Latest / Connecting the Dots / Autonomous Coders, Billion-Dollar Robots, and Turbulence at xAI

Transcript

Autonomous Coders, Billion-Dollar Robots, and Turbulence at xAI · 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, Sacramento. And a massive
• 0:18
  hello to everyone tuning in from around the globe.
• 0:21
  You're listening to Connecting the Dots, the
• 0:23
  show where we try to make sense of the daily
• 0:24
  flood of tech news. I'm Alex. And I'm Morgan.
• 0:27
  It is Wednesday, February 11th, 2026. And, you
• 0:31
  know, looking at the Morningstar report on my
• 0:33
  desk, it feels like the world shifted a little
• 0:35
  bit while we were sleeping. It really does. We
• 0:37
  have a show today that, and I'm not exaggerating,
• 0:40
  might change how you think about your job. Yeah.
• 0:43
  And maybe even the moon. Maybe, especially the
• 0:46
  moon. But before we get into all that, let's
• 0:48
  take a look outside. If you're here in Sacramento
• 0:50
  with us, you know the drill. grab that umbrella
• 0:54
  it is wet very wet yeah showers pretty much all
• 0:57
  day with a high around 58 degrees it's all part
• 0:59
  of this big pacific storm system that's moving
• 1:01
  inland so rain for us but if you're out in the
• 1:04
  great basin or the rockies you are just getting
• 1:07
  buried in snow right now and a quick shout out
• 1:09
  to our listeners in alaska a huge low pressure
• 1:12
  system is hitting them hard so please stay warm
• 1:15
  and safe up there okay let's take a quick look
• 1:17
  at the markets The big board is flashing a little
• 1:21
  green this morning. The Dow is up about 100 points,
• 1:25
  so a gain of 0 .20%. The S &P 500 is basically
• 1:29
  flat, just kind of hovering. But crypto has taken
• 1:31
  a bit of a hit. Bitcoin's down about 2 .5 % this
• 1:34
  morning. Yeah, trading around $67 ,000. So a
• 1:37
  little bit of red if you're looking at the digital
• 1:39
  wallet today. A little discount maybe. Now look.
• 1:41
  With a news day like today, I need some serious
• 1:45
  focus, not just the usual caffeine jitters. I
• 1:48
  know exactly what you mean. Which is why I am
• 1:50
  genuinely excited about our new partner, Strong
• 1:53
  Coffee. I have been drinking this stuff all week
• 1:55
  getting ready for the show. It's premium instant
• 1:57
  coffee, but, and this is the key thing, it really
• 2:00
  doesn't taste like instant coffee. It tastes
• 2:02
  like a real brew. Exactly. And the kicker is
• 2:05
  what's inside. 15 grams of protein, MCTs, and
• 2:09
  adaptogens like ashwagandha. Oh, ashwagandha
• 2:12
  is a total game changer. It gives you that clean,
• 2:15
  sustained energy. But without that awful crash,
• 2:17
  you get it too in the afternoon. Right. If you
• 2:19
  want to try it out, head over to strongcoffeecompany
• 2:22
  .com. Use our promo code SNARFL. That's S -N
• 2:25
  -A -R -F -U -L. And you'll get 20 % off your
• 2:28
  entire purchase. Seriously good stuff. And we
• 2:31
  are going to need it today because our lead story
• 2:33
  from the Morningstar report is... Well, it's
• 2:36
  a big one. We're calling it the Great AI Leap.
• 2:38
  Let's just jump right in. For the past year,
• 2:41
  it feels like the common wisdom was, oh, AI is
• 2:43
  kind of plateauing. It's getting cheaper, but
• 2:46
  not that much smarter. We got comfortable. And
• 2:48
  on February 5th, that idea just died. It completely
• 2:52
  evaporated. We saw two massive models get released
• 2:55
  at the same time, OpenAI's GPT -5 .3 codex and
• 2:59
  Anthropix's Claude Opus 4 .6. And looking at
• 3:01
  the specs, This isn't just a small update. This
• 3:04
  feels like a totally different kind of software.
• 3:06
  It's a fundamental shift. Let's break down the
• 3:08
  players. First, you've got OpenAI's GPT 5 .3
• 3:10
  codecs. The main thing here is speed and memory.
• 3:13
  It's 25 % faster. OK. But the number that really
• 3:16
  matters for business is the context window. 400
• 3:19
  ,000 tokens. All right. Let's stop there. We
• 3:21
  hear tokens thrown around all the time. For someone
• 3:25
  who isn't a machine learning expert, what does
• 3:28
  400 ,000 tokens actually mean? That's a great
• 3:31
  question. So think of a token as about three
• 3:34
  quarters of a word. Four hundred thousand tokens
• 3:36
  is like being able to read and remember three
• 3:40
  hundred novels all at once. Three hundred novels.
• 3:42
  Yep. You can feed it your entire company's documentation
• 3:45
  or years of legal files. And it holds all of
• 3:48
  that in its working memory for the whole conversation.
• 3:51
  That is That's terrifyingly powerful. It's like
• 3:54
  an employee who memorized the entire company
• 3:56
  archive in five seconds. Exactly. And then you
• 3:59
  have Anthropix Claude Opus 4 .6. And they're
• 4:01
  pushing it even further. Even further. They have
• 4:04
  a beta with a 1 million token context window.
• 4:07
  But where Claude is really standing out is in
• 4:09
  deep reasoning. It's more of a thinker. It can
• 4:10
  do long term strategic planning. But there was
• 4:13
  one detail in the technical documents for GPP
• 4:15
  5 .3 that just, it stopped me cold. We have to
• 4:19
  talk about this self -made angle. I had a feeling
• 4:21
  you'd catch that. The report says, and I'm quoting
• 4:23
  here, that the model was instrumental in creating
• 4:26
  itself. It was used to debug its own training.
• 4:29
  This is it. This is the feedback loop that researchers
• 4:31
  have been talking about for decades. It's called
• 4:33
  recursive self -improvement. So you're using
• 4:36
  today's AI to help write the code for tomorrow's
• 4:40
  AI. Precisely. And because the AI works so fast,
• 4:43
  it can find bugs and optimize things way faster
• 4:46
  than any human team ever could. The intelligence
• 4:49
  is compounding. So the AI helped build a better
• 4:52
  AI. That's it. And OpenAI is basically telling
• 4:55
  us that loop has officially started. It reminds
• 4:58
  me of that Matt Schumer article you sent me.
• 5:00
  The analogy he used gave me chills. He compared
• 5:02
  right now. February 2026 to February 2020. The
• 5:06
  before times. Exactly. Remember that. The virus
• 5:08
  was in the news, but life felt normal. We were
• 5:11
  still going out. But the math, you know, the
• 5:13
  math said the world was about to change. He calls
• 5:15
  this the water rising moment. It's a perfect
• 5:17
  analogy. You're standing on the beach, the water's
• 5:19
  at your ankles, and you think you're fine. But
• 5:21
  there's a tsunami coming. And that tsunami is
• 5:24
  aimed squarely at the job market. And not just
• 5:27
  for coders, right? That's the big misconception.
• 5:29
  No. They started with code because that's the
• 5:31
  language they used to build more AI. But now
• 5:34
  that they've nailed high -level reasoning, they're
• 5:36
  moving on to everything else. Legal, finance,
• 5:40
  creative writing. The CEO of Anthropic, Dario
• 5:42
  Omaday, predicted that 50 % of entry -level white
• 5:45
  -collar jobs could be eliminated in one to five
• 5:48
  years. One to five years. And looking at the
• 5:50
  performance data from METR, an organization that
• 5:53
  benchmarks AI one year seems more likely than
• 5:57
  five. What are the numbers? What does the data
• 5:59
  show? Well, a year ago, AI could do tasks that
• 6:02
  took a human a few minutes. Now. It can complete
• 6:06
  complex tasks that take a human expert Five hours.
• 6:09
  Five hours of work done. Instantly. Instantly.
• 6:12
  And that capability is doubling every few months.
• 6:15
  Imagine hiring someone who does five hours of
• 6:17
  expert work in a second for pennies. That's the
• 6:19
  new reality. Yeah. But you highlighted a really
• 6:21
  important point in the notes. A divide that's
• 6:24
  opening up between users. Yeah. Schimmer calls
• 6:26
  it the flip phone problem. I love that. What
• 6:28
  does he mean? It's simple. If you're only using
• 6:30
  the free version of these tools, you're basically
• 6:32
  using a flip phone in a smartphone world. The
• 6:35
  free versions are older, less powerful models.
• 6:39
  So people paying 20 bucks a month have access
• 6:41
  to like a nuclear reactor while everything else
• 6:44
  is playing with sparklers. That's a great way
• 6:46
  to put it. And it creates a false sense of security.
• 6:50
  You think AI is just OK because the free version
• 6:53
  makes mistakes, but you're going to be blindsided
• 6:55
  by someone using the paid tool to do your job
• 6:58
  10 times faster. OK, so let's get practical.
• 7:00
  What's the move? How do our listeners get ahead
• 7:02
  of us? The advice from the experts is pretty
• 7:05
  straightforward. It's the one hour rule. The
• 7:07
  one hour rule. Spend one hour a day, every day,
• 7:10
  using the paid tools. But don't ask it for a
• 7:12
  poem. Give it your actual work. Give it the messy
• 7:16
  spreadsheet. Give it the boring contract you
• 7:18
  have to summarize. Exactly. Force it to do the
• 7:21
  parts of your job you hate. Be the person who
• 7:23
  says, I already ran the analysis, not the one
• 7:25
  who says, I need a few days. I wonder if I can
• 7:28
  get it to do my coffee run. That's the one thing
• 7:30
  I think is still safe. It doesn't have legs,
• 7:33
  Alex, so your job is safe. For now. Well, that's
• 7:36
  a perfect setup for our next story. We're moving
• 7:38
  from digital brains to physical bodies. That's
• 7:42
  right, from the cloud to the factory floor. This
• 7:45
  story comes out of Austin, Texas. A startup called
• 7:48
  Aptronic just raised a huge funding extension.
• 7:51
  How huge? $520 million. Wow. Which brings their
• 7:56
  total round to nearly a billion dollars. And
• 7:59
  the company's valued at over five billion. That's
• 8:02
  a lot of cash for a hardware company. What are
• 8:03
  they building? They're building Apollo. It's
• 8:05
  a humanoid robot. Okay, so when I hear humanoid
• 8:08
  robot, my mind goes straight to like... Scary
• 8:11
  sci -fi movies or those videos of robots falling
• 8:14
  downstairs right and up tronic is being very
• 8:17
  smart about that They're focused on embodied
• 8:19
  AI, but with human psychology in mind Apollo
• 8:22
  is designed to look friendly friendly How I level
• 8:24
  cameras soft edges a face that isn't just a blank
• 8:27
  slate? The goal is for it to work alongside people,
• 8:30
  not locked away in a cage. A collaborator, not
• 8:32
  just a machine. Exactly. And this isn't just
• 8:34
  a concept anymore. They have some major partners.
• 8:37
  Mercedes -Benz is already using Apollo on their
• 8:39
  production lines. Wait, really? They're on the
• 8:41
  line right now? Yes. Doing the dull, dirty, and
• 8:45
  dangerous jobs. You've also got Google DeepMind
• 8:49
  as a partner, plus investments from John Deere
• 8:52
  and AT &T. It's interesting to compare them to...
• 8:54
  Tesla's optimist it is Tesla is juggling cars
• 8:58
  batteries solar a million things Eptronic is
• 9:02
  laser focused on just the robot that might be
• 9:04
  their biggest advantage master the factory then
• 9:07
  move to retail Yeah, and our homes. Yeah, I wouldn't
• 9:10
  mind an Apollo doing my dishes We're getting
• 9:11
  closer, but you see the trend right you've got
• 9:14
  the brain with GPT And now you have the body
• 9:16
  with Apollo the pieces are all coming together
• 9:18
  speaking of pieces coming together Yeah, and
• 9:20
  speaking of big thinkers We need to leave Austin
• 9:23
  and go somewhere a little farther out. We're
• 9:25
  going to the moon. Literally. At an all -hands
• 9:28
  meeting for XAI, Elon Musk dropped a bombshell.
• 9:32
  He told his employees the company needs a factory.
• 9:34
  Not in Texas. Not in Nevada. On the moon. I had
• 9:38
  to double check this wasn't a joke. A factory
• 9:40
  on the moon. To do what? To build AI satellites
• 9:43
  and launch them with a giant catapult. A mass
• 9:46
  driver, technically. A mass drive. Why? Why a
• 9:49
  catapult on the moon? It actually comes down
• 9:51
  to basic physics. The moon has almost no gravity
• 9:54
  and no atmosphere, so you don't need a huge rocket
• 9:57
  to get into orbit. You can just build a long
• 10:00
  magnetic track and basically fling things into
• 10:03
  space. So he wants to turn the moon into a giant
• 10:06
  launch pad to get more computing power into orbit
• 10:09
  than anyone else. That's the plan. He said it's
• 10:11
  to secure the future of civilization. But while
• 10:14
  he's looking at the stars, things back on Earth
• 10:16
  for XAI are, uh... Messy. Messy is a good word
• 10:20
  for it. They've had a staff exodus, right? A
• 10:22
  big one. Two co -founders and a bunch of key
• 10:24
  employees just resigned. And then there are the
• 10:26
  legal issues. The really serious one. Yeah. The
• 10:29
  California attorney general sent a cease and
• 10:31
  desist over their chat bot, Grok, being used
• 10:34
  to generate non -consensual deep fake images
• 10:36
  of real people. And on top of that, environmental
• 10:40
  problems in Memphis. Yep. Their data center there
• 10:43
  is under fire for using gas -burning turbines,
• 10:46
  and locals are complaining about a rotten egg
• 10:48
  -like stench. Lovely. So a pristine factory on
• 10:52
  the moon, but a rotten egg smell in Memphis.
• 10:55
  It's quite the contrast. It's the classic Musk
• 10:58
  paradox. An unbelievably huge vision, but the
• 11:02
  execution on the ground is just pure chaos. And
• 11:05
  amid all that chaos, he announced SpaceX is acquiring
• 11:08
  XAI. Which... If you're planning on building
• 11:10
  a moon factory, I guess that makes a certain
• 11:12
  kind of sense. SpaceX is the only ride to get
• 11:14
  you there. It's just, it's a lot to process.
• 11:17
  Self -improving code, friendly robots building
• 11:20
  cars, and moon catapults. The future is definitely
• 11:23
  not boring. Not at all. All right, before we
• 11:25
  wrap this all up, I want to give a quick mention
• 11:27
  to our friends at SCONI. Oh yeah, love my SCONIs.
• 11:29
  If you're a golfer, you really need to check
• 11:31
  out SCONI .com. They make premium golf shoes
• 11:34
  that are just incredibly comfortable and look
• 11:36
  great. And they've got a great deal for our listeners.
• 11:38
  Use that same promo code SNARFL and you'll get
• 11:40
  15 % off your entire purchase. Okay, let's do
• 11:43
  a quick recap. Three things you need to remember
• 11:45
  from today. What are they? First, the ground
• 11:48
  has shifted under our feet with AI. GPT 5 .3
• 11:52
  and Claude 4 .6 are here. They're self -improving.
• 11:55
  And if you're a knowledge worker, it is time
• 11:57
  to adapt. Now. Don't be the person with the flip
• 12:00
  phone. Second, robots are getting real bodies.
• 12:03
  Aptronic has the money and the partners to put
• 12:05
  humanoid robots in Mercedes factories. This isn't
• 12:08
  science fiction anymore. And third, Elon Musk
• 12:11
  is aiming for the moon, literally, with a plan
• 12:14
  for a factory and a catapult, all while fighting
• 12:17
  regulators and staffing problems back on Earth.
• 12:20
  It is a brave new world. And here's a thought
• 12:22
  to leave you with. We talked about AI having
• 12:25
  reasoning. and robots being collaborators. As
• 12:28
  these lines keep blurring, the big question isn't
• 12:30
  just what can machines do anymore. It's what's
• 12:33
  left that only we can do. That is the question
• 12:35
  of the decade. Think on that one. Thank you so
• 12:38
  much for joining us on Connecting the Dots. We're
• 12:40
  here to help at SNARFL, so reach out on our website
• 12:42
  if you have any more questions. That's a wrap
• 12:47
  on today's top stories. Remember, tech doesn't
• 12:50
  have to be complicated, and we're here to help
• 12:52
  connect the dots. If you've got questions or
• 12:55
  want to learn more, visit us at snarful .com.
• 12:59
  Thanks for listening, and we'll catch you next
• 13:01
  time.