Latest / Connecting the Dots / Heat Waves, SharePoint Exploits, Qwen3-Coder, and McKinsey’s China Reset

Transcript

Heat Waves, SharePoint Exploits, Qwen3-Coder, and McKinsey’s China Reset · Connecting the Dots

• 0:00
  Today's tech news moves fast, but understanding
• 0:03
  it shouldn't be hard. That's why Snarful 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 back to Connecting the Dots. Your
• 0:18
  daily look at the tech news that matters. It's
• 0:20
  Wednesday, July 23, 2025. Great to have you with
• 0:24
  us. Okay, quick check in before we get to the
• 0:26
  tech stories. Sacramento weather. Sunny. Nice
• 0:29
  day here. High of 83 degrees Fahrenheit. Pretty
• 0:31
  pleasant. Nice. Nationally though, bit of a different
• 0:34
  story. Still got that heat wave hitting the central
• 0:37
  and southeast U .S. hard. Yeah, and it looks
• 0:40
  like flash flood threats up in the north central
• 0:42
  U .S. Upper Midwest, Great Lakes, southeast,
• 0:45
  even Guam. Tropical disturbances causing trouble.
• 0:49
  Busy weather day. Alright, how about the markets?
• 0:51
  What's the pulse? Okay, so the Dow Jones closed
• 0:53
  yesterday at forty four thousand seven hundred
• 0:55
  and thirty nine point nine four That's up just
• 0:58
  about seventy nine points for the day point one
• 1:00
  eight percent up point eight four percent for
• 1:03
  the week All right slow climb there S &P S &P
• 1:05
  500 finished at six thousand three hundred and
• 1:07
  twenty eight point six barely moved really up
• 1:09
  one point seven six points That's point zero
• 1:11
  three percent about a point three eight percent
• 1:13
  over the week So, you know generally positive
• 1:14
  trend. Okay and Bitcoin always volatile Yeah,
• 1:17
  Bitcoin closed 118 ,203 .27. He's down yesterday,
• 1:22
  about 1 ,800 bucks, 1 .5%. Still up for the week,
• 1:25
  though, 0 .66%. So swings and roundabouts. Always
• 1:29
  is. OK, sports break. Did we get the Giants'
• 1:33
  score from last night? We did, and it's good
• 1:34
  news. Giants absolutely crushed the Braves 9
• 1:38
  -0. 9 -0, wow. Yeah, Landon Roop got the win.
• 1:41
  Braves' Matt Olson had a decent night, 3 for
• 1:43
  5, but it didn't matter. Just a huge win for
• 1:46
  the Giants. Feels good. Definitely does. Awesome.
• 1:50
  So feeling good after that win. Let's transition.
• 1:52
  Time for our Morningstar. Right. The Snarfl Tech
• 1:55
  article report. We've got three big ones today
• 1:57
  that really connect some interesting dots. Yeah,
• 2:00
  we're looking at a major security breach, some
• 2:02
  pretty cutting edge AI coding developments, and
• 2:05
  how geopolitics is directly hitting big tech
• 2:09
  firms. OK, let's dive into that first story.
• 2:11
  It's a concerning one. Microsoft is pointing
• 2:13
  fingers at China -linked hacking groups for some
• 2:15
  widespread attacks. Yeah, this is serious stuff.
• 2:18
  They've tied several groups, state -sponsored
• 2:20
  actors, they believe, to exploiting a previously
• 2:23
  unknown vulnerability. A, a zero day. In Microsoft
• 2:26
  SharePoint, right. But the on -premise version.
• 2:29
  Exactly. On -premise servers only. Not SharePoint
• 2:31
  Online and the cloud, which is the key distinction,
• 2:35
  but for organizations running their own SharePoint
• 2:37
  servers. Big problem. And they've named the groups.
• 2:40
  They have. Microsoft identified Linen Typhoon,
• 2:43
  Violet Typhoon, and Storm 2603. They used a method,
• 2:49
  Microsoft's calling it tool shell. to get into
• 2:51
  these servers. So a zero day, just to clarify
• 2:54
  for everyone, that means the attackers found
• 2:56
  a flaw before Microsoft knew about it, right?
• 2:58
  So there was no patch initially. Precisely. Like
• 3:01
  having a secret key nobody knew existed until
• 3:03
  the door was already open. It gives the attackers
• 3:05
  a huge advantage. No warning, no immediate defense.
• 3:08
  And the impact wasn't small, was it? No, not
• 3:10
  at all. We're talking dozens of organizations
• 3:12
  hit worldwide. Multinational companies, government
• 3:15
  entities. Sectors like telcos, software. Right.
• 3:19
  And This is the really alarming part. The U .S.
• 3:23
  National Nuclear Security Administration, the
• 3:25
  NNSA, they were infiltrated. Wow. The agency
• 3:29
  that manages the U .S. nuclear arsenal. That's
• 3:31
  the one. Now Microsoft and NNSA say no classified
• 3:34
  data was confirmed stolen. But just the fact
• 3:38
  that they got in, that's deeply worrying. Absolutely.
• 3:41
  It raises questions about what else they could
• 3:42
  access or what kind of foothold they might have
• 3:44
  established for later. Exactly. Pre -positioning
• 3:48
  is a major concern with nation -state actors.
• 3:50
  Getting access now for potential future use,
• 3:53
  Intel gathering, disruption. It's a long game.
• 3:56
  And it wasn't just the NNSA, right? Other federal
• 3:58
  agencies, too. Yeah, other US federal agencies,
• 4:01
  a private US university, even a California -based
• 4:03
  private energy provider. It's a pretty broad
• 4:05
  list. OK, so how did this tool shell actually
• 4:07
  work? What could they do once they were in? Well,
• 4:10
  the flaw essentially gave them remote access,
• 4:12
  unauthenticated access. So they didn't need valid
• 4:15
  logins initially. Yeah. From there, they can
• 4:18
  steal credentials, you know, usernames, passwords,
• 4:21
  that sort of thing. They could grab cryptographic
• 4:24
  keys, super sensitive stuff, and potentially
• 4:28
  install back doors to maintain access. So full
• 4:31
  run of the place, basically. accessing files,
• 4:33
  configurations, even running their own code on
• 4:35
  the server. That's what Microsoft indicated,
• 4:37
  yes. Full access to SharePoint content, file
• 4:39
  systems, internal settings, executing code over
• 4:42
  the network. It's pretty complete control. OK,
• 4:45
  so what's being done? Microsoft patched it quickly,
• 4:48
  I assume. They did. Released emergency patches.
• 4:51
  And CISA, the U .S. Cybersecurity Agency, they
• 4:54
  added it to their known exploited vulnerabilities
• 4:57
  catalog. That means federal agencies had to patch.
• 5:00
  Like, immediately. OK, so the immediate hole
• 5:02
  is plugged for those who patch. But, and this
• 5:04
  is the kicker, a proof of concept exploit is
• 5:06
  now public. So the instructions on how to do
• 5:10
  the hack are basically out there for anyone.
• 5:12
  Pretty much. It means less sophisticated hackers,
• 5:14
  maybe cyber criminals looking for ransomware
• 5:16
  opportunities, they can now use this technique
• 5:19
  relatively easily. The risk hasn't gone away
• 5:21
  just because there's a patch. It's actually spread.
• 5:24
  Right. So what's the big takeaway here for, you
• 5:26
  know, businesses, professionals listening? Well,
• 5:29
  number one. Patch management, especially for
• 5:31
  on -premise systems, is absolutely critical.
• 5:34
  You just cannot delay these updates. The window
• 5:36
  between vulnerability discovery and widespread
• 5:39
  exploitation is shrinking fast. Ignoring it is
• 5:42
  basically leaving the door wide open, especially
• 5:44
  now the method is public. Exactly. It puts critical
• 5:47
  business data, operations, everything at risk,
• 5:50
  and for everyone else, families. individuals,
• 5:54
  breaches of government agencies like NNSA or
• 5:56
  critical infrastructure like energy providers,
• 5:59
  they erode trust and they can have real -world
• 6:02
  ripple effects on services we all rely on. It
• 6:05
  just underscores how interconnected and frankly
• 6:07
  vulnerable our digital world can be. A stark
• 6:10
  reminder, definitely. Okay, so from cybersecurity
• 6:13
  threats, let's shift gears to innovation and
• 6:16
  our second story involves artificial intelligence.
• 6:19
  Right. This one comes from Alibaba, the Chinese
• 6:22
  tech giant. They've just launched something called
• 6:24
  Quen 3 Coder. Okay, Quen 3 Coder. What's the
• 6:27
  deal? Alibaba is calling it their most advanced
• 6:30
  open -source AI model specifically designed for
• 6:32
  software development. And the key phrase here
• 6:35
  is, agentic AI coding tasks. Agentic? What does
• 6:40
  that mean in this context? Like, AI agents doing
• 6:42
  the coding? Sort of, yeah. It means the AI isn't
• 6:46
  just suggesting code snippets. It's designed
• 6:48
  to tackle programming challenges more autonomously,
• 6:51
  like understanding a complex requirement, figuring
• 6:53
  out the steps, generating the code, maybe even
• 6:55
  debugging it, managing workflows. Wow. OK. So
• 6:59
  it's like a more independent coding partner.
• 7:00
  That's a good way to put it. They built it on
• 7:02
  a massive architecture of 480 billion parameters
• 7:04
  using a mixture of experts approach, which is
• 7:06
  quite sophisticated. And how does it stack up?
• 7:09
  Are they making performance claims? Oh, yeah.
• 7:11
  Big claims. They're saying Quen 3 Coder sets
• 7:13
  new benchmarks for open models in these Sage
• 7:16
  and Git coding tasks, plus things like controlling
• 7:19
  web browsers or using software tools. Better
• 7:21
  than other open models. That's the claim. They
• 7:24
  even say it performs on par with some top -closed
• 7:27
  models from the U .S., like Anthropix, Claude's
• 7:29
  Sonnet 4, or OpenAI's GPT -4, at least in certain
• 7:32
  areas. better than their domestic Chinese rivals.
• 7:37
  Interesting. And I saw something about a huge
• 7:39
  context window. Yes. This is pretty impressive.
• 7:42
  Two hundred and fifty six thousand tokens, which
• 7:45
  is already large, but they say it's extendable
• 7:47
  up to one million tokens. A million tokens. OK,
• 7:50
  translate that for us. Why is that a big deal
• 7:52
  for an AI that writes code? It's huge because
• 7:56
  code bases are often enormous and complex. A
• 7:59
  massive context window means the AI can essentially
• 8:03
  read and remember a much larger chunk of code,
• 8:06
  maybe the entire project, its libraries, its
• 8:08
  documentation all at once. Ah, so it understands
• 8:11
  the bigger picture, not just the little piece
• 8:12
  it's working on. Exactly. It can grasp the architecture,
• 8:15
  how different parts interact, maintain consistency,
• 8:17
  avoid conflicts. For developing really complex
• 8:20
  software, that ability to see the whole context
• 8:22
  is, well, a potential game changer. It helps
• 8:25
  generate more coherent, integrated code. OK,
• 8:27
  I can see that. So why does this matter to, say,
• 8:30
  a software developer listening right now? Well,
• 8:32
  it could be a massive productivity boost. Imagine
• 8:35
  offloading tedious, repetitive coding or even
• 8:38
  complex refactoring tasks to an AI like this.
• 8:42
  It frees up human developers to focus on the
• 8:45
  really creative, high level problem solving and
• 8:48
  design. Your coding sidekick, but a really, really
• 8:51
  smart one. Sounds potentially transformative
• 8:53
  for how software gets built and for businesses.
• 8:56
  Faster development cycles, definitely. That can
• 8:58
  mean lower costs, getting products to market
• 9:00
  quicker. big competitive advantages potentially.
• 9:03
  And for the rest of us, what's the broader significance?
• 9:05
  It just shows the incredible pace of AI development
• 9:08
  worldwide. And, you know, it highlights that
• 9:11
  China is making really significant strides in
• 9:13
  this area. This isn't theoretical. This kind
• 9:14
  of AI could soon be powering the tools and services
• 9:17
  we use every day. It's a global innovation race
• 9:20
  for sure. Really fascinating stuff. OK, so we
• 9:22
  have state -sponsored hacking, incredibly advanced
• 9:25
  AI coders. Our third story kind of ties these
• 9:28
  themes together through the lens of geopolitics.
• 9:30
  Yeah, this involves McKinsey, the huge global
• 9:32
  consulting firm. The report is that they've actually
• 9:35
  stopped their China -based operations from doing
• 9:37
  any consultancy work related to generative AI.
• 9:40
  Stopped it completely. Why? It seems to be a
• 9:43
  direct result of pressure and scrutiny coming
• 9:45
  from Washington, from U .S. lawmakers. Ah, okay.
• 9:49
  What's the specific concern from the U .S. side?
• 9:52
  Well, you had several Republican lawmakers, Representative
• 9:54
  Molinar, Senators Rubio and Ernst. calling for
• 9:57
  a federal investigation into McKinsey. An investigation?
• 10:00
  Into what exactly? Their core issue is this.
• 10:04
  McKinsey has big contracts with the U .S. Department
• 10:06
  of Defense, right? Over like $480 million worth
• 10:09
  since 2008. But at the same time, they've also
• 10:13
  been advising Chinese government entities, provincial
• 10:15
  governments, state -owned enterprises. Okay.
• 10:17
  I see the potential conflict. Right. The lawmakers
• 10:20
  are arguing that this creates a serious risk
• 10:22
  to U .S. national security. The worry is that
• 10:24
  McKinsey could be, even inadvertently, helping
• 10:27
  China's military or economy, especially in sensitive
• 10:30
  high -tech areas like generative AI. That's a
• 10:32
  heavy accusation. How has McKinsey responded?
• 10:35
  Well, in the past, they've defended their work,
• 10:38
  saying they have strict policies to avoid conflicts
• 10:40
  and that they don't work directly for the Chinese
• 10:42
  Communist Party or the central government in
• 10:44
  Beijing. They focus more on multinational companies
• 10:48
  operating in China and private Chinese businesses.
• 10:51
  But this move to halt AI work suggests the pressure
• 10:55
  worked, or they see the writing on the wall.
• 10:57
  It certainly looks that way. Reports also say
• 11:00
  McKinsey has already been cutting back on government
• 11:03
  -linked clients in China, and even reduced their
• 11:05
  workforce there by about a third. That's nearly
• 11:08
  500 people. So yeah, it looks as a significant
• 11:12
  strategic pullback, likely driven by this geopolitical
• 11:15
  heat. So why does this McKinsey story matter
• 11:17
  to, you know, businesses or professionals maybe
• 11:20
  not involved with huge consulting firms. It's
• 11:23
  a really clear example of how these high -level
• 11:25
  geopolitical tensions between the US and China
• 11:28
  are directly impacting how multinational companies
• 11:30
  have to operate. It's not just theory anymore.
• 11:33
  Right. If you do business in both countries,
• 11:35
  especially in tech, you're caught in the middle.
• 11:37
  Exactly. It's becoming a tightrope walk. Companies
• 11:40
  are being forced to navigate these national security
• 11:42
  concerns, potentially choose sides or drastically
• 11:46
  change. how they do business in sensitive sectors.
• 11:48
  It's not just about markets anymore. It's about
• 11:50
  alignment. And for the wider economy. It feeds
• 11:53
  into this whole trend. People talk about decoupling
• 11:56
  or de -risking this strategic separation between
• 11:59
  the U .S. and Chinese economies, especially in
• 12:02
  critical technologies like A .I. chips, quantum
• 12:06
  computing. And that separation can affect everything,
• 12:09
  right? Supply chains, investment, where innovation
• 12:11
  happens. Absolutely. It reshapes global trade
• 12:14
  flows, investment patterns. It forces companies
• 12:17
  to make really tough choices about where they
• 12:19
  put their resources, where they develop new tech,
• 12:21
  who their partners are. It has a whole layer
• 12:23
  of complexity and risk to international business.
• 12:26
  It really illustrates how interconnected tech
• 12:28
  business and global politics are now. You can't
• 12:31
  really separate them. Not anymore. This and McKinsey
• 12:33
  situation is probably just one visible example
• 12:36
  of what many companies are grappling with behind
• 12:38
  the scenes. How do you operate globally when
• 12:41
  the major powers are increasingly viewing each
• 12:43
  other through a lens of strategic competition,
• 12:45
  especially in technology? It's the defining challenge
• 12:49
  for global business right now. Wow. OK, quite
• 12:51
  a journey today. So let's quickly recap the three
• 12:54
  big stories we connected. Right. We started with
• 12:56
  those serious Microsoft SharePoint cyber attacks
• 12:59
  linked to Chinese state actors hitting critical
• 13:02
  organizations and showing the urgent need for
• 13:05
  cybersecurity vigilance. Then we jumped to the
• 13:08
  cutting edge of AI with Alibaba's Quen 3 Coder,
• 13:11
  an open source model aiming to automate complex
• 13:13
  software development, a potential game changer
• 13:15
  for tech productivity. And finally, we saw how
• 13:17
  geopolitical tensions are forcing major players
• 13:20
  like McKinsey to make tough choices, halting
• 13:22
  generative AI work in China due to U .S. scrutiny,
• 13:26
  highlighting that growing U .S.-China tech divide.
• 13:29
  From zero -day exploits to AI agents to global
• 13:32
  strategy shifts. It really shows how these different
• 13:35
  threads are weaving together to shape our world.
• 13:37
  Absolutely. Understanding these connections is
• 13:40
  really key to navigating what's coming next in
• 13:43
  technology and business. Well, hopefully we helped
• 13:45
  connect some of those dots for you today. Remember,
• 13:48
  we're here to help at Snarful, so reach out on
• 13:50
  our website if you have any more questions. That's
• 13:56
  a wrap on today's top stories. Remember, tech
• 13:59
  doesn't have to be complicated, and we're here
• 14:01
  to help connect the dots. If you've got questions
• 14:04
  or want to learn more, visit us at snarful .com.
• 14:08
  Thanks for listening, and we'll catch you next
• 14:10
  time.