Latest / Today’s AI News / A Sneaker Brand Pivoted to AI Compute and Stock Jumped 600%, Snap Cuts 1,000 Jobs, Gemini Hits Mac

Transcript

A Sneaker Brand Pivoted to AI Compute and Stock Jumped 600%, Snap Cuts 1,000 Jobs, Gemini Hits Mac · Today’s AI News

• 0:00
  Imagine taking a cozy, sustainable wool sneaker,
• 0:03
  tearing it down to the soles, and stuffing it full of humming
• 0:06
  graphics processing units. It sounds completely absurd, but
• 0:10
  that is exactly what is happening on Wall Street right
• 0:13
  now. Welcome back to today's AI News,
• 0:16
  brought to you by Nynex Productions.
• 0:18
  Yeah, thanks for having me. It's, it's going to be a really
• 0:20
  wild 1 today. I'm your resident AI expert and
• 0:23
  I'm ready to get into it. And I'm your host.
• 0:26
  Look, the mission of today's Dee dive is, well, it's tailored
• 0:29
  specifically for you, the listener.
• 0:30
  We are cutting through all the noise of the latest AI
• 0:33
  develoments to get to what actually matters.
• 0:35
  Because there's just so much noise out there, right?
• 0:37
  Now you know exactly. So let me set the stage for the
• 0:40
  journey ahead. Today we are promising to unpack
• 0:44
  everything, and I mean everything, from a wild
• 0:47
  corporate AI pivots and massive workforce shifts to the battle
• 0:51
  for your desktop muscle memory. Oh, that desktop battle is
• 0:54
  absolutely fascinating. It really is, yeah.
• 0:56
  And then we'll be ending with some real world community AI
• 0:59
  workflows that you can, you know, use right now in your
• 1:01
  daily life. But I want to start with what
• 1:04
  has to be the wildest AI pivot of the year.
• 1:08
  I think I knew exactly where your going with this.
• 1:09
  Yeah. We have to talk about all birds
• 1:12
  because the trajectory here is just insane.
• 1:14
  So back in 2021, this was a company that hit a $4 billion
• 1:19
  peak at their IPO, 4 billion. Which is just a massive
• 1:22
  valuation for a shoe company. Right, Fast forward to this past
• 1:25
  March and they just, they sold all their brand assets to
• 1:28
  American Exchange Group for a mere $39 million.
• 1:31
  Wow. Talk about a drop.
• 1:33
  Exactly. They basically gutted it.
• 1:34
  But here's the kicker. Now they are closing a $50
• 1:38
  million financing deal to become something called New Bird AI.
• 1:44
  Wait, new bird AI? Yeah, and they are pivoting to
• 1:47
  become a GPU as a service rental shop.
• 1:49
  That is, I mean, that's just bonkers going from sustainable
• 1:52
  wool shoes to renting out graphics processing units.
• 1:55
  And the market reaction was immediate.
• 1:58
  Shares of their ticker Be Our Dick just exploded.
• 2:02
  They went from $3 to over $20. It's it's like a 600% pop.
• 2:06
  For a market cap that had doodled to what, 22 million?
• 2:10
  Dollars exactly 22 million. And next month, the shareholders
• 2:13
  are actually going to vote to strip their public benefit
• 2:16
  status. So they are officially killing
• 2:19
  the sustainable footwear mission entirely.
• 2:23
  But OK, let's unpack this. Let's do it.
• 2:25
  Because to me, this feels like a dying video rental store
• 2:28
  suddenly declaring their cryptocurrency mine back in
• 2:31
  2017. Is this just, you know, pure
• 2:33
  corporate desperation? Well, what's fascinating here is
• 2:36
  how this actually connects to the broader compute crunch we're
• 2:39
  seeing. I mean, sure, it sounds like
• 2:41
  desperation, but you have to look at the underlying
• 2:42
  mechanics. Right, because everyone needs GP
• 2:44
  us right now. Exactly, we hear CE OS claim all
• 2:47
  the time that every company will become an AI company, but
• 2:50
  gutting a retail business for parts just to rent out AI
• 2:53
  compute under long term contracts?
• 2:55
  That isn't what anyone expected. No, nobody had that on their
• 2:58
  bingo card. Right.
• 3:00
  But there is such a severe shortage of GP US right now,
• 3:03
  it's basically providing the perfect pitch cover for these
• 3:07
  dying tickers. They have the public shell, they
• 3:09
  get the capital injection and they just pivot to
• 3:12
  infrastructure. That makes sense.
• 3:14
  It's wild, but it makes sense. And Speaking of Wall Street
• 3:17
  rewarding these aggressive moves, that actually brings us
• 3:19
  to our next story. Yeah, because Wall Street is
• 3:22
  definitely rewarding 2 specific AI moves right now.
• 3:25
  You've got these wholesale pivots like all birds and then
• 3:28
  you have AI driven efficiency layoffs.
• 3:31
  Which is exactly what is happening over at Snap.
• 3:33
  They just announced they are cutting 1000 jobs. 1000 Jobs.
• 3:38
  Which represents roughly 16% of their entire workforce.
• 3:41
  And the craziest part is that CEO Evan Spiegel didn't didn't
• 3:44
  blame this on inflation or shareholder pressure.
• 3:47
  He explicitly blamed it on AI efficiency.
• 3:49
  That's a. Huge shift in the narrative.
• 3:51
  Usually CE OS try to hide behind macroeconomic factors.
• 3:54
  Exactly. But Spiegel is saying they're
• 3:56
  literally replacing traditional engineering teams with these
• 3:59
  small, what he calls AI, augmented pods.
• 4:02
  And the numbers he gave are staggering.
• 4:05
  He says AI is now writing 65% of all new code at snap 65% and
• 4:12
  it's fielding over 1,000,000 monthly internal queries.
• 4:15
  Which is why SNAP stock rose, what, 7 to 9% on this news?
• 4:19
  Yeah, roughly 9%, even though they're still down 30% year to
• 4:22
  date. But they are aiming for $500
• 4:25
  million in annual savings by the end of 2026.
• 4:30
  So what does this all mean? For the human worker, you mean?
• 4:33
  Yeah, exactly. I mean, especially when you
• 4:35
  remember that Block cut 4000 jobs, which was 40% of their
• 4:38
  staff back in February. We've seen over 70,000 tech jobs
• 4:43
  erased this year alone. Well, if we connect this to the
• 4:46
  bigger picture, there's this growing, really uncomfortable
• 4:49
  disconnect. You have markets cheering on one
• 4:52
  side and workers terrified on the other.
• 4:54
  It's it's a very stark contrast. It is because Spiegel's framing
• 4:58
  here the idea that AI reduces repetitive work and increases
• 5:01
  velocity. This is the new corporate
• 5:03
  blueprint. Right, they aren't just trimming
• 5:04
  the fat anymore. Exactly.
• 5:06
  It's fundamentally changing how tech companies scale.
• 5:09
  You don't need a massive team of junior doves if a small pod
• 5:13
  using AI can output the exact same amount of code.
• 5:16
  Which is a perfect segue, really, because while AI is
• 5:19
  fundamentally reshaping enterprise workforces, it is
• 5:22
  simultaneously fighting this massive battle for prime real
• 5:25
  estate on your personal devices. Yes, the desktop wars, yeah.
• 5:30
  Google just rolled out a native Mac desktop app for its Gemini
• 5:34
  Assistant, and you literally just invoke it by pressing
• 5:37
  Option plus Space. Just option space.
• 5:39
  That's prime keyboard real estate, right?
• 5:41
  There it is, and the features are pretty wild.
• 5:43
  It's got full screen sharing capabilities, it can access your
• 5:46
  Google Drive and Google Photos directly, plus it has Nano
• 5:49
  Banana for image generation and VO for video generation.
• 5:52
  Built right in which? Is a huge step up from how they
• 5:54
  handled Windows. Right.
• 5:55
  Oh totally. The Windows app release
• 5:57
  basically just jammed Gemini and Google Lens into a search bar,
• 6:01
  and it's shipped English only, but this Mac App is a full
• 6:04
  global rollout. They are definitely taking the
• 6:06
  Mac user base very seriously. But here's where it gets really
• 6:10
  interesting. Gemini is showing up essentially
• 6:14
  a full year late compared to rivals like Claude and ChatGPT.
• 6:19
  That's true. They are late to the native
• 6:21
  desktop party. And Gemini is still just a chat
• 6:24
  first assistant, right? You type, it talks back.
• 6:27
  But Claude and ChatGPT can directly execute tasks on a
• 6:31
  user's machine now, so isn't Google basically bringing a
• 6:34
  knife to a gunfight here? That's a great question, and it
• 6:37
  comes down to understanding agentic abilities.
• 6:39
  You're right that Gemini currently lags in executing
• 6:42
  those complex autonomous tasks. Yeah, I can't click around your
• 6:45
  screen for you yet. Exactly.
• 6:47
  But the desktop battle isn't entirely about who has the
• 6:50
  smartest agent right this second.
• 6:52
  It's entirely about muscle memory.
• 6:53
  It's about becoming the daily driver for the user.
• 6:56
  Right, because if I'm hitting space every time I need
• 6:59
  something. You're in Google's ecosystem.
• 7:01
  Google has massive distribution power.
• 7:03
  If they capture that muscle memory now, they can make up for
• 7:06
  lost time and ease of use deficits incredibly fast.
• 7:09
  I just push an update later that makes it smarter.
• 7:11
  Exactly. They get you hooked on the
• 7:12
  shortcut first. OK, that makes sense.
• 7:15
  But this raises an important question.
• 7:18
  What does true agentic AI actually look like right now for
• 7:21
  the user who wants those autonomous tasks today?
• 7:25
  It's the perfect bridge, really, because you don't have to wait
• 7:28
  for Google to figure. It out right, you can actually
• 7:31
  use true AI agents today inside Notion.
• 7:35
  They just rolled out these prebuilt clod powered AI agents
• 7:38
  and the stand out one is called the Business Workspace Auditor.
• 7:41
  I saw this, it's pretty incredible how seamless it is.
• 7:44
  Almost too easy. So for the listener, here's the
• 7:47
  step by step breakdown. You literally just install the
• 7:49
  template from Notion AI. You give the agent access to
• 7:52
  specific pages or databases. Right, just basic permission.
• 7:56
  Then you just at mention the databases and you simply type
• 7:59
  audit this page. That's it.
• 8:00
  No complex prompting needed. None.
• 8:03
  And the output is this highly formatted issue report.
• 8:06
  It comes complete with severity levels and recommended fixes.
• 8:10
  It literally looks like a report from a highly paid consultant.
• 8:13
  That's amazing. And it gets Wilder.
• 8:15
  If you give the agent edit permissions, it can actually
• 8:18
  make the fixes itself. Wait, it just goes in and edits
• 8:21
  the database autonomously? Exactly, it spots the missing
• 8:24
  tag or the broken link and it just fixes it.
• 8:27
  And there are other pro tips here too.
• 8:28
  You can use their business process Audit to diagnose
• 8:31
  bottlenecks in your workflows, or you can use the Task Triagger
• 8:35
  for a complete brain dump after a meeting.
• 8:37
  Which is so valuable because this really bridges the gap
• 8:40
  between chat and action. Right.
• 8:42
  It's not just talking, it's doing.
• 8:44
  Exactly, and it highlights the value of native integration.
• 8:48
  When an AI lives directly where your data lives, like inside
• 8:51
  Notion, the friction just disappears.
• 8:55
  It makes these agents incredibly potent for everyday efficiency.
• 8:58
  You don't have to copy paste things back and forth.
• 9:00
  Precisely. The data and the agent are in
• 9:03
  the same room, so to speak. Which naturally pivots us right
• 9:07
  into our next segment, because the evolution of specific models
• 9:10
  like clawed inside Notion, leads perfectly into our quick hits.
• 9:14
  Let's cover everything else happening in the AI landscape
• 9:16
  today. Let's run through them.
• 9:17
  There's a lot. OK, Trending tools first, we
• 9:20
  already mentioned Gemini for Mac and that whole desktop app roll
• 9:23
  out. Right, the option space muscle
• 9:25
  memory grab. Exactly.
• 9:26
  Next up is Lyra 2 Point O. This is Nvidia's new mind
• 9:31
  bending tech. It basically turns text and
• 9:33
  camera paths into fully explorable 3D scenes.
• 9:36
  Yeah. Lyra is fascinating.
• 9:37
  Generating actual 3D volume from text is a massive leap forward
• 9:41
  for spatial computing. It's wild.
• 9:43
  And we have Gemini 3.1 Flash TTS.
• 9:45
  This is Google's new speech model.
• 9:47
  It covers over 70 languages. But the cool part is it uses
• 9:50
  these audio tags to actually steer the tone, the pace, and
• 9:54
  even the accent. It's currently sitting at #2 on
• 9:56
  the Artificial analysis leaderboard, by the way.
• 9:59
  Very impressive latency. Wow #2 OK and finally for tools
• 10:03
  we have Ernie Image. This is Baidu's powerful new
• 10:06
  text to image model and they released it with open weights.
• 10:09
  Which is great for the developer community.
• 10:11
  Open weight models drive so much grassroots innovation.
• 10:13
  Absolutely. OK, moving on to the everything
• 10:16
  else is news. Adobe just debuted their Firefly
• 10:18
  AI assistant. This is a big push into Agenta
• 10:20
  creativity. It runs multi app workflows
• 10:22
  across Photoshop, Premiere, Lightroom, and Firefly.
• 10:26
  Right, so you aren't just prompting A1 app, the AI is
• 10:28
  orchestrating across the whole Adobe suite.
• 10:30
  Exactly next GBT 5.4 Pro just produced a mathematical proof
• 10:36
  for a 60 year old math problem. Oh, I read about this.
• 10:39
  Yeah, the mathematician Jared Lichtman said.
• 10:41
  It found a path that humans had completely overlooked.
• 10:44
  That's the real breakthrough. It's not just regurgitating
• 10:47
  textbook math, it's actually synthesizing new logical paths.
• 10:51
  It's crazy to wrap your head around.
• 10:53
  Yeah. And our last quick hit,
• 10:54
  Anthropic is switching their Claude Enterprise pricing.
• 10:58
  They're all going to charge businesses based on actual token
• 11:01
  consumption now. Which could cause some serious
• 11:04
  bill shock for a power user. Oh for sure, if you've been
• 11:07
  uploading massive PDFs all day, your finance department's going
• 11:09
  to have a heart attack next month.
• 11:11
  Yeah, people really need to watch their context windows now.
• 11:14
  But to synthesize all these hits briefly, whether it's solving 60
• 11:18
  year old math problems or generating 3D scenes, the
• 11:21
  overarching theme here is AI moving from passive generation
• 11:24
  to active problem solving. Active problem solving.
• 11:27
  I love that. And that actually bring us to
• 11:29
  our final segment, community AI Workflows, because we have to
• 11:33
  ask, how are everyday people actually applying this problem
• 11:37
  solving tech in their daily lives?
• 11:39
  Right, getting out of the enterprise space and into the
• 11:41
  living. Room exactly.
• 11:43
  So today we have a story from Kirsten AR in London, UK.
• 11:46
  So Kirsten had this massive pile of high ticket clothing that she
• 11:50
  wanted to sell secondhand. OK, sure.
• 11:52
  But she was completely daunted by the listing process because
• 11:56
  it is incredibly time consuming to measure, write description's,
• 11:59
  rice check, all of that. Oh, it's the worst.
• 12:01
  Total friction. Right O she built a custom
• 12:04
  system. She just takes photos of the
• 12:07
  item on her phone and she can put some basic keywords like the
• 12:10
  brand, the size and a rough price right into an air table
• 12:14
  form. OK, air table.
• 12:15
  I like it. And from there, plot takes over.
• 12:18
  It automatically generates highly optimized, compelling
• 12:21
  copy for every single text field.
• 12:23
  We're talking from the SEO headline down to the bulleted
• 12:26
  description. That saves so much typing.
• 12:28
  But it gets better. Claude even gauges the pricing
• 12:31
  based on existing listings on the platform and adjusts her
• 12:34
  price to make sure it's competitive.
• 12:36
  See, that is that is just brilliant, isn't it?
• 12:39
  It is the perfect embodiment of everything we've talked about in
• 12:42
  today's deep dive. She's taking those agentic
• 12:45
  concepts we discussed with Notion and Adobe and she's
• 12:48
  applying them to a highly practical, money making personal
• 12:51
  system. It really proves you don't need
• 12:54
  to be a corporate giant like Snap to leverage AI for serious
• 12:58
  efficiency. Exactly, you don't need 1000
• 13:00
  person layoff to build an AI pod.
• 13:02
  You can just build 1 yourself using off the shelf tools to
• 13:05
  sell your old jackets. It's the ultimate
• 13:07
  democratization of tech. Well, that brings us to the end
• 13:11
  of our deep dive today. But before we go, I know you
• 13:15
  always like to leave us with a final thought to Mull over.
• 13:17
  I do, and considering everything we covered today, I want to
• 13:20
  leave you with this. If AI is now writing 65% of the
• 13:24
  code at major tech companies and it's solving century old math
• 13:28
  problems and running our desktop workflows right, how long until
• 13:32
  an autonomous AI agent decides it needs more compute power and
• 13:36
  starts quietly renting GPU's from a rebranded sneaker company
• 13:39
  on its own? Oh wow, now that is a
• 13:42
  terrifyingly plausible thought to end on an AI secretly funding
• 13:47
  new bird AI. It could happen.
• 13:49
  It really could. Well, thank you so much for
• 13:51
  joining us for this deep dive. If you want to stay updated on
• 13:54
  all these crazy shifts, please make sure you subscribe to the
• 13:56
  show. Yeah, please subscribe.
• 13:57
  And we would warmly invite you to rate the show with five stars
• 14:01
  if you found this valuable. It really helps us out.
• 14:04
  Have a great rest of your day and we'll see you back for
• 14:06
  tomorrow's episode.