Latest / Connecting the Dots / AI Infrastructure Volatility, Stablecoins Go Mainstream, and Adobe’s Firefly Upgrade

Transcript

AI Infrastructure Volatility, Stablecoins Go Mainstream, and Adobe’s Firefly Upgrade · 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 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. Welcome to Connecting the Dots. I'm Alex.
• 0:18
  And I'm Morgan. And it is Tuesday, December 16th,
• 0:21
  2025. Today, we are connecting some, uh... really
• 0:25
  fascinating dots that link these massive market
• 0:28
  jitters with a huge shift in financial plumbing.
• 0:31
  And you know, the quest for absolute precision
• 0:33
  in creative AI, it's a day of really conflicting
• 0:37
  data. It really is. If you just look at the headlines,
• 0:40
  you see fear, but then you look at the actual
• 0:42
  contracts being signed and you just see this
• 0:45
  explosive, undeniable growth. So we're going
• 0:48
  to walk you through that push and pull this morning.
• 0:50
  But before we jump in, let's get the national
• 0:52
  picture. OK. If you're dealing with that heavy
• 0:54
  lake effect snow east of Lake Ontario, please
• 0:57
  stay safe. And for our listeners in the Pacific
• 0:59
  Northwest, get ready. Yeah that second Pacific
• 1:02
  cold front is coming. Another atmospheric river
• 1:04
  arriving tomorrow. It's bringing gusty winds,
• 1:07
  mountain snow, and unfortunately a renewed threat
• 1:09
  of flooding. And right here in our backyard in
• 1:12
  Sacramento we're seeing that classic dense fog
• 1:14
  this morning. It should clear up a bit though
• 1:16
  to a high of 53 degrees with a slight chance
• 1:18
  of showers later. Alright, let's talk markets
• 1:20
  because this volatility really sets the stage
• 1:23
  for our whole discussion today. It does. The
• 1:25
  Dow Jones closed Monday just slightly down at
• 1:27
  48 ,279 .40. That's a dip of what, 0 .21 % on
• 1:32
  the day. Right. But it's still holding on to
• 1:34
  some weekly gains. The S &P 500 was basically
• 1:37
  flat, closed at 6 ,803 .41. And we keep watching
• 1:41
  Bitcoin, which had a nice little daily bump.
• 1:44
  It gained almost a percent to close at $87 ,270.
• 1:48
  It's still working to claw back from a pretty
• 1:50
  significant drop over the last week. Right. And
• 1:52
  that mix of these minor daily ups and downs set
• 1:57
  against this bigger weekly fear. That's exactly
• 1:59
  what we're seeing reflected in our first big
• 2:01
  story today. The AI infrastructure sector. Absolutely.
• 2:04
  But first, let's talk about something a little
• 2:06
  more stable. Thanks to our first sponsor. Good
• 2:08
  one. Are you looking for that perfect gift for,
• 2:10
  you know, the techie friend who also secretly
• 2:13
  loves classic rock or their favorite team. We're
• 2:16
  huge fans of Old Glory. They're an iconic music
• 2:18
  and sports fan merch store. They have great stuff
• 2:20
  and it's just in time for the holidays. And if
• 2:23
  you shop now and use the promo code SNARFL at
• 2:26
  checkout, you will get 15 % off your entire purchase.
• 2:29
  That's Old Glory at OldGlory .com. Okay, let's
• 2:33
  unpack this Morningstar report. We have to start
• 2:36
  with the whiplash happening in the AI infrastructure
• 2:39
  world. It's hitting massive players. We're talking
• 2:42
  Broadcom, Oracle, CoreWeave. How can we be in
• 2:45
  a situation where investors are selling off shares?
• 2:48
  I mean Oracle is down almost 50 % since September.
• 2:52
  The entire tech industry is screaming about AI
• 2:56
  demand. That is the ultimate paradox. And it
• 2:58
  all boils down to a really fundamental financial
• 3:00
  question. Capital expenditure versus return on
• 3:03
  investment. The classic CapEx versus ROI. Exactly.
• 3:07
  The CapEx, the sheer amount of money needed to
• 3:09
  build the foundational AI infrastructure, the
• 3:11
  chips, the servers, the data centers, it's astronomical.
• 3:15
  So investors are asking, you know, are the returns
• 3:17
  going to be fast enough and large enough to justify
• 3:20
  the billions being spent right now? They're worried
• 3:22
  the cost is just outpacing the revenue. They
• 3:24
  fear it's outpacing a realistic revenue timeline.
• 3:27
  Yeah. It sounds like, you know, typical high
• 3:29
  -growth bubble anxiety. But is that fear justified,
• 3:32
  especially when the companies themselves are
• 3:34
  reporting these massive wins? I don't think it's
• 3:37
  entirely justified. The underlying demand is
• 3:39
  relentless. I mean, we saw Broadcom and Oracle
• 3:41
  actually beat their recent revenue expectations.
• 3:45
  And the demand signal from the customer side
• 3:47
  is unanimous. As one analyst put it, and this
• 3:50
  is a great quote he said, Every single AI company
• 3:52
  on the planet is saying, if you give me more
• 3:55
  compute, I can make more revenue. So the bottleneck
• 3:58
  is supply, not demand. It's all supply. And Broadcom
• 4:01
  just provided this stunning counter narrative
• 4:04
  to all that negative sentiment. That $10 billion
• 4:07
  order is just staggering. It is a massive market
• 4:10
  moving vote of confidence. It really is. Their
• 4:13
  shares surged 8 .34 % yesterday on that news.
• 4:17
  $10 billion AI chip order. And an order of that
• 4:20
  size. That's not coming from a small player.
• 4:22
  Oh, no. That's coming from a single hyperscale
• 4:24
  cloud provider that's building out their entire
• 4:26
  next generation infrastructure. $10 billion is
• 4:29
  enough to make any investor just put their worries
• 4:33
  on hold for a bit. For a little while, yeah.
• 4:35
  So what is Broadcom doing differently to pull
• 4:37
  in a contract like that, especially when NVIDIA
• 4:39
  just dominates the conversation? They have a
• 4:42
  brilliant competitive strategy. It's all focused
• 4:45
  on high margin customization. They avoid selling
• 4:49
  what's called merchant silicon. Okay, what's
• 4:51
  that? Think of that as like the general purpose
• 4:53
  off -the -shelf chips that many companies can
• 4:56
  use. That's Inzidia's traditional playground.
• 4:59
  Instead, Broadcom specializes in custom accelerators.
• 5:03
  Okay, so help us connect that idea. What does
• 5:04
  a custom accelerator mean in, you know, in practice?
• 5:08
  Okay, think of it like this. Nvidia sells you
• 5:10
  a high -performance engine for a really fast
• 5:13
  SUV. It works great for basically everyone. Broadcom
• 5:16
  works with a company, say a Google or a Meta,
• 5:19
  to design a bespoke engine specifically for their
• 5:22
  unique Formula One racing car. Ah, so it's not
• 5:25
  general purpose at all. It's totally specialized.
• 5:28
  Exactly. It's tuned for their specific workload,
• 5:31
  their specific data center cooling, their specific
• 5:33
  software architecture, and that customization,
• 5:36
  that generates higher margins for Broadcom. And
• 5:39
  I'm guessing it locks the customer in. Critically,
• 5:41
  it creates stickier customer relationships. Once
• 5:44
  a tech giant bases their multi -billion dollar
• 5:47
  AI stack on Broadcom's custom silicon, switching
• 5:50
  to a competitor becomes prohibitively expensive,
• 5:53
  they lock them in. That's a genius move. It avoids
• 5:56
  the direct street fight with NVIDIA. Completely.
• 5:59
  Okay, so let's talk about CoreWeave. They've
• 6:01
  been the definition of volatility. A 50 % crash
• 6:04
  followed by a rapid surge. What's their story?
• 6:07
  Core Weave's financials are strong. I mean, $1
• 6:09
  .36 billion in Q3 revenue. But they are a major
• 6:13
  beneficiary of all this broader AI speculation.
• 6:16
  The stock's jumpy. Very jumpy. The volatility
• 6:18
  shows investor nerves, for sure. But their biggest
• 6:21
  recent move addresses a really crucial pain point
• 6:23
  that's going to win them customers, regardless
• 6:25
  of market sentiment. And that pain point is data
• 6:28
  transfer fees or egress fees, which I know they
• 6:32
  sound boring, but they are absolutely crushing
• 6:34
  for large businesses. They are devastating. Imagine
• 6:37
  you have a massive warehouse of proprietary data
• 6:40
  and that warehouse is located inside a major
• 6:43
  cloud provider like AWS. OK. If you decide to
• 6:46
  move that entire data warehouse over to a competitor
• 6:49
  like CoreWeave, the egress fees that AWS charges
• 6:53
  you just to pull your data out. can sometimes
• 6:55
  rival the cost of storing the data for an entire
• 6:58
  year. Wow. It's basically a financial punishment
• 7:01
  for switching. That's what it is. It kills competition
• 7:03
  and it traps customers. So what's CoreWeave's
• 7:06
  answer? It's their zero egress migration program.
• 7:09
  They are simply eliminating those costly data
• 7:11
  transfer fees when customers move their data
• 7:13
  sets from AWS, Google Cloud, or Azure over to
• 7:17
  CoreWeave. That's smart. By removing that financial
• 7:19
  barrier, they make their whole proposition instantly
• 7:22
  more attractive, especially for companies trying
• 7:24
  to optimize AI workloads, which require moving
• 7:27
  just colossal amounts of data. It's smart, aggressive
• 7:30
  competition. It proves they're thinking about
• 7:32
  customer friction and they have the commitments
• 7:34
  to back it up, don't they? Yeah, they do. Despite
• 7:37
  the market nerves, their customer pipeline is
• 7:39
  solid. They have a total backlog of fifty five
• 7:42
  point six. billion dollars, including performance
• 7:46
  obligations of 50 billion with major clients
• 7:49
  like meta and open AI. The market might be skeptical
• 7:53
  about ROI, but the customers are absolutely committed
• 7:55
  to building. Okay. So moving from that volatile
• 7:58
  AI infrastructure to something more. Well, stable
• 8:04
  financial infrastructure. We have a massive piece
• 8:07
  of news about visa and their embrace of digital
• 8:10
  currency here in the U S. This is a truly foundational
• 8:13
  change. What's the announcement? Visa announced
• 8:15
  they're going to allow U .S. institutions, we're
• 8:18
  talking banks like Cross River Bank and Lead
• 8:20
  Bank, to settle transactions using Circle's USDC
• 8:23
  stablecoin. And this is running on the Solana
• 8:25
  blockchain. That's right. This is the first full
• 8:28
  U .S. deployment of stablecoin settlement for
• 8:30
  a massive payments network. So for the average
• 8:32
  listener tapping their card at the grocery store,
• 8:34
  what changes for them? Functionally, nothing
• 8:37
  changes for the cardholder. You still swipe,
• 8:39
  the payment is authorized instantly, same as
• 8:41
  it ever was. What this revolutionizes is the
• 8:44
  messy, slow, expensive backend, the rails of
• 8:47
  the system. Explain that. When you buy something,
• 8:50
  the two banks involved, the one that issued your
• 8:53
  card and the merchant's bank. they have to settle
• 8:55
  up that money movement. Traditionally, they rely
• 8:58
  on systems like ACH or Fedwire, which only operate
• 9:02
  during specific banking hours and can take three
• 9:05
  to five days to finalize. And those delays essentially
• 9:08
  just tie up cash for days. Exactly. Think of
• 9:11
  it like having your cash stuck in transit on
• 9:13
  a multi -day journey. If you're a major bank,
• 9:16
  you have billions of dollars tied up in these
• 9:18
  slow settlement cycles. Which affects their liquidity,
• 9:21
  their ability to lend. It affects their liquidity
• 9:23
  management, exactly. By using USDC on Solana,
• 9:26
  that settlement happens in minutes. 247 .365.
• 9:29
  So the banks care because instant settlement
• 9:31
  means their capital is instantly freed up. It's
• 9:34
  all about maximizing that working capital. That
• 9:36
  is the core business benefit. It reduces operational
• 9:39
  friction, it improves treasury efficiency, and
• 9:41
  it gives banks much better liquidity control.
• 9:44
  And we know this works. We do. Visa's global
• 9:47
  stablecoin pilots have been working flawlessly.
• 9:49
  They're already handling an annualized run rate
• 9:52
  of over $3 .5 billion in volume outside the U
• 9:55
  .S. So bringing this to the U .S. is the crucial
• 9:58
  final step. It is. And Visa is deeply integrated
• 10:02
  here. They're not just using the ecosystem. What
• 10:05
  do you mean? Visa is a design participant for
• 10:07
  Circle's upcoming ARK blockchain. And the timing
• 10:11
  here is also crucial. We've seen these pilots
• 10:13
  globally, but The U .S. rollout is happening
• 10:16
  now, partly due to what sources describe as a
• 10:19
  supportive, relaxed regulatory environment. It
• 10:23
  allows them to modernize this crucial domestic
• 10:26
  infrastructure without, let's say, immediate
• 10:29
  heavy political resistance. They're leveraging
• 10:31
  the moment to upgrade the financial system for
• 10:34
  the digital age. Okay, finally, let's jump into
• 10:36
  the world of creative tech with Adobe Firefly.
• 10:39
  I think anyone who's played with generative AI
• 10:40
  knows the frustration. You get 90 % of what you
• 10:43
  want, but that last 10%, the weird artifact,
• 10:46
  the misplaced object, the camera angle that's
• 10:48
  just slightly off it, ruins the whole clip. And
• 10:51
  historically you had to scrap it and start over.
• 10:53
  Exactly. That lack of precision, that lack of
• 10:56
  fine -grained control, that has been the major
• 10:58
  barrier preventing professional creators from
• 11:01
  fully adopting generative video. So what's Adobe
• 11:03
  doing? They're aiming to solve that directly
• 11:06
  with a really significant update to their Firefly
• 11:08
  platform, which now introduces precise prompt
• 11:12
  -based editing. You've all been there, trying
• 11:14
  to generate a simple picture, only for the character
• 11:16
  to have, like, Seven fingers or the Sun is in
• 11:19
  the wrong place right that level of frustration
• 11:21
  just kills a professional workflow So how does
• 11:24
  this new prompt to edit feature actually work
• 11:27
  instead of the AI being an all -or -nothing generator?
• 11:30
  It becomes a malleable editing tool you apply
• 11:33
  natural language instructions directly to the
• 11:35
  existing AI generated clip OK, so you don't trash
• 11:38
  it. You don't trash it. You just tell it change
• 11:39
  the sky to overcast or lower the saturation on
• 11:42
  the background or zoom in slightly on the main
• 11:45
  subject. The edits are applied in context, preserving
• 11:47
  everything else you liked. That's a game changer.
• 11:50
  That level of direct refinement suddenly makes
• 11:52
  generative AI viable for film and broadcast work,
• 11:56
  where total control is essential. It really does.
• 11:59
  And they've added proper creator tools, too.
• 12:01
  There's a full timeline view for adjusting frames
• 12:04
  and sound. What else? What I find particularly
• 12:07
  innovative is the new control over camera motion.
• 12:10
  You can now upload a start frame and a reference
• 12:12
  video that shows a specific camera motion, say
• 12:15
  a slow crane shot, and instruct the Firefly video
• 12:19
  model to recreate that precise movement in your
• 12:22
  generated scene. It takes a randomness out of
• 12:25
  the AI's camera work. Here's where it gets really
• 12:27
  interesting though. Adobe isn't trying to do
• 12:29
  it all themselves. No, they're building Firefly
• 12:32
  as an all -in -one hub by integrating the best
• 12:35
  third -party models on the market. It's the classic
• 12:38
  Adobe playbook. It is. They understand they are
• 12:41
  the career partner for creative pros. Their goal
• 12:44
  is to make Firefly the single subscription hub
• 12:46
  for everything, much like Photoshop dominated
• 12:49
  photography for decades. To do that, they have
• 12:51
  to integrate the bleeding edge. So what are some
• 12:53
  of those key integrations? They've integrated
• 12:55
  Topaz Labs Astra, which is a highly respected
• 12:58
  model used to upscale video quality to professional
• 13:01
  resolutions 1080p or even 4K. That's critical
• 13:05
  for production. Absolutely. And they've integrated
• 13:08
  Black Forest Labs FLUX .2, which is a new image
• 13:11
  generation model that's now available across
• 13:14
  Firefly and the Photoshop desktop apps. So they're
• 13:17
  giving users the best tools instantly, whether
• 13:20
  they're built in -house or by a partner. That's
• 13:22
  a good gift. And to encourage everyone to explore
• 13:25
  this new level of precision and power, I see
• 13:28
  they're offering an incentive. They are betting
• 13:30
  heavily that once you experience this level of
• 13:32
  control and quality, you won't go back. Adobe
• 13:34
  is offering unlimited generations for specific
• 13:37
  subscription tiers, covering both their image
• 13:39
  and video models until January 15th. A big push
• 13:43
  to get creators on board. Massive push to get
• 13:45
  everyone into this new ecosystem that blends
• 13:48
  their native Firefly models with the best -in
• 13:51
  -class third -party tech. Fantastic breakdown
• 13:53
  of the tech world's conflicting signals today.
• 13:56
  Yeah. Before we wrap up, we have to thank another
• 13:58
  great sponsor, Scony. If you are looking for
• 14:01
  premium golf shoes, Sconee is the place to go.
• 14:04
  And just like with Old Glory, you can use the
• 14:06
  promo code SNARFUL at checkout to get 15 % off
• 14:09
  your entire purchase. That's sconee at sconee
• 14:12
  .com. So what does this all mean for you, the
• 14:15
  listener? We covered three key points today,
• 14:17
  all showing how technology is simultaneously
• 14:20
  solving these major friction points across three
• 14:23
  very different domains. One, the AI infrastructure
• 14:26
  market is highly volatile. It's split between
• 14:29
  this massive, necessary capital investment like
• 14:31
  Broadcom's $10 billion order and this deep investor
• 14:35
  concern over when the profits will actually show
• 14:37
  up. We see that in the stock volatility of Oracle
• 14:39
  and CoreWeave. Two, Visa is fundamentally modernizing
• 14:43
  the plumbing of the U .S. financial system. They're
• 14:46
  deploying the stablecoin USDC on the Solana blockchain.
• 14:49
  Which cuts operational friction and drastically
• 14:51
  improves liquidity and speed for banks and fintechs.
• 14:54
  It's pushing traditional finance into the 2047
• 14:56
  digital era. And three. Adobe is fighting AI
• 14:59
  imprecision by giving creators control. By adding
• 15:02
  prompt -based editing and integrating top third
• 15:04
  -party models like Astra and FLUX .2 into Firefly,
• 15:08
  they're transforming generative AI from a fun
• 15:10
  novelty into a precision tool for professionals.
• 15:14
  Yeah, the overarching theme here is the ongoing
• 15:16
  evolution of control. Whether it's financial
• 15:18
  control over working capital, competitive control
• 15:21
  over customers by removing those huge egress
• 15:23
  fees, or creative control over a single frame
• 15:27
  of video, the market rewards those who eliminate
• 15:30
  friction.