Latest / Connecting the Dots / AI Chips, Smuggling Controls, and a Supersonic Pivot to Power

Transcript

AI Chips, Smuggling Controls, and a Supersonic Pivot to Power · Connecting the Dots

• 0:00
  Today's tech news moves fast, but understanding
• 0:03
  it shouldn't be hard. That's why Snarfle Solutions
• 0:06
  Group brings you Connecting the Dots, 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. Good morning, I'm Alex. And I'm Morgan.
• 0:18
  Welcome to Connecting the Dots. It is Wednesday,
• 0:20
  December 10th, 2025. And we have some incredible
• 0:23
  stories today. It feels like everything is connecting.
• 0:26
  Global policy, tech hardware, even the power
• 0:29
  grid itself. It really is. We're going to be
• 0:31
  looking at how these huge high level decisions
• 0:34
  actually connect back to our economy and, you
• 0:37
  know, our daily lives. Before we jump in, though,
• 0:39
  let's get a quick check on the weather and markets.
• 0:41
  Here in Sacramento, we're looking at areas of
• 0:43
  dense fog today. Yep. And a high of only 46 degrees,
• 0:47
  so bundle up. Nationally, that atmospheric river
• 0:50
  is still hitting the Pacific Northwest. So that
• 0:52
  means more gusty winds, heavy rain, and unfortunately,
• 0:55
  potential for flooding and landslides. Right.
• 0:58
  And on the other side of the country, a clipper
• 0:59
  system is crossing the Great Lakes and Northeast,
• 1:02
  bringing some pretty heavy snow. OK, let's look
• 1:05
  at the markets. As of Tuesday's close, the Dow
• 1:07
  Jones was at 47 ,560 .29. Down just a bit, about
• 1:11
  0 .34%. The S &P 500 closed at 6 ,840 .51, which
• 1:16
  was almost perfectly flat. And Bitcoin, as of
• 1:18
  this morning, is sitting around 91 ,979, down
• 1:22
  about 0 .78%. Alright, before we get to our main
• 1:25
  stories, we want to thank our sponsor, Old Glory.
• 1:28
  With the holidays coming up, if you have a techie
• 1:30
  friend who also loves music or sports, Old Glory
• 1:33
  is the place to go. It's an iconic fan merch
• 1:36
  store. They have really great stuff and we've
• 1:38
  got a great offer for you. Use our promo code
• 1:40
  SNARFL at checkout. That's S -N -A -R -F -U -L
• 1:44
  and you'll get 15 % off your entire purchase.
• 1:47
  Just head over to oldglory .com. Okay, let's
• 1:50
  get into it. Our internal SNARFL Tech article
• 1:53
  report, what we call the Morningstar, is just
• 1:55
  packed today. We've got three major stories.
• 1:58
  And they all seem to feed into each other. We're
• 2:00
  talking about AI chips, a huge U .S. trade policy
• 2:04
  reversal, and an aerospace company pivoting.
• 2:08
  to power the AI grid. That's the one. It sounds
• 2:10
  wild, but it makes a surprising amount of sense
• 2:12
  when you connect the dots. I can't wait to get
• 2:14
  to that one. But let's start with the chipmaker
• 2:16
  at the center of everything, NVIDIA. Right. So
• 2:18
  our first story is about NVIDIA trying to solve
• 2:21
  this growing problem of chip smuggling. They're
• 2:23
  developing a new software update, and it's an
• 2:25
  optional one that can actually verify the country
• 2:28
  where its GPUs are operating. It's essentially
• 2:30
  location tracking for your graphics card. And
• 2:33
  the way it works is pretty clever. It uses the
• 2:37
  chip's confidential computing features. Okay,
• 2:40
  let's pause there. What exactly are confidential
• 2:43
  computing features? That's a great question.
• 2:45
  Think of it like a secure vault built directly
• 2:48
  into the chip. It's an isolated area where data
• 2:51
  can be processed without the main operating system
• 2:54
  or even a hacker being able to see it. So it's
• 2:57
  a feature designed for high security. Exactly.
• 3:00
  And NVIDIA is using that secure trusted part
• 3:02
  of the chip to measure the time delay when it
• 3:05
  communicates with NVIDIA's servers. And that
• 3:07
  delay lets them estimate the chip's physical
• 3:09
  location. Precisely. It's a technical solution
• 3:12
  to a huge political problem. The White House
• 3:14
  and Congress have been pushing for a way to stop
• 3:17
  these chips from being smuggled into restricted
• 3:19
  countries. We've seen those criminal cases, right?
• 3:21
  Yeah. Smuggling rings trying to move over 160
• 3:23
  million dollars worth of Nvidia chips to China.
• 3:26
  That's the exact problem this is meant to solve.
• 3:28
  But, you know, it creates this immediate tension.
• 3:31
  Of course, because China's cybersecurity regulator
• 3:34
  is now questioning Nvidia, worried that this
• 3:37
  feature could be a U .S. back door. And that's
• 3:39
  the fear, isn't it? that it bypasses the chip
• 3:42
  security. NVIDIA officially denies any backdoors.
• 3:46
  What's their official line? They're calling it
• 3:48
  a customer -installed software agent that just
• 3:51
  helps data center operators monitor fleet health,
• 3:54
  integrity and inventory. But the perception is
• 3:56
  what matters. It's a real tightrope walk for
• 3:58
  them. They have to satisfy U .S. regulators without
• 4:01
  losing the trust of their biggest international
• 4:03
  markets. It's a huge precedent. A piece of hardware
• 4:05
  that can call home to verify its location. That
• 4:09
  really blurts the line between technology and
• 4:12
  policy enforcement. And that idea leads us perfectly
• 4:15
  into our second story, which is an absolute bombshell.
• 4:19
  A massive one. President Trump has approved Nvidia
• 4:22
  to sell its H200 AI chips to China. This is a
• 4:26
  complete reversal. For years, U .S. policy was
• 4:28
  all about restricting China's access to high
• 4:30
  -end AI chips. Exactly. And the H200 is not a
• 4:34
  small step up. We need to be clear about that.
• 4:35
  It's part of the older Hopper generation, so
• 4:37
  it's not their absolute best chip. Right. The
• 4:39
  new Blackwell and upcoming Rubin chips are more
• 4:42
  powerful. But compared to what China was previously
• 4:45
  allowed to buy, the H20 chip, The H -200 has
• 4:48
  nearly 10 times the processing performance. It's
• 4:51
  a gigantic leap. So this is a major injection
• 4:54
  of computing power into China's AI development.
• 4:57
  What's the catch? The catch is financial. The
• 5:00
  approval comes with a very specific condition.
• 5:03
  A 25 % surcharge, basically a tariff, must be
• 5:07
  paid to the U .S. government on the revenue from
• 5:09
  each sale. Wait, so the U .S. government is now
• 5:12
  taking a direct cut from Nvidia's sales in China?
• 5:15
  That's right. It's a fascinating policy pivot.
• 5:17
  Instead of an outright ban, it's now a tax. The
• 5:20
  U .S. government is positioning itself to profit
• 5:22
  directly from the international demand for its
• 5:25
  strategic tech. It's an attempt to balance everything,
• 5:28
  letting Nvidia get back billions in lost business
• 5:30
  while still withholding the absolute top -tier
• 5:33
  chips and, you know, creating a new revenue stream
• 5:35
  for the government. But the reaction domestically
• 5:37
  was immediate. and very strong. I saw that. Democratic
• 5:43
  senators like Elizabeth Warren called it a colossal
• 5:46
  economic and national security failure. The argument
• 5:49
  being that this gives Chinese competitors like
• 5:51
  DeepSeek or Huawei a massive advantage in training
• 5:54
  their AI models. It really highlights that core
• 5:57
  debate. Do you prioritize U .S. corporate revenue
• 6:01
  or the U .S. tech lead? And this 25 % tariff
• 6:05
  is an attempt to do both, which is why it's so
• 6:07
  controversial. But here's the other twist in
• 6:09
  the story. Even with U .S. approval, it's not
• 6:12
  clear if China will even let its companies buy
• 6:14
  the H -200. Oh, right. Because Beijing had been
• 6:17
  pushing its domestic companies to use Chinese
• 6:19
  -made chips instead of the less powerful H -20.
• 6:22
  Exactly. They might see this as another form
• 6:24
  of dependence. So Beijing is considering its
• 6:26
  own conditions, like requiring special applications
• 6:28
  for use, which could make the chips less attractive.
• 6:31
  So this whole major policy shift from the U .S.
• 6:33
  could just fizzle out if Beijing decides to quietly
• 6:36
  block it. It could. It just goes to show how
• 6:38
  every single move in the tech space is now a
• 6:41
  geopolitical chess move. Wow. Okay, from geopolitics
• 6:45
  to, well, literal power. Our third story about
• 6:49
  Boom Supersonic. This one is something else.
• 6:52
  It really is. So, boom, the company building
• 6:55
  the Mach 1 .7 Overture airliner just announced
• 6:58
  they are re -sequencing their strategy. Re -sequencing.
• 7:00
  What does that mean? It means they're now focusing
• 7:03
  first on developing an electricity generating
• 7:05
  natural gas turbine called super power. A supersonic
• 7:08
  jet company is now making power plants. That
• 7:10
  sounds like a delay, a major detour from their
• 7:13
  goal. It sounds like it, but it's actually the
• 7:15
  opposite. It's an accelerant. They realized their
• 7:18
  engine technology solves a massive, immediate
• 7:21
  problem. Which is what? The AI energy crisis.
• 7:24
  Data centers are facing a genuine energy crunch.
• 7:27
  We're seeing reports of multi -million dollar
• 7:29
  GPU racks sitting idle just because they can't
• 7:32
  get enough electricity. So data center operators
• 7:34
  are starting to build their own power plants
• 7:35
  right on site. Yeah. Behind the meter power.
• 7:38
  Exactly. And this is where Boom's technology
• 7:40
  comes in. The existing turbines that power these
• 7:42
  facilities are based on old subsonic jet engines.
• 7:46
  Those engines are optimized for cruising at high
• 7:48
  altitudes, where it's like minus 50 degrees Fahrenheit.
• 7:51
  They're built for the cold. Which is the exact
• 7:53
  opposite of a data center in a hot climate like
• 7:56
  Arizona or Texas. Precisely. In 110 degree heat,
• 7:59
  those legacy turbines lose 20 to 30 percent of
• 8:02
  their power output. It's called thermal de -rate.
• 8:05
  Plus, they need huge amounts of water for cooling.
• 8:08
  So what's Boom's advantage? Their engine core,
• 8:11
  the Symphony, was designed for the Overture.
• 8:13
  It's built to operate continuously at Mach 1
• 8:16
  .7 at 60 ,000 feet, where effective temperatures
• 8:19
  can reach 160 degrees. It's designed for extreme
• 8:23
  heat. So their super power turbine doesn't lose
• 8:25
  power in hot weather. It maintains its full 42
• 8:28
  megawatt output, even at 110 degrees. And, critically,
• 8:32
  it's entirely waterless. It's a perfect solution
• 8:35
  for the AI industry's biggest bottleneck right
• 8:37
  now. And this is already happening. It is. They've
• 8:39
  secured $300 million in new funding and have
• 8:42
  a launch order for 29 turbines from a customer
• 8:45
  named Crusoe. So the business strategy is to
• 8:47
  use the money from selling these power turbines.
• 8:49
  To fund the final, most expensive part of developing
• 8:53
  and certifying their supersonic jet. The AI energy
• 8:56
  crisis is literally funding the return of supersonic
• 9:00
  flight. That is brilliant. And all the data they
• 9:02
  get from running these turbines 24 -7 will help
• 9:05
  certify the jet engine for the Overture. It's
• 9:07
  an incredible pivot. They're solving a huge current
• 9:10
  problem to enable their future vision. That's
• 9:12
  connecting the dots on a whole other level. Before
• 9:14
  we wrap up with our final thoughts, we'd like
• 9:16
  to thank our second sponsor today, SCONI. SCONI
• 9:19
  is a premium golf shoe store. If you've got a
• 9:22
  golfer in your life, they have some fantastic
• 9:24
  high quality options. And you can use that same
• 9:27
  promo code SNARFL at checkout for 15 % off your
• 9:30
  entire purchase. Check them out at scony .com.
• 9:33
  Okay, so let's recap today. This has been a wild
• 9:36
  ride. It really has. We started with Nvidia using
• 9:38
  this complex, confidential computing tech to
• 9:41
  track its chips around the globe to comply with
• 9:43
  U .S. law, which in turn creates friction with
• 9:46
  China. Then that massive policy reversal. The
• 9:49
  U .S. allowing H200 AI chips to be sold to China,
• 9:52
  but with this unprecedented 25 % tariff going
• 9:56
  straight to the U .S. government. And we wrapped
• 9:58
  with Boom Supersonic, a company aiming for the
• 10:01
  skies, but first pivoting to solve the energy
• 10:04
  crisis on the ground, using its supersonic engine
• 10:07
  tech to power the very AI data centers that are
• 10:11
  changing our world. The theme today is just so
• 10:13
  clear. Technology, geopolitics, and basic industrial
• 10:17
  capacity like energy are no longer separate fields.
• 10:20
  Not at all. The solutions and the constraints
• 10:23
  are coming from the most unexpected places. It's
• 10:25
  not just about who can build the fastest chip
• 10:27
  anymore. It's also about who can actually power
• 10:30
  it. And that really leaves us with a final thought
• 10:33
  for you to consider. When a company like BoomSuccess
• 10:35
  depends on solving a national energy crisis first,
• 10:39
  and another like NVIDIA effectively becomes a
• 10:41
  tax collector for the government, you have to
• 10:44
  ask. Are we seeing the end of a truly global
• 10:47
  free market tech industry? Is the entire AI economy
• 10:50
  now just becoming a kind of geopolitical utility?
• 10:53
  We're here to help it snarfle, so reach out on
• 10:55
  our website if you have any more questions. That's
• 11:00
  a wrap on today's top stories. Remember, tech
• 11:03
  doesn't have to be complicated, and we're here
• 11:06
  to help connect the dots. If you've got questions
• 11:09
  or want to learn more, visit us at snarful .com.
• 11:12
  Thanks for listening, and we'll catch you next
• 11:14
  time.