Latest / Connecting the Dots / Disruptive Acquisitions, AI's Public Market, and Global Tech Geopolitics

Transcript

Disruptive Acquisitions, AI's Public Market, and Global Tech Geopolitics · Connecting the Dots

• 0:00
  Today's tech news moves fast, but understanding
• 0:03
  it shouldn't be hard. That's why Snarfl Solutions
• 0:06
  Group brings you Connecting the Docs, where we
• 0:09
  break down what's happening in tech and why it
• 0:11
  matters to you. Now, let's dive in with your
• 0:14
  hosts. Progress has literally always meant building
• 0:32
  smaller and smaller chips to get faster results.
• 0:35
  Well, absolutely Moore's law right shrink it
• 0:37
  down shrink it down and in business We're taught
• 0:40
  that technology just seamlessly crosses borders
• 0:43
  to conquer the globe But and this is what we're
• 0:46
  getting into today What happens when you wake
• 0:48
  up one morning and realize all those laws of
• 0:51
  physics have just suddenly reversed man It creates
• 0:53
  a serious sense of vertigo. Honestly, I mean
• 0:56
  the strategies that built the biggest companies
• 0:57
  in the world over the last 20 years, they suddenly
• 1:00
  look completely obsolete. Yeah, completely. The
• 1:03
  map we've been using just no longer matches the
• 1:05
  territory we're actually standing on. And, well,
• 1:08
  helping you navigate that strange new territory
• 1:10
  is exactly what we do here. So welcome to Connecting
• 1:13
  the Dots, the daily radio show where we break
• 1:15
  down the most consequential tech news and figure
• 1:18
  out its real, everyday impact on your life, your
• 1:21
  work, and your wallet. I'm your host, Alex. And
• 1:24
  I'm Morgan. We are broadcasting today from the
• 1:26
  Snarful Solutions Group in Sacramento, California,
• 1:30
  and today is Monday, May 4, 2026. Beautiful morning
• 1:34
  for it too. It really is. Before we get into
• 1:36
  these massive shifts happening in retail and
• 1:38
  global computing, let's quickly set the stage
• 1:40
  for the day. Looking out the window here in Sacramento,
• 1:44
  it's a pretty comfortable morning. Very nice.
• 1:46
  Yeah, sitting at 64 degrees Fahrenheit with a
• 1:48
  light 7 mile per hour breeze. Though, I mean,
• 1:51
  we might see some light rain later. Nationally
• 1:53
  though, it's a bit more turbulent. Oh yeah, definitely.
• 1:56
  There are nearly 300 active U .S. weather alerts
• 1:58
  right now. About 89 of those are over land, so
• 2:02
  definitely keep an eye on weather .gov if you're
• 2:04
  traveling today. Well, you know that turbulence
• 2:06
  is actually a really great metaphor for the financial
• 2:08
  markets right now, too. Oh, for sure! How are
• 2:11
  we looking? So if you look at the major indices,
• 2:14
  you're seeing just a lot of mixed signals. The
• 2:16
  Dow closed down on Friday at $49 ,499 .27. Right.
• 2:22
  But, and here's the catch, if you step back and
• 2:24
  look at the whole week, it secured solid gains.
• 2:27
  And the S &P 500 did the exact same thing. Slight
• 2:30
  daily dip, but a weekly gain. It's that classic
• 2:33
  push and pull. Meanwhile, the crypto space is
• 2:35
  just like leaning entirely into the pull. Oh,
• 2:38
  big time. Yeah, Bitcoin opened this morning over
• 2:40
  the $78 ,000 mark. Wow. I know, right? And it's
• 2:44
  up over 4 % for the week. There is real momentum
• 2:47
  building in digital assets. And, you know, that
• 2:50
  mix of traditional market tension and digital
• 2:52
  momentum brings us perfectly to the mission for
• 2:54
  today's show. Let's hear it. We are going to
• 2:56
  look closer at three massive tech stories from
• 2:59
  the Morningstar Tech Report. These are actively
• 3:02
  shifting the balance of power in retail and artificial
• 3:04
  intelligence right now. Yeah, these aren't just,
• 3:06
  you know, boring boardroom dramas. Exactly. These
• 3:09
  are moves that will dictate how our families
• 3:11
  shop, how professionals work, and honestly, what
• 3:14
  the global tech ecosystem looks like by the end
• 3:16
  of the decade. Absolutely. But first, let's pay
• 3:18
  the bills with a quick break. That's good. Today's
• 3:21
  episode is brought to you by Strong Coffee, the
• 3:23
  premium instant coffee that tastes like a traditional
• 3:25
  brew but is packed with 15 grams of protein,
• 3:28
  NCTs, and adaptogens like ashwagandha. Get clean,
• 3:32
  sustained energy without the jitters. Use promo
• 3:34
  code SNARFULL for 20 % off at strongcoffeecompany
• 3:39
  .com. Alright, so let's unpack this first story
• 3:41
  because it has absolutely sent shockwaves through
• 3:44
  Wall Street this morning. I mean, shockwaves
• 3:46
  is an understatement. Truly. GameStop, yes. The
• 3:49
  mall -based video game retailer has made an unsolicited
• 3:53
  $55 .5 billion takeover bid for the e -commerce
• 3:56
  pioneer eBay. Okay, we really need to pause and
• 3:59
  just absorb the sheer absurdity of those numbers.
• 4:02
  This is wild. They are offering $125 per share,
• 4:05
  right, in a 50 -50 split of cash and stock. Yep,
• 4:08
  that's the offer. My initial reaction to this
• 4:10
  was just complete disbelief. I mean, we're talking
• 4:13
  about a minnow trying to swallow a whale here.
• 4:15
  Oh, completely. Going into Friday... GameStop
• 4:18
  had a market valuation of roughly $12 billion.
• 4:21
  Right. And eBay is sitting at about $46 billion.
• 4:25
  So I have to ask, is this just leftover, you
• 4:28
  know, 2021 era meme stock hubris, or is there
• 4:32
  an actual grounded business logic here? Well,
• 4:35
  it's incredibly tempting to dismiss it as a meme
• 4:38
  stock stunt. I mean, everyone's first thought
• 4:39
  was here we go again. Right, right. But when
• 4:42
  you look at the actual term sheet that GameStop
• 4:44
  CEO Ryan Cohen is presenting, the math is shockingly
• 4:48
  sober. Really? Yeah. Cohen hasn't just issued
• 4:51
  some flashy press release, he's quietly billed
• 4:54
  a 5 % stake in eBay already. Wow, okay. And he
• 4:57
  is threatening to take this hostile if the eBay
• 5:00
  board rebuffed him. Plus, he's backed by a $20
• 5:02
  billion bank loan from TD Securities. Wait, hold
• 5:05
  on. A $20 billion loan from TD Securities, that's
• 5:09
  real money. Getting a loan like that from a major
• 5:11
  traditional bank means the adults in the room
• 5:13
  have looked at this and signed off. So what is
• 5:16
  Cohen's pitch to the banks? I mean, how does
• 5:18
  a $12 billion company justify buying a company
• 5:22
  four times its size? So the core of his argument
• 5:26
  rests on this ruthless $2 billion cost cutting
• 5:31
  plan. Okay. He essentially believes eBay has
• 5:34
  become super bloated and just lost its identity
• 5:36
  over the years. Specifically, he is pointing
• 5:39
  a massive spotlight at eBay's marketing and sales
• 5:42
  budget. Which is how much? In 2025, eBay spent
• 5:45
  $2 .4 billion in that department alone. $2 .4
• 5:48
  billion on just marketing. What did they get
• 5:51
  for that? I mean, what was the return on investment
• 5:53
  there? Well, according to the letter Cohen sent
• 5:55
  to eBay's chair, That 2 .4 billion dollars spend
• 5:58
  resulted in a net increase of active buyers of
• 6:01
  just point seven five percent You've got to be
• 6:03
  kidding me. Nope in raw numbers. They spent 2
• 6:06
  .4 billion dollars to acquire barely 1 million
• 6:08
  new active users Okay, wait, let me let me do
• 6:10
  that math in my head really quick That means
• 6:12
  they paid roughly twenty four hundred dollars
• 6:14
  to acquire a single new user basically. Yeah
• 6:16
  On a platform where people are mostly buying
• 6:19
  like secondhand clothes and used electronics,
• 6:22
  that is a staggering customer acquisition cost.
• 6:25
  It's completely unsustainable. So his argument
• 6:27
  is basically you are a legacy Internet brand
• 6:30
  with near universal recognition. Why on earth
• 6:33
  are you spending billions on digital ads that
• 6:35
  clearly aren't working? That is exactly the crux
• 6:38
  of the financial argument. So he wants to slash
• 6:41
  1 .2 billion directly from sales and marketing
• 6:44
  right off the bat. Makes sense. Then cut another
• 6:47
  300 million from product development and shave
• 6:50
  500 million from admin departments like HR and
• 6:53
  IT. But, I mean, you can't just cut your way
• 6:56
  to growth, right? Right. And Cohen knows that.
• 6:58
  The cuts are only half the strategy. The other
• 7:00
  half, which is the actual growth engine, is what
• 7:03
  GameStop brings to the physical world. And this
• 7:05
  is where the story shifts from, you know, a Wall
• 7:07
  Street spreadsheet to our actual neighborhoods.
• 7:10
  Exactly. Because GameStop still has about 1600
• 7:12
  physical retail steamers left right after spending
• 7:15
  the last few years closing down all the underperforming
• 7:17
  ones. Yeah, about 1600 locations. So under this
• 7:20
  proposal, those 1600 stores wouldn't just be
• 7:23
  selling video games anymore. They suddenly become
• 7:26
  this massive physical nationwide logistical network
• 7:29
  for eBay. Precisely. Think about the biggest
• 7:32
  friction point of using eBay for the average
• 7:35
  person right now. Shipping or getting scammed.
• 7:38
  Both. It's the logistics and the trust. Cohen
• 7:42
  envisions these physical stores as national drop
• 7:44
  -off and shipping nodes. But more importantly,
• 7:47
  he's focusing on authentication. Oh, that's huge.
• 7:50
  Right now, if you buy, say, an expensive trading
• 7:52
  card or a vintage watch or high -end hardware
• 7:54
  on eBay, there's always that lingering fear,
• 7:57
  right? Yeah, that it's counterfeit. Exactly.
• 7:59
  But GameStop's retail staff already inspect,
• 8:03
  grade, and authenticate used hardware and collectibles.
• 8:05
  every single day. That's their core business.
• 8:08
  Okay, so imagine the practical application here
• 8:10
  for a family or a small business. You have a
• 8:12
  rare collectible. Instead of taking photos, writing
• 8:15
  a long description, hoping buyers trust your
• 8:16
  zero review profile, and then dealing with buying
• 8:19
  shipping materials. Which is a huge hassle. You
• 8:22
  just walk into your local strip mall, you hand
• 8:25
  the item to a human being behind the GameStop
• 8:27
  counter, they verify it on the spot, and boom,
• 8:31
  your eBay listing is instantly generated and
• 8:33
  it carries a guaranteed trust badge. It's seamless.
• 8:36
  Yeah. And they're even proposing using the back
• 8:39
  rooms of these physical stores as live broadcasting
• 8:42
  studios for high -end auctions. That is so smart.
• 8:45
  It really is. It is a direct play to build a
• 8:47
  legitimate structural competitor to Amazon. Which
• 8:50
  is incredible. incredibly hard to do. Right,
• 8:52
  because Amazon dominates due to its incredibly
• 8:55
  dense multi -billion dollar logistical network.
• 8:59
  Building that from scratch today is nearly impossible.
• 9:01
  Yeah, nobody has the capital for that. But this
• 9:03
  combined entity... It would merge massive digital
• 9:07
  reach with an immediate pre -existing physical
• 9:10
  footprint in almost every community in America.
• 9:12
  That would fundamentally change how everyday
• 9:14
  families interact with online resale. Though,
• 9:17
  I mean, eBay's leadership isn't just going to
• 9:19
  surrender without a fight. Oh, definitely not.
• 9:21
  I saw they just reported first quarter revenue
• 9:23
  of 3 .9 billion dollars, which actually beat
• 9:26
  Wall Street's expectations. They did beat expectations,
• 9:29
  yeah. And eBay's current strategy is essentially
• 9:31
  what Cohen is trying to interrupt. they've been
• 9:34
  trying to buy relevance, basically. How so? Well,
• 9:37
  acquiring companies like this secondhand fashion
• 9:39
  app, Depop, for $1 .2 billion just to capture
• 9:43
  younger demographics. Right, right. And crucially,
• 9:46
  eBay actually credits their recent Q1 success
• 9:50
  to leaning heavily into artificial intelligence.
• 9:53
  Specifically, they're using AI to drastically
• 9:56
  improve communication and search matching between
• 9:59
  buyers and sellers. It's fascinating that eBay
• 10:02
  is relying on AI to save its margins. But, you
• 10:05
  know, that software reliance exposes a massive
• 10:08
  physical bottleneck for the entire industry.
• 10:10
  Huge bottleneck. Because the AI that eBay and
• 10:12
  everyone else wants to use requires immense computing
• 10:15
  power. And right now, we are hitting the absolute
• 10:17
  physical limits of how small we can shrink traditional
• 10:20
  silicon chips to power those systems. Yeah, physics
• 10:22
  is undefeated. Exactly. Which is why this company
• 10:25
  called Soebras is trying to solve the problem
• 10:27
  by going in the exact opposite direction. Let's
• 10:30
  talk about the startup baking a giant pizza to
• 10:32
  take on a three trillion dollar king man It is
• 10:35
  one of the most audacious hardware swings. We've
• 10:38
  seen in decades Cerebras is an AI chip maker,
• 10:42
  and they're launching their second attempt at
• 10:44
  a U .S. initial public offering. Okay. They are
• 10:47
  seeking evaluation of up to $40 billion, aiming
• 10:51
  to raise roughly $3 .5 billion by selling 28
• 10:55
  million shares priced between $115 and $125 a
• 10:58
  piece. Okay, I really want to explore the physical
• 11:01
  hardware here because it feels counterintuitive
• 11:03
  to literally everything we know about computing.
• 11:05
  It really does. For decades, like we said, Moore's
• 11:08
  law dictated that progress meant making chips
• 11:11
  smaller and packing more transistors into a tiny
• 11:13
  space. Most people think of a computer chip as
• 11:16
  a tiny little square of silicon, like a single
• 11:18
  slice of pizza, right? Sure. Nvidia's flagship
• 11:20
  B200 chip, which powers most of this AI revolution,
• 11:24
  is essentially just a really, really powerful
• 11:26
  slice. But Cerebras isn't making a slice. They
• 11:29
  are baking an entire massive table -sized pizza.
• 11:32
  That's massive. Their flagship processor, the
• 11:34
  WSE3, is the largest commercial silicon chip
• 11:37
  in the world. It is 58 times larger than Nvidia's
• 11:41
  B200. 58. I have to ask, though, if bigger is
• 11:43
  better. Why didn't Nvidia or Intel do this years
• 11:47
  ago? Well, the reason nobody did it is because
• 11:49
  of the manufacturing defect rate, which the industry
• 11:52
  commonly calls the yield problem. Okay, the yield
• 11:55
  problem. Right. When you manufacture silicon
• 11:58
  wafers, there are always microscopic flaws. Just
• 12:01
  part of the process. Traditionally, you make
• 12:03
  a large wafer, and then you slice it up into
• 12:06
  hundreds of tiny chips. Oh, I see. That way,
• 12:09
  if there's a microscopic flaw on one part of
• 12:11
  the wafer, you only throw away the one tiny slice
• 12:14
  that contains the flaw, and you keep all the
• 12:16
  rest. Right, you save the good slice. Exactly.
• 12:19
  But if you try to make one giant chip out of
• 12:21
  the whole wafer, a single microscopic flaw could
• 12:24
  ruin the entire multi -million dollar piece of
• 12:27
  hardware. So how did Cerebras fix that? They
• 12:29
  figured out a proprietary way to route around
• 12:32
  those microscopic defects on the fly, allowing
• 12:35
  them to keep the entire wafer intact. OK, so
• 12:37
  they solved the manufacturing hurdle. But why
• 12:39
  does the giant pizza approach actually make the
• 12:42
  AI faster, like physically? It all comes down
• 12:44
  to latency. You know, the time it takes for data
• 12:47
  to travel. Think of those standard chips, the
• 12:49
  single slices of pizza. If you need to feed a
• 12:52
  massive AI model. One slice isn't enough right
• 12:55
  you need a lot of you have to wire thousands
• 12:57
  of those slices together So the data has to constantly
• 13:00
  leave one chip travel down a copper wire and
• 13:03
  enter another chip which takes time Passing that
• 13:06
  data back and forth creates a massive traffic
• 13:08
  jam But because cerebras bakes an entire table
• 13:11
  -sized pizza on one continuous piece of silicon
• 13:13
  The data never has to leave the chip the communication
• 13:16
  bottleneck just disappears so the AI gets its
• 13:19
  answers almost instantly exactly but um Let me
• 13:22
  push back here for a second. NVIDIA is a $3 trillion
• 13:26
  absolute juggernaut. They essentially built the
• 13:29
  modern AI era from the ground up. Is having a
• 13:32
  physically larger chip really enough to dethrone
• 13:35
  them? See, that is the $40 billion question right
• 13:38
  there. And answering it requires us to understand
• 13:40
  a crucial distinction in how artificial intelligence
• 13:43
  works. The difference between training and inference.
• 13:47
  Okay, break that down for me. Cerebras is not
• 13:50
  actually trying to dethrone Nvidia entirely.
• 13:53
  Nvidia completely dominates the training phase.
• 13:56
  That is when an AI model is essentially going
• 13:58
  to school. It's reading the entire internet,
• 14:01
  learning the patterns and building its neural
• 14:03
  network. Got it. That process takes months and
• 14:06
  requires massive clustered arrays of traditional
• 14:09
  GPUs. So if training is the AI going to school
• 14:12
  and like reading the encyclopedia. What is inference?
• 14:16
  Inference is when the AI actually takes the test.
• 14:19
  Oh, I like that. It is what happens when you
• 14:21
  sit at your desk, you type a prompt into chat
• 14:23
  GPT, and you wait for the answer. That is the
• 14:25
  AI applying what it already learned. to generate
• 14:28
  a response for you in real time. OK, so inference
• 14:31
  is the everyday usage. Exactly. And Cerebras
• 14:33
  is carving out a stronghold, specifically in
• 14:36
  the inference market. Their massive chips are
• 14:38
  uniquely designed to provide incredibly fast
• 14:40
  memory bandwidth, which is exactly what inference
• 14:43
  requires. And clearly, the Major players believe
• 14:46
  in this approach, right? Because Cerebras just
• 14:48
  secured a transformative deal with OpenAF. Oh,
• 14:50
  it's huge. A multi -year compute agreement worth
• 14:52
  more than $10 billion. Wow. They are providing
• 14:56
  up to 750 megawatts of computing power through
• 15:00
  2028. Just to put that in perspective, 750 megawatts
• 15:04
  is enough electricity to power a medium -sized
• 15:06
  city. That is unbelievable. For a company that
• 15:08
  generated $510 million in revenue in 2025, Which,
• 15:13
  by the way, was a 76 % jump from the previous
• 15:16
  year, swinging them into a 238 million dollar
• 15:19
  profit. A huge leap. Landing a $10 billion contract
• 15:23
  is totally game -changing. It serves as an anchor
• 15:26
  that proves to Wall Street they are a serious
• 15:28
  infrastructural player. It's an incredible turnaround,
• 15:31
  especially considering they had a major geopolitical
• 15:33
  roadblock recently. I mean, their first IPO attempt
• 15:36
  back in 2024 was actually halted due to national
• 15:39
  security concerns, right? That's right, yeah.
• 15:41
  The Committee on Foreign Investment in the United
• 15:43
  States, or CFIUS, stepped in. Why did they intervene?
• 15:46
  Well, the issue was that Cerebras' largest customer
• 15:48
  and a major shareholder was G42. They're a massive
• 15:52
  tech conglomerate based in Abu Dhabi. And at
• 15:55
  the time, G42 had deep ties to Chinese partners.
• 15:59
  At one point, G42 accounted for 87 % of Cerebras'
• 16:04
  revenue. The U .S. government was very worried
• 16:06
  about advanced U .S. chip technology bleeding
• 16:08
  into China through the Middle East. So what did
• 16:11
  Cerebras do? They had to pause the IPO. Wait
• 16:14
  for G42 to divest its Chinese investments and
• 16:18
  convert their holdings to non -voting shares.
• 16:20
  They finally cleared that hurdle in March 2025.
• 16:24
  It's wild to think that a tech startup in California
• 16:26
  had its IPO pause because the Middle East has
• 16:29
  become a proxy battleground for the tech war
• 16:31
  between the U .S. and China. It really shows
• 16:33
  how connected everything is. But the geopolitical
• 16:35
  tension isn't just affecting startups. It is
• 16:37
  causing massive headaches for the biggest players
• 16:40
  in the world. Which actually brings us to our
• 16:42
  third story. The tech iron curtain. Yeah, and
• 16:45
  this is arguably the most profound shift in global
• 16:47
  commerce we've seen this decade. Nvidia CEO Jensen
• 16:50
  Huang recently announced that his company's market
• 16:53
  share of AI accelerators in China has plummeted
• 16:55
  to zero percent. Zero. Zero percent. Just two
• 16:59
  years ago, analysts estimated Nvidia controlled
• 17:02
  roughly 66 percent of that massive market. Now
• 17:05
  it's completely evaporated. OK, going from two
• 17:08
  thirds of the market to absolute zero in a country
• 17:11
  the size of China is staggering. It's unprecedented.
• 17:14
  And Hong isn't blaming a competitor, is he? He
• 17:17
  is placing the blame squarely on U .S. export
• 17:20
  policy. Correct. He's saying the government restrictions
• 17:23
  meant to keep advanced chips out of China have
• 17:26
  largely backfired. It feels like the ultimate
• 17:29
  textbook definition of the Streisand effect.
• 17:32
  Oh, absolutely. By trying to choke off China's
• 17:34
  access to our cutting -edge AI hardware, we didn't
• 17:37
  stop them from wanting it or needing it, we just
• 17:40
  gave them the ultimate financial and strategic
• 17:42
  incentive to build their own alternative faster
• 17:45
  than they ever would have otherwise. Yeah, it
• 17:47
  essentially acted as a massive protective tariff
• 17:50
  for Chinese domestic chipmakers. Wow, yeah. Huang
• 17:53
  argues that by conceding a market of that scale,
• 17:55
  The U .S. basically forced companies like Huawei,
• 17:58
  Cambricon, and more threads to accelerate their
• 18:01
  own self -sufficiency. Have they actually caught
• 18:03
  up? Well, tech research firm Bernstein estimates
• 18:06
  these domestic vendors are now moving to cover
• 18:08
  up to 80 % of China's total demand. Huan stated
• 18:12
  bluntly that conceding an entire market the size
• 18:14
  of China probably does not make a lot of strategic
• 18:17
  sense and that U .S. policy needs to be dynamic.
• 18:20
  I understand the hardware side, for sure, but
• 18:23
  I have to ask about the software. Because I thought
• 18:25
  Nvidia's real secret weapon wasn't just the physical
• 18:28
  silicon, it was their software ecosystem. Specifically
• 18:31
  CUDA. Doesn't America still maintain a massive
• 18:34
  advantage there? Oh, the CUEMOTE is absolutely
• 18:37
  real. Okay, good. CUDA is Nvidia's proprietary
• 18:40
  software platform that developers use to communicate
• 18:43
  with the hardware. To use an analogy, imagine
• 18:47
  if every master chef in the world was trained
• 18:49
  to only use one specific brand of highly complex
• 18:52
  oven. Okay. Even if a rival company builds a
• 18:54
  slightly better oven, no chef wants to use it
• 18:57
  because they'd have to completely relearn how
• 18:59
  to cook. That makes perfect sense. That's what
• 19:01
  Nvidia did with software developers. And impartial
• 19:04
  assessments confirm that this software stack
• 19:07
  remains the main frontier that American companies
• 19:10
  still hold firmly. It is incredibly difficult
• 19:13
  to replicate the years of development and the
• 19:15
  global developer loyalty built around CEDA. OK,
• 19:18
  so if we still have the chefs, so to speak, how
• 19:22
  is China circumventing that advantage? Well,
• 19:25
  this is where Huang breaks AI down into what
• 19:27
  he calls the five layer cake. You have the physical
• 19:30
  hardware at the bottom, then system software
• 19:33
  like CUD, then frameworks, then the AI models
• 19:35
  themselves, and finally the applications we use.
• 19:38
  China might be lagging in the system software
• 19:41
  layer, but they have a massive advantage in human
• 19:43
  capital. Huang referred to their vast pool of
• 19:46
  math and science researchers as one of their
• 19:48
  national treasures. They just have the sheer
• 19:50
  numbers. Exactly. Couple that raw talent with
• 19:53
  significantly cheaper energy costs. And remember,
• 19:57
  training these models requires city levels of
• 19:59
  electricity and China remains a highly formidable
• 20:02
  competitor. They are essentially using cheap
• 20:05
  energy and massive brain power to brute force
• 20:08
  their way around the hardware and software gaps.
• 20:11
  So what does this actually mean for you, the
• 20:12
  listener commuting to work right now? For the
• 20:15
  last 20 years, technology was largely a global...
• 20:18
  monoculture. Right, everyone used the same stuff.
• 20:20
  We all used USB standards, similar internet protocols,
• 20:23
  and the same underlying silicon chips. But looking
• 20:26
  forward, we are looking at a fractured ecosystem.
• 20:29
  The smartphones, the enterprise software, and
• 20:31
  the AI platforms your family uses might look
• 20:34
  function and cost completely differently depending
• 20:36
  on which side of the globe they were developed
• 20:38
  on. Exactly. And a fractured ecosystem often
• 20:41
  means slower overall global deployment and higher
• 20:43
  costs for consumers everywhere. Instead of building
• 20:45
  on a shared foundation, companies and nations
• 20:48
  are duplicating efforts. It is a massive disruption
• 20:51
  on every single level. And while speaking of
• 20:54
  keeping things running smoothly, before we wrap
• 20:56
  up today's show, let's take one more quick break.
• 20:59
  Sounds good. Before we wrap up, a quick thanks
• 21:01
  to Old Glory, an iconic music and sports fan
• 21:04
  merch store, perfect for the techie in your life.
• 21:07
  Use promo code SNARFL for 15 % off at oldglory
• 21:11
  .com. So, looking back at everything we've unpacked
• 21:14
  today, the underlying current is a reversal of
• 21:17
  what we thought was permanent. The ambitions
• 21:19
  of meme stock veterans are attempting to fundamentally
• 21:21
  reshape physical and digital retail by weaponizing
• 21:25
  brick and mortar stores. Right. A mall video
• 21:27
  game store attempting to swallow an e -commerce
• 21:30
  pioneer just to solve the physical trust bottleneck
• 21:33
  of the internet. It's wild. Then you have a startup
• 21:35
  solving impossible manufacturing physics to bake
• 21:37
  a giant silicon chip to challenge a $3 trillion
• 21:40
  incumbent. completely rethinking how we process
• 21:43
  everyday AI inference. And finally, the geopolitical
• 21:47
  splintering of the hardware market. The U .S.
• 21:50
  export bans inadvertently turbocharging Chinese
• 21:53
  domestic innovation, completely reshaping who
• 21:56
  builds the future and where it gets deployed.
• 21:59
  The map has entirely changed. It really makes
• 22:02
  you evaluate your own assumptions about how the
• 22:04
  market works. Which leaves us with a question
• 22:06
  for you, the listener, to mull over as you go
• 22:09
  about your day. We started by talking about the
• 22:12
  rules of the game being broken. Right. If a retail
• 22:14
  video game store can attempt to buy an internet
• 22:17
  giant and a startup can build a chip 58 times
• 22:20
  larger than the industry standard just to run
• 22:22
  the AI we use every day, what other seemingly
• 22:25
  impossible tech rules are about to be breaking
• 22:27
  in our daily lives by the end of this year? It
• 22:30
  definitely makes you wonder what we consider
• 22:33
  impossible today. that will just be standard
• 22:35
  practice tomorrow. Absolutely. Keep your eyes
• 22:37
  open because the dots are only going to get more
• 22:39
  interesting to connect. We're here to help at
• 22:41
  Snarful, so reach out on our website if you have
• 22:43
  any more questions.