 18:19They bought into Intel back in 2021 thinking the assets were
- 18:22cheap, but as the true multi year magnitude of this foundry
- 18:26build out became apparent, he aggressively trimmed the
- 18:29position. And Michael Burry took a broader
- 18:31angle. Right.
- 18:32Essentially shorting the entire AI and semiconductor exuberance
- 18:35via ETFs, betting the capital cycle would eventually crush
- 18:39margins and. Then you have Monish Pabrai, who
- 18:41completely avoids lodging foundries altogether,
- 18:44specifically because of the brutal capital requirements to
- 18:47stay on that EUV treadmill we talked about.
- 18:50So when you look at the mechanics of the discounted cash
- 18:52flow analysis, the DC, you understand exactly why these
- 18:56value managers are staying away. You really do.
- 18:58So what does this all mean for the actual math breakdown?
- 19:01How the DCF models this out? Well, a discount cash flow model
- 19:05attempts to find the intrinsic value of a company by estimating
- 19:08all the future free cash it will ever produce and discounting
- 19:12that back to today's dollars, right?
- 19:14But remember, free cash flow is operating cash minus capital
- 19:18expenditures. Which is negative right now.
- 19:19Exactly because Intel's current free cash flow is negative $5
- 19:23billion, a standard model breaks.
- 19:26You have to normalize it. OK, so how do they do that?
- 19:28The report assumes that by 2027, the peak factory build out
- 19:32finishes revenue stabilize and Intel returns to historical
- 19:36margins. Under those assumptions, they
- 19:38establish A baseline normalized free cash flow of $5.0 billion
- 19:44annually. OK.
- 19:45Assuming they stabilize and start making 5 billion in free
- 19:48cash a year, what does the model do with that number?
- 19:51If we run a nominal case scenario, meaning a moderately
- 19:54successful turn around where Intel Foundry start securing
- 19:57volume from external customers and PC cycles remain steady, we
- 20:01assign a 5% nominal growth rate to that five point O billion
- 20:05dollar cash flow into the future.
- 20:06OK. Discounting that back to the
- 20:08present, the model calculates an intrinsic equity value of around
- 20:1211.37 per share. Wait, hold on a valuation of
- 20:1511.37 that assumes A measly 5% growth on normalized cash flows.
- 20:20Given they just posted 25% top line revenue growth in Q2, isn't
- 20:25that DCF model being incredibly pessimistic?
- 20:27Like why should we trust evaluation that low?
- 20:29It sounds pessimistic, but you have to separate short term
- 20:32revenue bounces from long term cash generation.
- 20:34That 25% revenue growth is a cyclical bounce off a horrific
- 20:38bottom in 2024. Long term, competing in the
- 20:41foundry business against TSMC is a low margin, incredibly high
- 20:46CapEx slugfest. So the costs just keep eating
- 20:49the revenue. The model assumes that even if
- 20:51revenue grows, the relentless capital expenditures required to
- 20:55buy next generation lithography tool will consume the vast
- 20:58majority of the cash generated. And the stock is currently
- 21:01trading around 22.80. Yep.
- 21:03So at current levels, the market is completely rejecting the 5%
- 21:06growth model. At 22.80, the equity market is
- 21:09pricing in a growth and margin recovery scenario that vastly
- 21:13exceeds a 5% baseline. They're pricing in a miracle.
- 21:16Basically, investors buying today are underwriting a
- 21:19massive, flawless return to the absolute glory days of 2018 to
- 21:232020, where Intel was consistently generating 12 to
- 21:26$15 billion in annual free cash flow.
- 21:28That implies there's practically 0 margin of safety, like if this
- 21:32incredibly complex turn around stumbles, IF18A yields don't
- 21:35tune fast enough, or if the ARM software abstraction
- 21:38accelerates. The downside risk to the stock
- 21:40is profound. That is the ultimate take away
- 21:43from a deep value perspective here.
- 21:45Intel is not a traditional value stock right now, no matter what
- 21:48the price to book ratio says. It is a highly cyclical, capital
- 21:54heavy special situation. You are betting on the flawless
- 21:58execution of an incredibly complex manufacturing and
- 22:02geopolitical road map, not just buying a cheap stream of cash
- 22:05flows. It is a master class for all of
- 22:08you listening in. Why you have to look past the
- 22:11top line revenue when the capital expenditure required
- 22:14just to survive is so astronomical.
- 22:17Traditional value metrics completely breakdown.
- 22:20Absolutely. Which brings us back to our
- 22:21restaurant owner staring at the spreadsheet.
- 22:24They've realized that to keep the line out the door, they
- 22:26can't just buy ingredients. They have to buy a new multi
- 22:30$1,000,000 state-of-the-art oven every single year just to keep
- 22:34up with the restaurant across the street.
- 22:35It's a tough business to be in. It really is OK, And that leaves
- 22:38us with a final thought to Mull over.
- 22:40We talked about the AI hardware arms race, right?
- 22:43If this relentless March of innovation eventually leads to
- 22:47compute power becoming as cheap and commoditized as electricity,
- 22:50where nobody cares what chip architecture is running in the
- 22:53background, they just just expect the data to slow
- 22:56instantly, Does owning the most advanced physical factories in
- 22:59the West become Intel's ultimate unassailable monopoly, or is it
- 23:04just the heaviest piece of infrastructure that finally
- 23:06sinks the ship? That is the multibillion dollar
- 23:09question that will define the next decade of silicon.
- 23:12Thanks for diving deep with us today.
- 23:14Keep questioning, keep learning, and we'll catch you on the next
- 23:16deep dive.