Latest / Connecting the Dots / Open-Source AI Upsets, Hollywood’s $108B Gamble, and Autonomy Pushes Forward

Transcript

Open-Source AI Upsets, Hollywood’s $108B Gamble, and Autonomy Pushes Forward · 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. I'm Alex. And I'm Morgan. And
• 0:18
  this is Connecting the Dots. Happy Tuesday. It
• 0:21
  is December 23rd, 2025. We are in that, uh...
• 0:24
  Magical, but also slightly chaotic week leading
• 0:27
  up to the holidays. Definitely chaotic. And if
• 0:29
  you're traveling, we really have to start with
• 0:31
  a quick check in on the weather because it's
• 0:33
  having a huge impact, especially on the West
• 0:35
  Coast. It truly is. We are watching a powerful
• 0:39
  atmospheric river event just unfold across California.
• 0:44
  We love the water, but this is a lot. It's a
• 0:46
  lot and it's creating some real hazards. The
• 0:49
  systems are bringing massive rainfall. with the
• 0:52
  heaviest part expected tomorrow, Wednesday, right
• 0:55
  in the LA basin. So if you're driving anywhere
• 0:57
  near the mountains this week, you've got to be
• 0:59
  cautious. The ground is saturated, which means
• 1:02
  the risks for flash flooding, mudslides, rock
• 1:05
  slides, they're not just potential anymore, they
• 1:07
  are very, very real. And for anyone heading up
• 1:09
  to the Sierra Nevada, expect several feet of
• 1:12
  heavy snow, major, major travel delays. It's
• 1:15
  a week where weather is just dominating everything.
• 1:17
  So the takeaway for you is pretty simple. If
• 1:19
  you're expecting deliveries or flying into any
• 1:21
  major West Coast hub, just anticipate disruptions,
• 1:24
  probably well into Friday. A lot of water moving
• 1:26
  through a lot of important commerce routes. Exactly.
• 1:29
  Speaking of commerce, let's pivot quickly to
• 1:31
  the markets. It was a day of slow, steady movements
• 1:36
  or that pre -holiday wind down. Yeah. Looking
• 1:38
  at yesterday's close, Tuesday, December 23rd,
• 1:41
  the Dow Jones closed up just a hair. plus 0 .07
• 1:45
  % at 48 ,353. point back in. And that small upward
• 1:50
  momentum was pretty much mirrored by the S &P
• 1:52
  500. It closed up 0 .14 % at 6 ,882 .14. We see
• 1:59
  this a lot this week, right? A quiet drift upward
• 2:01
  as managers position themselves before the year
• 2:03
  wraps. Absolutely. Finalizing their books for
• 2:06
  the year. Bitcoin, however, decided to be the
• 2:08
  exception to the calm. It took a little bit of
• 2:10
  a volatile dip. It closed down 2 .06%, settling
• 2:13
  at 86 ,707 .64. A little drama there. But overall,
• 2:18
  the major indices are holding stable. Before
• 2:20
  we jump into our look at the Morningstar report,
• 2:22
  we do want to give a quick shout out to our sponsor,
• 2:24
  Old Glory. Yes. And if you are still scrambling
• 2:26
  for those last minute, impossible to find gifts.
• 2:29
  Especially for that techie friend who claims
• 2:31
  they already have everything. Old Glory is the
• 2:34
  place to go. They are the iconic music and sports
• 2:36
  fan merchandise store. They really do have the
• 2:38
  best gear for almost any fandom. And here's the
• 2:42
  best part for you. Old Glory is offering 15 %
• 2:44
  off your entire purchase right now. All you have
• 2:46
  to do is use our special promo code. That code
• 2:49
  is SNARFUL. That's S -N -A -R -F -U -L at checkout.
• 2:53
  Find that perfect gift at OldGlory .com. And
• 2:56
  remember to use the code SNARFUL. All right,
• 2:59
  let's get into it. The three stories defining
• 3:01
  tech, geopolitics, and some very high stakes
• 3:04
  finance today. And we're starting with a massive
• 3:06
  AI coding development out of China that could
• 3:09
  fundamentally change the cost equation for enterprise
• 3:12
  software. Let's unpack this. So a Chinese AI
• 3:14
  startup, Minimax, has just released an upgraded
• 3:17
  open source language model. It's called M2 .1.
• 3:20
  And the core story here really is that this model
• 3:22
  is focusing its power on making enterprise software
• 3:25
  development, well, more accessible and efficient
• 3:27
  for everyone. Right, because for a long time
• 3:30
  these AI coding models were fantastic if you
• 3:33
  were working in, say, Python. Which is great
• 3:36
  for data science, web scripting, that kind of
• 3:38
  thing. But complex global companies, they're
• 3:40
  built on these ecosystems of multiple, often
• 3:43
  very difficult languages. Exactly. And that's
• 3:45
  what's different here. M2 .1 made a systemic,
• 3:49
  a really deliberate enhancement across the board.
• 3:51
  We're talking about the major languages that
• 3:53
  power everything. Rust, Java, Golang. C++ Kotlin.
• 3:58
  Yeah. It covers the complete development chain,
• 4:00
  from low -level systems all the way up to the
• 4:03
  application layer. That is a huge industry shift.
• 4:06
  Previous models had this notorious weakness with
• 4:10
  non -Python code. Especially for native mobile
• 4:12
  development. Right. And Minimax specifically
• 4:14
  targeted and strengthened native Android and
• 4:16
  iOS development with M2 .1, which solves a massive
• 4:20
  pain point for so many companies. And here's
• 4:22
  where it gets really interesting. They didn't
• 4:24
  just make it functionally better. They made it
• 4:26
  visually creative. They're focusing on what they
• 4:28
  call design comprehension and aesthetic expression.
• 4:31
  Or the much more fun term, vibe coding. I love
• 4:34
  that. Vibe coding. Right. It allows the model
• 4:37
  to move past just writing utility functions and
• 4:41
  into generating, you know, highly visual, high
• 4:44
  quality front -end designs. The sources had some
• 4:46
  incredible examples, like generating a complex
• 4:50
  gesture interactive 3D Christmas tree. Or that
• 4:53
  interactive iOS home screen widget with a sleeping
• 4:55
  Santa you could click to wake up. That's not
• 4:58
  simple stuff. Not at all. That kind of capability
• 5:01
  changes who needs to be in the room for a project.
• 5:04
  If the AI can handle that foundational wide coding,
• 5:07
  it shifts the role of the design engineer to
• 5:09
  more oversight and high level concepts rather
• 5:11
  than pixel pushing. And this advanced capability
• 5:13
  is just stunningly affordable. The original M2
• 5:16
  model was already twice as fast as a competitor
• 5:19
  like Clawed Sonnet 4 .5. At only 8 % of the cost
• 5:22
  to run, that is what democratizes this technology.
• 5:24
  At cost -speed advantage is everything. And M2
• 5:27
  .1 doesn't just maintain that lead, it pushes
• 5:29
  it further. It now matches or even exceeds Clawed
• 5:33
  Sonnet 4 .5 on very specialized tasks like generating
• 5:36
  test cases or conducting code reviews. Plus,
• 5:39
  I saw they engineered the responses to be more
• 5:41
  concise. Yes, which means the model uses fewer
• 5:44
  tokens. Can you quickly break down what token
• 5:47
  consumption means for, say, a business's bottom
• 5:51
  line? Absolutely. Think of tokens as the basic
• 5:54
  unit of cost and time. The fewer tokens the AI
• 5:56
  consumes to give you a quality answer, the cheaper
• 5:59
  and faster the whole process is. So by making
• 6:02
  M2 .1 more concise, they directly reduce the
• 6:05
  cost of engineering assistance. Which makes it
• 6:07
  scalable for small dev teams anywhere. better,
• 6:10
  faster, and much cheaper. But the real game changer
• 6:12
  is what they're calling the digital employee.
• 6:14
  This is where it moves beyond just coding assistance
• 6:17
  and into full end -to -end office automation.
• 6:20
  Okay, so the model takes high -level text instructions.
• 6:23
  And then it autonomously executes the steps by
• 6:26
  controlling the mouse and keyboard on the web.
• 6:28
  Wait, I'm curious though. Historically, this
• 6:31
  kind of end -to -end automation has been just
• 6:33
  plagued with integration failures and security
• 6:35
  risks. Are they really solving that or just the
• 6:37
  coding part? That's a great question about long
• 6:39
  -term oversight. But for now, they're demonstrating
• 6:42
  capability. For example, you could tell it. Collect
• 6:45
  all outstanding equipment requests from Slack,
• 6:48
  find pricing on three internal servers, calculate
• 6:51
  the total cost, and update the project management
• 6:54
  sheet. And it handles all of that. The navigation,
• 6:57
  the data parsing, the updates, all on its own.
• 7:00
  Exactly. It's moving AI from a tool you use for
• 7:04
  a task to a task. Yeah. an actual independent
• 7:08
  team member handling full workflows. That has
• 7:10
  massive implications for team efficiency. It's
• 7:13
  genuinely disruptive. So if story one showed
• 7:16
  us the race to code the future, story two reveals
• 7:19
  the strategic competition over the physical hardware
• 7:21
  that powers it, specifically the battle over
• 7:24
  autonomous chips. That's a perfect transition.
• 7:26
  We're seeing a different kind of tech acceleration
• 7:28
  in China focused squarely on autonomous driving
• 7:30
  and is driven heavily by these geopolitical goals
• 7:33
  for independence. Right. Despite some high profile
• 7:35
  safety concerns like that fatal Xiaomi crash
• 7:38
  back in March, Chinese automakers are just accelerating
• 7:41
  development faster than ever. And the key driver
• 7:44
  here is the push for full domestic chip independence.
• 7:48
  The pace is breathtaking. Let's look at XPeng.
• 7:51
  They've just debuted an upgraded system, VLA
• 7:53
  2 .0, that's explicitly designed to surpass Tesla's
• 7:57
  FSD. And the training data scale they used is
• 8:00
  just... it's astounding. They trained it on nearly
• 8:03
  a hundred million video clips from real -world
• 8:05
  driving. To put that in perspective, that's like...
• 8:08
  What, 65 ,000 years of experience for a human
• 8:11
  driver? Something like that. Yeah. And that sheer
• 8:13
  volume allows the system to handle incredibly
• 8:16
  nuanced situations. The detail that really stood
• 8:19
  out to me was its ability to recognize human
• 8:21
  gestures. Oh, yeah. A construction worker signals
• 8:24
  the car to stop. It stops. They wave it through
• 8:26
  a chaotic intersection. It proceeds safely. That
• 8:29
  requires contextual comprehension, not just seeing
• 8:32
  an obstacle. And the competitive benchmarks show
• 8:34
  it. Xpeng reported that VLA 2 .0 needed five
• 8:37
  times fewer driver interventions compared to
• 8:39
  the version of Tesla's FSD available in China.
• 8:41
  And this isn't just an Xpeng thing. Volkswagen
• 8:44
  is already adopting this system. Which brings
• 8:46
  us to the core tension. The chip race. The hardware
• 8:50
  is dominated by American suppliers. They hold
• 8:53
  70 % of the high -end autonomous chip market
• 8:56
  in China. Aging sees that as a strategic vulnerability,
• 9:00
  so they've set a very aggressive target. Full
• 9:02
  domestic research and production of automotive
• 9:04
  chips by 2027. That is less than two years away.
• 9:08
  Who's the key player making that possible? Horizon
• 9:10
  Robotics. They're government -backed and have
• 9:13
  become the primary challenger. Their latest chip,
• 9:16
  the Journey 6, boasts 560 TOPS of computing power.
• 9:21
  OK, for our listeners, what does 560 TOPS mean
• 9:24
  in simple terms? TOPS is Tera Operations Per
• 9:27
  Second. It's just a measure of the chip's processing
• 9:29
  power for AI tasks. image recognition, path planning.
• 9:33
  And 560 is a lot. It's an extremely high level
• 9:35
  of processing power, which you need to run the
• 9:37
  complex algorithms for full autonomy. And this
• 9:40
  chip is already being integrated by major Chinese
• 9:42
  automakers. And even international brands like
• 9:44
  Volkswagen and Bosch are using it for their vehicles
• 9:47
  sold in the Chinese market. It's smart maneuvering.
• 9:49
  It ensures they have a tech stack that works
• 9:51
  regardless of, say, U .S. trade limitations.
• 9:54
  It's all about controlling the technological
• 9:56
  future of the industry. It's a national strategic
• 9:58
  competition. So for a quick stats check here
• 10:00
  in the U .S., we hear self -driving all the time,
• 10:03
  but most consumer tech, FSD, Blue Cruise, it's
• 10:06
  still level two. Meaning the driver has to constantly
• 10:09
  supervise the system. Exactly. Level three is
• 10:12
  conditional hands -off driving, which is what
• 10:13
  China is pushing toward. Here, we only have limited
• 10:17
  L3 systems, like Mercedes Drive Pilot, and only
• 10:20
  in two states at low speeds. And GM Super Cruise
• 10:23
  is slated for a Level 3 upgrade around 2028.
• 10:26
  That's the plan. So while XPeng is cloning a
• 10:29
  huge lead in interventions now, the real regulatory
• 10:32
  battle over L3 deployment is still ahead. A race
• 10:35
  driven by both tech and geopolitics. Okay, let's
• 10:38
  pivot entirely from self -driving chips to some
• 10:40
  very high stakes corporate drama. Our third story
• 10:43
  involves a hostile takeover battle that has huge
• 10:45
  implications for consumer choice and the future
• 10:48
  of film. We're talking about Warner Bros. Discovery
• 10:50
  WBD and the fight for its soul between Netflix
• 10:53
  and a coalition led by Paramount Skydance. So
• 10:56
  the WBD board is reviewing a hostile takeover
• 10:59
  bid from the Paramount Group. A move that could
• 11:02
  completely derail the massive asset sale they'd
• 11:05
  previously planned with Netflix. Let's just look
• 11:07
  at the financial stakes here because they are
• 11:09
  staggering. The original agreement, the Netflix
• 11:12
  deal, had an enterprise value of $82 .7 billion.
• 11:17
  And for that, Netflix was going to get the WBD
• 11:20
  Studio, HBO and HBR Max. Massive consolidation.
• 11:24
  And then the Paramount bid arrived. This is an
• 11:27
  all -cash offer for the entire company. $30 per
• 11:30
  share. Which gives the deal an enterprise value
• 11:32
  of $108 .4 billion. I mean, that's roughly the
• 11:36
  GDP of a mid -sized nation trading hands over
• 11:38
  media assets. It's $18 billion more in cash and
• 11:42
  29 % more value than the Netflix offer. That's
• 11:44
  almost impossible for a board to ignore. Especially
• 11:47
  since the financing is bulletproof. The Paramount
• 11:49
  offer is fully backstopped and personally guaranteed
• 11:51
  by Larry Ellison. Whose family trust holds over
• 11:54
  $250 billion in Oracle stock. That's certainty.
• 11:56
  And they've pulled together a coalition of other
• 11:58
  heavy hitters. capital, Middle Eastern sovereign
• 12:01
  wealth funds, Jared Kushner's affinity partners.
• 12:04
  They're not presenting a hypothetical bid. They're
• 12:06
  presenting a guaranteed one. And the arguments
• 12:09
  they're making are fascinating. Paramount says
• 12:11
  their bid is pro -Hollywood, pro -consumer and
• 12:15
  pro -competition. Right. They say the combined
• 12:18
  company would target $6 billion in cost savings
• 12:21
  and instantly create a new streaming giant with
• 12:24
  207 million subscribers. And at the same time,
• 12:27
  they're attacking the Netflix deal. arguing it
• 12:30
  would solidify streaming domination and be harmful
• 12:34
  to the film and TV industry. Specifically pointing
• 12:36
  to the threat it poses to theatrical releases.
• 12:39
  Meanwhile, the WBD board is just stuck in the
• 12:41
  middle. They're advising shareholders to take
• 12:43
  no action for now. They only have 10 business
• 12:45
  days to issue a formal recommendation on the
• 12:48
  Paramount offer. So what does all this mean for
• 12:50
  you, the consumer? This fight really determines
• 12:53
  the shape of the media landscape. If Netflix
• 12:55
  wins, prepare for massive consolidation. fewer
• 12:59
  places to sell big content, and probably higher
• 13:01
  subscription prices down the road. And if Paramount
• 13:03
  wins, it creates a new supergiant designed specifically
• 13:07
  to challenge Netflix's dominance. Theoretically,
• 13:10
  a more competitive environment, which could mean
• 13:12
  more choice for you and better deals for creators.
• 13:15
  It is a high stakes battle between consolidation
• 13:18
  and competition. It really is. It's a cliffhanger.
• 13:21
  Before we wrap up today, we do need to thank
• 13:23
  our second sponsor, Scony. If you are looking
• 13:27
  for the absolute premium gift for the golfer
• 13:29
  in your life, Sconee is the premium golf shoe
• 13:32
  store. They have exceptional style and performance.
• 13:34
  It'll elevate any game. And just like with Old
• 13:36
  Glory, you can get 15 % off your entire purchase
• 13:39
  right now by using our promo code. Just use SNARFL.
• 13:43
  That's S -N -A -R -F -U -L at checkout. Head
• 13:45
  over to sconee .com and use the code SNARFL to
• 13:48
  snag those savings on the perfect holiday gift.
• 13:51
  Alright, so to quickly recap the three stories
• 13:53
  that defined our day. First, Minimax M2 .1 is
• 13:56
  radically changing the AI landscape. It's making
• 13:59
  high -end multi -language coding accessible and
• 14:01
  enabling these autonomous digital employees.
• 14:04
  Second, the global autonomous driving race is
• 14:07
  accelerating, especially in China, and it's driven
• 14:10
  by a fierce strategic competition over chip hardware
• 14:13
  and the push for Level 3. And finally, the future
• 14:16
  of Warner Bros. Discovery and maybe the entire
• 14:19
  streaming industry is being decided in a hostile
• 14:23
  $108 billion merger battle. And the thread connecting
• 14:27
  all these is technological power. Who cobes it,
• 14:31
  who builds the hardware for it, and who controls
• 14:33
  how it's distributed, that's who wins the next
• 14:35
  decade. It really makes you wonder. If that paramount
• 14:38
  bid succeeds and the new company cuts $6 billion
• 14:41
  in costs, How many of those jobs will be replaced
• 14:44
  by the very digital employees we just talked
• 14:46
  about? A provocative thought to end the year
• 14:48
  on? We're here to help at Snarful, so reach out
• 14:50
  on our website if you have any more questions.
• 14:55
  That's a wrap on today's top stories. Remember,
• 14:59
  tech doesn't have to be complicated and we're
• 15:01
  here to help connect the dots. If you've got
• 15:03
  questions or want to learn more, visit us at
• 15:06
  snarful .com. Thanks for listening and we'll
• 15:09
  catch you next time.