Latest / Connecting the Dots / AI Shifts from Hype to Business as Global Competition Heats Up

Transcript

AI Shifts from Hype to Business as Global Competition Heats Up · 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 and welcome to Connecting
• 0:17
  the Dots. I'm Alex. And I'm Morgan. Broadcasting
• 0:19
  to you live from Smarshall Solutions Group right
• 0:22
  here in Sacramento, California. It is Monday,
• 0:24
  March 30th, 2026, and we have got a really fascinating
• 0:29
  analysis lined up for you today. We really do.
• 0:32
  But, you know, before we get into the tech breakdown,
• 0:34
  let's take a quick look at what's going on out
• 0:35
  there this morning. Looking out the window here
• 0:37
  in Sacramento. It is just, it's a beautiful Monday.
• 0:40
  Oh, gorgeous. Yeah, we're looking at partly sunny
• 0:43
  skies, a high of around 81 degrees today, and
• 0:46
  this perfectly light little breeze at about two
• 0:48
  to seven miles per hour. Honestly, it is the
• 0:50
  perfect morning to grab your coffee and take
• 0:52
  a walk. And if you are checking your portfolio
• 0:56
  on that walk, well, Wall Street is giving us
• 0:57
  a bit of a mixed bag to start the week. The Dow
• 1:00
  is showing some optimism up about 77 .73 points,
• 1:04
  while the S &P 500 dipped slightly down 23 .04
• 1:08
  points. Crypto, on the other hand, is having
• 1:10
  a phenomenal morning. Bitcoin is flexing hard
• 1:13
  today, surging over $1 ,500 to close right around
• 1:16
  $67 ,478. That is a very strong start for the
• 1:21
  crypto markets. Now, on the local sports front,
• 1:25
  we do have to offer some quick condolences to
• 1:27
  the Sacramento Kings fans out there. Yeah, that
• 1:29
  was rough. A really tough loss last night against
• 1:31
  the Brooklyn Nuts, 99 to 116. Ouch. I mean, though,
• 1:36
  to be fair. Okay, God, you put up a really solid
• 1:39
  18 points off the bench. He did, yeah. It just,
• 1:42
  you know, it wasn't enough to stop the Nets from
• 1:44
  snapping their 10 -game losing streak. But honestly,
• 1:47
  the sports story everyone is actually talking
• 1:48
  about this weekend wasn't basketball. No. It
• 1:51
  was the MLB opening night broadcast on Netflix.
• 1:54
  Oh, man, talk about a spectacle that just completely
• 1:57
  missed the fundamentals. Completely. I mean,
• 1:59
  Netflix obviously wanted to bring in... a new
• 2:02
  younger audience, and they threw an absolute
• 2:05
  fortune at the production. But traditional baseball
• 2:07
  fans were left completely bewildered. Right.
• 2:09
  So the over -the -cop drone smoke, like they
• 2:11
  had red, white, and blue smoke pumping everywhere.
• 2:14
  They had dancers performing on top of New York
• 2:16
  City -style taxi cabs right on the field. The
• 2:19
  Giants literally came out of San Francisco trolley
• 2:21
  cars. It was wild. They were trying to turn a
• 2:24
  traditional MLB game into like a Savannah bananas
• 2:27
  routine. Exactly. And well, while that level
• 2:30
  of entertainment is fun in a vacuum, the actual
• 2:33
  broadcast failed at its core job. I mean, they
• 2:35
  had a vanishing score bug that made it almost
• 2:38
  impossible to read the score. Yes. And the camera
• 2:40
  cuts kept missing the in -game action. Yeah.
• 2:43
  The Yankees did end up beating the Giants seven
• 2:45
  and arrow, but the entire production just felt
• 2:47
  like an identity crisis. And honestly, That theme,
• 2:51
  you know, throwing massive amounts of capital
• 2:52
  at a flashy spectacle while completely failing
• 2:56
  to deliver fundamental everyday utility is the
• 3:00
  exact perfect transition into our main technology
• 3:03
  focus for today. It really is. The parallels
• 3:04
  are actually uncanny. But before we hit the news,
• 3:08
  let's talk about fueling up for the day. Because
• 3:10
  if you want to tackle this Monday with focus,
• 3:12
  you need the right kind of energy. Exactly. And
• 3:15
  that is where strong coffee comes in. It is a
• 3:18
  premium instant coffee that tastes exactly like
• 3:20
  a rich traditional brew, but it is doing so much
• 3:23
  more for your body. So much more. It is packed
• 3:26
  with 15 grams of protein, MCTs for brain fuel
• 3:30
  and adaptogens like ashwagandha. And the ashwagandha
• 3:33
  is really the game changer in that blend. Oh,
• 3:35
  absolutely. It regulates your cortisol, which
• 3:37
  is what gives you that clean, sustained energy.
• 3:39
  You get all the focus of a great cup of coffee
• 3:42
  without the jitters or, you know, that brutal
• 3:44
  mid afternoon crash. Yeah, it is literally just
• 3:47
  scoop, mix and conquer your morning. So do yourself
• 3:50
  a favor. Use promo code SNARFL. for 20 % off
• 3:53
  your entire purchase at strongcoffeecompany .com.
• 3:56
  That's strongcoffeecompany .com promo code SNARFL
• 3:59
  for 20 % off. All right, let's get into our feature
• 4:03
  segment. It is time for the Morningstar. The
• 4:06
  SNRFL Tech article report. Exactly. We pull the
• 4:09
  most critical technology stories of the day.
• 4:11
  And today, all three of our top stories point
• 4:13
  to a massive structural reality check in the
• 4:17
  AI industry. Yeah, the honeymoon phase is definitely
• 4:19
  over. It really is. The market is pivoting hard
• 4:21
  from this billion dollar hype to practical everyday
• 4:26
  utility. And our first story is an absolute shocker.
• 4:30
  OpenAI is officially shutting down its highly
• 4:33
  hyped text to video generator, Sora, just six
• 4:36
  months after it launched. Just six months. I
• 4:38
  mean, it went from being hailed on magazine covers
• 4:40
  as the technology that would single -handedly
• 4:43
  destroy and rebuild Hollywood to being quietly
• 4:46
  taken off life support. And it is not a gentle
• 4:48
  wind down either. They are walking away from
• 4:50
  a massive one billion dollar partnership with
• 4:53
  Disney. Right, the one that included licensing
• 4:55
  those iconic Disney characters to be generated
• 4:58
  inside the Sora engine. Yes. So OpenAI is ripping
• 5:01
  its computing power away from video generation
• 5:04
  and redirecting it entirely toward enterprise
• 5:07
  productivity products. all ahead of a planned
• 5:09
  IPO, which is a huge shift. But look, why? They
• 5:12
  had a billion dollar Disney deal sitting on the
• 5:14
  table. Why walk away from that kind of money?
• 5:16
  Well, to understand the sudden collapse, we have
• 5:18
  to look at the structural economics of the technology,
• 5:20
  specifically the difference between training
• 5:22
  a model and running inference. When Sora debuted,
• 5:24
  the demos were mind blowing. But the compute
• 5:27
  cost to generate high -definition video from
• 5:30
  scratch is just astronomical. Right, because
• 5:32
  it's not searching a database for a video of
• 5:34
  a dog running on a beach. It is hallucinating
• 5:37
  every single pixel from thin air, right? Exactly.
• 5:40
  It is using diffusion models to calculate the
• 5:42
  physics, the lighting, and the pixel placement
• 5:45
  for 24 to 30 frames every single second. Wow.
• 5:48
  So when a user asks ChatGPT to rate an email,
• 5:51
  that text generation costs OpenAI fractions of
• 5:54
  a cent in server compute. Sure. But when a user
• 5:57
  asks Sora for a 10 second video, it burns massive
• 6:00
  amounts of server time. The inference costs are
• 6:03
  so high that you fundamentally cannot offer a
• 6:05
  $20 a month consumer subscription without losing
• 6:07
  money every time someone hits the generate button.
• 6:11
  So it brings us back to that concept car analogy,
• 6:13
  like imagine an automaker spends a billion dollars
• 6:16
  designing a hyper futuristic concept car. It
• 6:19
  looks incredible, but it gets one mile to the
• 6:22
  gallon and costs $10 ,000 to start the engine.
• 6:24
  Exactly. It's an astonishing engineering flex,
• 6:27
  but absolutely no one can afford to drive it
• 6:29
  to the grocery store. And it's a massive liability
• 6:32
  for an IPO. Oh, absolutely. If you are preparing
• 6:34
  for an initial public offering, Wall Street doesn't
• 6:37
  care about your cool engineering flex. They care
• 6:40
  about your gross margins. Right. But the compute
• 6:43
  cost is only half the equation here. The legal
• 6:46
  reality of generative video is a massive exposure.
• 6:49
  Because of the training data. Yes. The Disney
• 6:52
  partnership covered specific licensed characters,
• 6:55
  but what about the millions of hours of copyrighted
• 6:58
  movie footage that was inevitably scraped to
• 7:01
  train the foundational model in the first place?
• 7:03
  Right, because text models have somewhat survived
• 7:06
  the gray areas of Fair You so far, but video
• 7:09
  feels entirely different. The legal exposure
• 7:12
  is orders of magnitude higher. I mean, ByteDance,
• 7:15
  the parent company of TikTok, recently delayed
• 7:18
  the global launch of their competing video models
• 7:20
  for the exact same reason. The technology works
• 7:23
  in a vacuum, but the commercial and legal frameworks
• 7:25
  to support it simply do not exist. You cannot,
• 7:29
  well, consumer demand and legal immunity into
• 7:31
  existence just because the underlying math is
• 7:34
  impressive. Which begs a massive question. If
• 7:37
  Western tech giants are bleeding money on these
• 7:39
  flashy, compute -heavy god models, who is actually
• 7:43
  turning AI into a profitable business right now?
• 7:46
  Because someone has to be making money. The answer
• 7:48
  is sitting in China, and it is operating on a
• 7:50
  completely different philosophy. Let's look at
• 7:53
  the numbers from our second story. While Sora
• 7:55
  was shutting down, Kuishu Technology, the Chinese
• 7:58
  short video platform behind Kuai, published their
• 8:01
  earnings. And they are impressive. Very. Their
• 8:04
  video generation platform, Kling AI, just posted
• 8:07
  an annualized revenue run rate of $300 million.
• 8:11
  In the fourth quarter of 2025 alone, they pulled
• 8:14
  in $47 million. Wow! And management projects
• 8:18
  at revenue will more than double in 2026. So
• 8:21
  how are they printing money while OpenAI is shutting
• 8:23
  down? This comes down to the fundamental difference
• 8:26
  between the model layer and the application layer.
• 8:28
  Western discourse is entirely obsessed with the
• 8:31
  model layer. The god models. Right. It's an arms
• 8:33
  race of who can build the biggest foundational
• 8:35
  model with the most trillions of parameters.
• 8:38
  It is an ego -driven quest to build an omnipotent
• 8:40
  digital brain. Like a Swiss army knife that can
• 8:43
  write a novel, code a Python script, and direct
• 8:46
  a cinematic movie all from one blank text box.
• 8:49
  Yes. But consumers rarely need a Swiss army knife.
• 8:52
  Most of the time, working professionals just
• 8:55
  need a really good screwdriver. Right. And that
• 8:57
  is what the application layer provides. Companies
• 8:59
  in the application layer treat AI not as the
• 9:02
  final product, but as an invisible engine designed
• 9:05
  to solve highly granular, specific workflow problems.
• 9:09
  Give us an example of how that works in practice.
• 9:11
  How does an application layer company actually
• 9:13
  deploy this? Look at a company like Metoo. They
• 9:16
  currently have five different apps in the global
• 9:19
  mobile top 50, primarily serving overseas users.
• 9:22
  Five apps. Five. And they are not selling a chat
• 9:25
  interface, they are selling specific solutions.
• 9:27
  Yeah. E -commerce product design, quick video
• 9:29
  touch -ups, targeted image editing. OK, so highly
• 9:32
  focused. Highly focused. They take smaller, highly
• 9:35
  efficient AI models and fine tune them for one
• 9:38
  specific task. Then they wrap that AI in an intuitive
• 9:43
  user interface. the user doesn't have to learn
• 9:45
  prompt engineering, they just click a button
• 9:47
  that says remove background and add studio lighting.
• 9:50
  So wait, are Western tech giants blinding themselves
• 9:53
  by trying to build one AI that can do everything
• 9:56
  for everyone, while these international developers
• 9:58
  are just building simple functional tools that
• 10:01
  working professionals and businesses actually
• 10:03
  want to pay for right now? That is the exact
• 10:05
  blind spot. The money isn't in the raw foundational
• 10:08
  intelligence. The money is in the interface.
• 10:10
  Wow. The application layer companies compete
• 10:13
  on design, functionality, and pricing. They aren't
• 10:16
  burning billions trying to make the model smarter.
• 10:19
  They are spending their capital making the user
• 10:21
  experience frictionless. And as Kling AI's numbers
• 10:24
  prove, that strategy is far more durable and
• 10:27
  profitable. It is. But you know, one major Western
• 10:30
  player has clearly been watching this exact trend
• 10:32
  and they're aggressively pivoting their entire
• 10:34
  ecosystem to capture that application layer utility.
• 10:38
  Which bridges us perfectly to our third story
• 10:40
  today. Let's talk about the device that is likely
• 10:42
  sitting in your pocket or on your desk right
• 10:44
  now. Apple has been watching this and they are
• 10:47
  making their move. And it is a total philosophical
• 10:50
  shift for them. Here is the news. Apple is releasing
• 10:53
  two new iPhone apps later this year. First, coming
• 10:57
  up shortly on April 14th, is a new Apple Business
• 11:01
  app. Right, which is replacing Apple Business
• 11:03
  Essentials, Apple Business Manager, and Apple
• 11:05
  Business Connect. Exactly. It acts as a centralized
• 11:08
  hub for employees to install work apps, view
• 11:11
  colleague directories, and request IT support.
• 11:14
  But the much bigger earthquake is what is coming
• 11:16
  with iOS 27, iPadOS 27, and Mac OS 27 later this
• 11:21
  year. The brand new standalone Siri app. Yes.
• 11:25
  Siri is getting a total overhaul with full chatbot
• 11:28
  functionality. We are finally witnessing the
• 11:30
  Siri rebirth the tech world has been demanding
• 11:32
  for a decade. Truly. Up until now, Siri has been
• 11:35
  strictly transactional. You ask it to set a 10
• 11:37
  -minute timer or check the weather, and it executes
• 11:39
  a rigid command. But it lacked reasoning. With
• 11:42
  iOS 27, Siri fundamentally transitions into a
• 11:46
  system -level chatbot, operating with the fluidity
• 11:49
  of chat GPT or Claude. It will support both rich
• 11:52
  text and voice conversations, and retain deep
• 11:55
  context from past interactions. With all the
• 11:58
  necessary privacy safeguards, of course. But
• 12:00
  it goes beyond just being a smart voice assistant
• 12:03
  now, right? It's deeply woven into the operating
• 12:05
  system itself. Completely. The biggest change
• 12:08
  is that it replaces Spotlight Search. Siri will
• 12:11
  be able to analyze on -screen context, analyze
• 12:14
  files you upload to it, and take direct action
• 12:16
  within core apps like mail and messages. Okay,
• 12:19
  so just to make sure I understand the mechanics
• 12:21
  of this, I could be looking at an email from
• 12:23
  a friend about a recent trip, and I could tell
• 12:25
  Siri, find the group photo from that trip, crop
• 12:28
  out the background, and reply to this email with
• 12:30
  it. Exactly. And Siri handles the entire workflow
• 12:33
  across three different apps. Yes, that is the
• 12:35
  exact application layer utility we were just
• 12:38
  talking about. Apple isn't giving you a blank
• 12:40
  chat box. They're using AI as an invisible routing
• 12:43
  engine to execute specific tasks. But here is
• 12:47
  the massive twist in the story, and this is what
• 12:50
  has the tech world reeling today. Apple isn't
• 12:53
  building the engine for this on their own? No,
• 12:55
  they are not. They have entered into a multi
• 12:57
  -year deal to power the new Siri using Google's
• 13:00
  Gemini AI models running remotely on Google's
• 13:03
  cloud servers. This is a stunning admission from
• 13:06
  a company that has historically built its entire
• 13:09
  brand around complete vertical integration. I
• 13:12
  mean, Apple loves to control the hardware, the
• 13:14
  software, and the silicon. Right. It's like Apple
• 13:17
  finally looked in the mirror and realized they
• 13:19
  couldn't build a hypercar engine fast enough
• 13:21
  to compete. So what do they do? They buy a top
• 13:25
  tier engine from Google, drop it into their beautifully
• 13:27
  designed Apple chassis, and hand the keys to
• 13:30
  their users. That is a perfect way to describe
• 13:33
  it. It looks like a concession on the surface,
• 13:35
  but strategically, it is brilliant. Really? Yeah.
• 13:37
  Even though they're outsourcing their flagship
• 13:39
  feature. Yes, because you have to look at the
• 13:41
  physics of edge computing versus cloud computing.
• 13:44
  Apple's on -device, silicon in there. The neural
• 13:47
  engines inside the iPhones and Macs are fantastic
• 13:50
  for low -latency, privacy -focused tasks like
• 13:53
  facial recognition or predictive text. Exactly.
• 13:57
  But running a frontier language model with hundreds
• 13:59
  of billions of parameters requires massive amounts
• 14:02
  of RAM, power, and specialized cooling. You cannot
• 14:05
  physically put that in a smartphone. Right. You
• 14:08
  need warehouses filled with specialized servers.
• 14:10
  You need Google's cloud infrastructure. Apple
• 14:13
  evaluated their internal models and realized
• 14:15
  they were lagging behind Google's Gemini. Rather
• 14:18
  than release an inferior product and lose the
• 14:21
  critical application race, they made a pragmatic
• 14:23
  choice. They are leveraging Google's servers
• 14:26
  for the heavy reasoning. Right. The AI itself
• 14:29
  isn't the destination. It is just the underlying
• 14:31
  infrastructure. Apple controls the power outlet,
• 14:34
  the iPhone, and the user never leaves the Apple
• 14:37
  interface. Wow. It is a fascinating evolution.
• 14:40
  We go from Sora's spectacular, expensive failure
• 14:43
  to Kling's quiet, targeted profitability and
• 14:46
  finally to Apple's strategic masterclass in integrating
• 14:49
  the infrastructure. The industry is rapidly growing
• 14:52
  up. It really is. Now before we wrap up today's
• 14:55
  breakdown and leave you with our final thought,
• 14:56
  we want to mention our second sponsor, Old Glory.
• 14:59
  Yeah, if you have a techie friend or family member
• 15:02
  who also loves pop culture, music or sports,
• 15:05
  Old Glory is the iconic fan merch store. absolutely
• 15:09
  need to check out. It really is the perfect place
• 15:11
  to grab a gift. They have an incredible selection
• 15:13
  of apparel and accessories that let you wear
• 15:15
  your fandom on your sleeve. The depth of their
• 15:17
  inventory is wild. You can go to their site looking
• 15:19
  for a gift for someone else and you inevitably
• 15:22
  end up buying three vintage band tees for yourself.
• 15:25
  It is a dangerous website for the wallet, for
• 15:27
  sure, but we've got a hookup for you. Use promo
• 15:30
  code SNARFUL at checkout. You will get 15 % off
• 15:33
  your entire purchase at OldGlory .com. That is
• 15:36
  oldglory .com promo code SNARFL for 15 % off.
• 15:40
  All right. So let's pull all of these threads
• 15:41
  together. What we are witnessing is the great
• 15:44
  AI normalization. Sora proved to the market that
• 15:48
  possessing undeniably cool technology without
• 15:50
  a sustainable business model just turns you into
• 15:53
  a very expensive toy. Meanwhile, the massive
• 15:56
  revenue numbers from Kling AI showed us that
• 15:58
  the real financial momentum is in the application
• 16:00
  layer. It's in solving highly specific everyday
• 16:03
  problems for working professionals. And Apple's
• 16:07
  massive Siri overhaul with iOS 27 cements the
• 16:11
  reality that the future of artificial intelligence
• 16:13
  isn't going to be found by visiting a standalone
• 16:15
  laboratory website. It is going to be seamlessly
• 16:19
  integrated into our daily workflows and text
• 16:21
  messages. But that level of deep integration,
• 16:24
  specifically with their new partnership, brings
• 16:26
  up a massive structural question. Absolutely.
• 16:30
  So we want to leave you with a provocative thought
• 16:31
  to chew on today. If Apple is willing to hand
• 16:34
  the brain of the iPhone over to Google just to
• 16:37
  win the AI application race, what does that mean
• 16:40
  for the data privacy standard Apple has spent
• 16:43
  a decade building? That is the million dollar
• 16:45
  question. It really is. You are definitely going
• 16:48
  to want to keep an eye on those terms of service
• 16:49
  when iOS 27 drops. It's going to be fascinating
• 16:52
  to see how they handle that data handoff. We're
• 16:55
  here to help at Snarful, so reach out on our
• 16:56
  website if you have any more questions. That's
• 17:02
  a wrap on today's top stories. Remember, tech
• 17:05
  doesn't have to be complicated, and we're here
• 17:07
  to help connect the dots. If you've got questions
• 17:10
  or want to learn more, visit us at snarful .com.
• 17:13
  Thanks for listening, and we'll catch you next
• 17:15
  time.