Latest / Tech Talks With Kinsoft / Last Week in Tech – SpaceX Buys Cursor for $60B, Washington Pulls Anthropic's Top Models, and the Klue Breach Spreads

Transcript

Last Week in Tech – SpaceX Buys Cursor for $60B, Washington Pulls Anthropic's Top Models, and the Klue Breach Spreads · Tech Talks With Kinsoft

• 0:00
  Imagine spending, I mean, literally millions
• 0:02
  of dollars integrating the world's most advanced
• 0:04
  AI into your business. Right. Your development
• 0:08
  team loves it. Your daily operations rely on
• 0:10
  it. Your revenue is practically tied to it at
• 0:15
  this point. Yeah, it's a load -bearing pillar
• 0:16
  for the company. Exactly. And then overnight,
• 0:19
  you wake up to find a government has legally
• 0:22
  forced the provider to just pull the plug. Just
• 0:25
  shut it all down. Yeah. The software hasn't crashed.
• 0:28
  The servers aren't down. It has simply been seized
• 0:31
  and your business is left completely in the dark.
• 0:34
  Which is a terrifying thought. It really is.
• 0:36
  So welcome to our deep dive. Whether you are
• 0:39
  catching up for a major tech strategy meeting
• 0:41
  this week or you are just, you know, insanely
• 0:44
  curious about the fragile digital landscape we
• 0:46
  are all navigating, we built this deep dive specifically
• 0:48
  for you. We really did. Today, we are pulling
• 0:51
  apart a single incredibly dense update from Tech
• 0:54
  Talks with Kinsoft. This is... dated late June
• 0:57
  2026. And our mission here is to just cut through
• 1:00
  the daily noise and examine a massive rapid shift
• 1:03
  happening right now regarding who actually controls
• 1:06
  the digital tools your business relies on every
• 1:09
  single day. So, OK, let's unpack this. Yeah,
• 1:12
  it really is a fascinating collision to unpack.
• 1:14
  We are looking at a critical moment in tech history
• 1:16
  where incredibly rapid unchecked innovation is
• 1:20
  suddenly smashing head on into, well, extreme
• 1:23
  corporate consolidation on one side and then
• 1:26
  forceful government intervention on the other.
• 1:27
  Right. It is just a fragile new reality for anyone
• 1:30
  who builds, manages or simply uses technology
• 1:33
  today. And I mean, nothing screams extreme corporate
• 1:36
  consolidation quite like the sheer scale of the
• 1:38
  numbers in this first story. Oh, absolutely.
• 1:40
  Let's look at what happens when tech giants decide
• 1:42
  to swallow the very tools that build the tech
• 1:44
  industry. So on June 16th, 2026, just days after
• 1:48
  pulling off its own blockbuster IPO, SpaceX filed
• 1:51
  to acquire any sphere. The company behind Cursor.
• 1:54
  Exactly. For context, for you listening, AnySphere
• 1:58
  is the company behind that massively popular
• 2:00
  AI coding tool called Cursor. And the price tag
• 2:04
  for this acquisition is an all -stock transaction
• 2:06
  worth around $60 billion. $60 billion. I mean,
• 2:11
  $60 billion is a staggering figure. It's wild.
• 2:15
  It is, especially when you consider that, according
• 2:17
  to the source material, this makes it the largest
• 2:20
  acquisition of a venture -backed startup in history.
• 2:23
  Wow. In history? Yeah. But when you look at the
• 2:26
  underlying mechanics of any sphere's business,
• 2:29
  you start to see the strategic math behind it.
• 2:31
  Cursor was carrying somewhere between $2 .5 and
• 2:34
  $4 billion in annualized revenue. Okay, so a
• 2:37
  massive cash engine already. Exactly. Yeah. And
• 2:39
  more importantly, they had over a million paying
• 2:41
  users. And it's not just Cursor operating at
• 2:43
  this scale, right? The source points out that
• 2:45
  nearly every serious AI coding tool is now sitting
• 2:49
  under a mega corporate umbrella. Yeah, the whole
• 2:51
  board has been bought up. Right. You've got Copilot
• 2:53
  owned by Microsoft. You've got Windsurf and Codex
• 2:55
  orbiting OpenAI. You have Claude Code over at
• 2:59
  Anthropic. And now Cursor is absorbed by SpaceX.
• 3:02
  It's a clean sweep. It really feels like the
• 3:05
  era of the railroad barons where the independent
• 3:07
  companies laying the tracks are suddenly being
• 3:09
  bought up by three or four massive empires. That's
• 3:12
  a really good way to put it. But, I mean, push
• 3:15
  back here for a second. Companies grow, right?
• 3:17
  And startups get acquired by bigger fish all
• 3:20
  the time in Silicon Valley. That's the whole
• 3:22
  model. Sure. So is this actually bad for the
• 3:25
  average developer or is it just the natural evolution
• 3:28
  of tech? Well, what's fascinating here is that.
• 3:31
  It entirely shifts the power dynamic in a way
• 3:33
  that introduces massive vendor risk for the end
• 3:36
  user. How so? Let's explore how a software development
• 3:40
  team actually functions today. When a business
• 3:42
  decides to use an AI coding tool like Cursor
• 3:45
  or Copilot, they don't just casually try it out
• 3:48
  on a weekend. Right. It's not like downloading
• 3:50
  a new notes app. Exactly. They integrate it into
• 3:53
  their core systems. They standardize their coding
• 3:55
  workflows on it. Their developers build intricate
• 3:58
  habits and processes around how that specific
• 4:01
  tool functions, you know, how it suggests code,
• 4:03
  how it debugs. Okay, yeah, I see that. So if
• 4:06
  my entire engineering team spends a year learning
• 4:09
  the exact quirks and shortcuts of one specific
• 4:12
  AI tool, changing it out isn't just swapping
• 4:15
  a piece of software. No, not at all. It's practically
• 4:18
  retraining the whole department. It becomes pure
• 4:20
  muscle memory. Yes. And that muscle memory is
• 4:24
  what creates absolute lock -in. And this is where
• 4:27
  we have to understand the mechanics of an API
• 4:29
  and application programming interface. Right.
• 4:31
  In the past, you bought software in a box, right?
• 4:34
  Yeah, you install it and it's yours. Exactly.
• 4:37
  Right. But today, you are just running access
• 4:39
  to a data pipe hosted in the cloud. Once that
• 4:42
  tool is deeply embedded in your workflow, your
• 4:45
  leverage as a buyer drops to near zero. Because
• 4:48
  you can't just leave. Right. If the independent
• 4:51
  startup you initially signed a flexible contract
• 4:53
  with is suddenly absorbed by a $60 billion titan
• 4:57
  like SpaceX or Microsoft, the rules of the game
• 4:59
  change. You have to know who ultimately owns
• 5:02
  the tools you are betting your business on. Because
• 5:04
  your leverage just shifted dramatically. Entirely.
• 5:07
  If that new mega owner decides to double the
• 5:10
  price or drastically alter the privacy policy
• 5:13
  regarding how your proprietary code is fed back
• 5:16
  into their machines to train future models, I
• 5:19
  mean, you have very little recourse. The market
• 5:22
  is shrinking to a handful of giants and they
• 5:24
  hold all the leverage. Wow. So you might think
• 5:28
  you are working with a plucky, independent startup,
• 5:30
  but you could wake up tomorrow and find out you
• 5:32
  are actually just a tenant on one of the mega
• 5:34
  empire's land. And you are subject to the rent
• 5:36
  hikes. But as powerful as these corporate empires
• 5:40
  are, there is a higher power that can step in
• 5:42
  and remind them they aren't completely in charge.
• 5:44
  Yes, there absolutely is. So you might be renting
• 5:46
  assets from one of these mega empires. But what
• 5:49
  happens when a government decides to seize the
• 5:52
  building? Yeah, this is where the landscape abruptly
• 5:54
  shifts from corporate strategy to national security.
• 5:57
  Right. And it introduces a level of operational
• 5:59
  risk most businesses are completely unprepared
• 6:03
  for. So here's where it gets really interesting.
• 6:05
  In early June, after Anthropic launched its most
• 6:08
  powerful new models, the U .S. Commerce Department
• 6:11
  stepped in. And they did not mess around. No,
• 6:13
  they didn't. They issued an export control directive
• 6:16
  ordering Anthropic to suspend worldwide access
• 6:19
  to those top tier models. Worldwide. Worldwide.
• 6:22
  And not just for the general public either. This
• 6:24
  suspension even applied to Anthropic's own foreign
• 6:27
  national staff working inside the company. Which
• 6:29
  is just a massive move. The mechanism behind
• 6:32
  this unprecedented shutdown, according to the
• 6:34
  source material, was a discovered jailbreak.
• 6:37
  OK, a jailbreak. Right. The government realized
• 6:39
  these models contained a vulnerability that could
• 6:42
  potentially turn them into unrestricted cyber
• 6:44
  tools. Wow. Yeah. To understand why the Commerce
• 6:47
  Department reacted so aggressively, we need to
• 6:50
  quickly define what we mean by frontier models.
• 6:52
  Right, because a frontier model isn't just a
• 6:54
  slightly smarter chatbot. Exactly. It is a highly
• 6:57
  advanced AI system that matches or exceeds the
• 7:01
  capabilities of the most powerful existing models
• 7:03
  on the market. Because they are pushing the absolute
• 7:05
  boundaries of what AI can do, they introduce
• 7:08
  entirely novel security risks. And one of those
• 7:12
  risks is that jailbreak mechanism. Large language
• 7:16
  models are fundamentally designed to predict
• 7:18
  text based on the prompts they receive, right?
• 7:21
  Yeah. They have safety guardrails built in to
• 7:24
  prevent them from, say, writing malicious code
• 7:26
  or generating instructions for a biological weapon.
• 7:29
  Right. The stuff you definitely don't want them
• 7:31
  doing. Exactly. Yeah. But a jailbreak, often
• 7:34
  achieved through a technique called prompt injection,
• 7:36
  uses adversarial, highly specific language to
• 7:41
  confuse the model. It essentially tricks it into
• 7:44
  ignoring its safety training. Oh, I see. Yeah.
• 7:47
  And once those rails are bypassed, an advanced
• 7:50
  frontier model can theoretically be used to write
• 7:52
  zero -day exploits or automate cyber attacks
• 7:55
  at scale. So the Commerce Department viewed this
• 7:58
  text generator not as a software product, but
• 8:00
  as a potential cyber weapon. Just to be clear
• 8:03
  on the facts of the dispute here, Anthropic did
• 8:06
  comply with the directive. They did, yes. They
• 8:08
  took the models offline globally, but they also
• 8:10
  publicly disagreed with the government's action,
• 8:13
  pointing out that this exact same weakness affects
• 8:15
  other frontier models out there, not just theirs.
• 8:18
  Right. They felt singled out. Yeah. And, you
• 8:21
  know, we are looking purely at the facts of the
• 8:23
  event, not taking sides here. But if I'm a business
• 8:26
  leader. who just built my company's operations
• 8:29
  on one of these load -bearing ai models and it
• 8:33
  vanishes overnight i mean what do i even do is
• 8:36
  this just the new digital act of god well if
• 8:40
  we connect this to the bigger picture it absolutely
• 8:42
  is a digital act of god yeah and it forces a
• 8:45
  complete rewrite of how businesses handle disaster
• 8:48
  planning and continuity risk. Oh, so. Think about
• 8:51
  the physical mechanics of traditional software.
• 8:53
  If you bought an on -premise database system
• 8:55
  10 years ago, you installed it on physical servers
• 8:58
  sitting in a room you controlled. Right, under
• 9:00
  lock and key in your own office? Yes. If the
• 9:03
  government sanctioned the creator of that software
• 9:05
  or the company went bankrupt, your software didn't
• 9:07
  suddenly uninstall itself from your hard drives.
• 9:10
  The lights stayed on. Right. I might not get
• 9:12
  the next security patch, but my customer service
• 9:14
  team could still log in the next morning. Exactly.
• 9:17
  But today, with frontier AI models, you don't
• 9:20
  possess the software. You are renting a continuous
• 9:22
  stream of data from a cloud server. Just a pipe.
• 9:25
  Just a pipe. A year ago, the idea that a government
• 9:29
  could step in and legally force a provider to
• 9:31
  sever that data stream, instantly breaking the
• 9:34
  core functionality of businesses all over the
• 9:36
  world. That simply wasn't a threat model most
• 9:39
  IT departments were seriously considering. No,
• 9:42
  you'd think about server outages, not federal
• 9:44
  seizures. Right. But this event proves that a
• 9:47
  frontier AI model can be switched off globally
• 9:50
  overnight on national security grounds. So if
• 9:53
  your operations rely on AI to function, if that
• 9:56
  AI is load -bearing for your company, you suddenly
• 9:59
  have to ask yourself, what happens if your primary
• 10:02
  model becomes a geopolitical football tomorrow
• 10:04
  morning? You need a backup plan. Yeah, a backup
• 10:07
  plan that doesn't just rely on a different tool,
• 10:09
  but perhaps doesn't even rely on the same national
• 10:11
  jurisdiction. It is a stark reminder that while
• 10:14
  megacorporations control the tools, sovereign
• 10:16
  power still dictates the ultimate terms of service.
• 10:19
  It highlights a massive vulnerability at the
• 10:22
  very top of the tech food chain. But, you know,
• 10:24
  the irony here is staggering. Oh, for sure. We
• 10:26
  just talked about massive $60 billion corporate
• 10:29
  buyouts and top tier models that the government
• 10:32
  can turn off overnight like cyber weapons. But
• 10:34
  you don't even need a federal directive or a
• 10:37
  state sponsored quantum hack to bring a company
• 10:40
  to its knees today. Not at all. According to
• 10:43
  the Kinsoft update, the most widespread threat
• 10:45
  right now isn't futuristic at all. It's basic
• 10:48
  plumbing. You just need a sloppy digital handshake.
• 10:51
  The deeply unglamorous reality of tech hygiene.
• 10:53
  Exactly. So let's look at the Clue breach. Clue
• 10:57
  is a market intelligence software vendor, and
• 11:00
  they were compromised not by a sophisticated
• 11:02
  zero -day exploit, but through a forgotten integration
• 11:05
  credential. A forgotten credential. Yeah. And
• 11:07
  the fallout is massive, with the breach spreading
• 11:10
  to nearly 200 organizations. And to understand
• 11:13
  the sheer scale of this, we really need to look
• 11:15
  at how this mechanically happened. Okay, let's
• 11:17
  hear it. The attackers managed to harvest what
• 11:19
  are known as OAuth tokens. Now, for our listeners,
• 11:23
  OAuth is the industry standard protocol that
• 11:26
  allows one application to access data in another
• 11:28
  application without you ever having to hand over
• 11:31
  your actual password. Okay. So when you click
• 11:34
  a button that says sign in with Google, or when
• 11:36
  you authorize a third -party app to connect to
• 11:39
  your Salesforce environment, you are using O.
• 11:41
  Oh, right. So what does this all mean? It's like
• 11:44
  a hotel valet key. A valet key, exactly. You
• 11:47
  hand the parking attendant a key that allows
• 11:50
  them to turn on the engine and park the car,
• 11:52
  but it physically prevents them from opening
• 11:55
  the trunk or the glove box where you keep your
• 11:57
  valuables. O is a digital valet key. That is
• 12:01
  a perfect analogy. And so Clue was handed a digital
• 12:05
  valet key to connect seamlessly into its customers'
• 12:08
  Salesforce environments to pull the data it needed
• 12:10
  for market intelligence. Right. And that valet
• 12:13
  key analogy explains exactly how the attackers
• 12:16
  bypassed all the expensive front door security.
• 12:18
  Because they just went for the keys. Yes. When
• 12:21
  an oath connection is made, an access token is
• 12:23
  generated and stored in a database on a server.
• 12:26
  Some of these tokens are incredibly long -lived.
• 12:29
  I mean, they don't automatically expire unless
• 12:31
  someone manually revokes them. Wow. They just
• 12:34
  stay active forever. Essentially, yes. The attackers
• 12:37
  didn't try to pick the complex locks on the front
• 12:40
  doors of these 200 companies. They broke into
• 12:43
  the valet stand. Stole the master ring of digital
• 12:46
  valet keys and quietly used those existing trusted
• 12:50
  tokens to query the customer's CRM data directly.
• 12:54
  That is wild. They bypassed the firewalls, the
• 12:56
  biometric logins, the complex password requirements,
• 12:59
  all because they had a pre -approved digital
• 13:01
  pass. Yeah. And this wasn't just hitting small,
• 13:04
  unsuspecting companies without IT departments,
• 13:06
  right? No, not at all. The Kinsoft to SourceNuts,
• 13:09
  the victim list is a who's who of elite security
• 13:11
  firms. We are talking about hundreds. Interest
• 13:14
  tenium, recorded future. These are organizations
• 13:18
  whose entire business model is catching cyber
• 13:20
  threats, and they were compromised through a
• 13:22
  vendor's forgotten token. It's like spending
• 13:25
  a million dollars on a state -of -the -art bank
• 13:26
  vault, but leaving the spare key under the doormat
• 13:29
  for the window cleaner. Exactly. And this raises
• 13:31
  an important question, because it fundamentally
• 13:33
  challenges how we view digital security. We spend
• 13:37
  millions of dollars and countless hours worrying
• 13:40
  about the strength of the vault door. But this
• 13:43
  incident proves the real danger lies in the trusted
• 13:46
  side channels. Huntress, Tanium, Recorded Future.
• 13:50
  These organizations are filled with world -class
• 13:54
  experts. If they can get caught by a compromised
• 13:56
  OOOTH token, anyone can. Because it isn't a failure
• 14:00
  of intelligence or a lack of sophisticated security
• 14:02
  software? No, it is a pure failure of hygiene.
• 14:05
  The takeaway here is the compounding nature of
• 14:08
  vulnerability. One single... trusted app connection
• 14:12
  can become a doorway into hundreds of organizations.
• 14:15
  Wow. You can build the most secure internal network
• 14:17
  in the world, but the moment you grant a third
• 14:20
  -party app an Outh token to reach your data,
• 14:22
  your security posture is instantly lowered to
• 14:25
  match that third -party vendor's security. You're
• 14:27
  only as strong as your weakest integration. Exactly.
• 14:29
  If they forget to rotate a credential or if a
• 14:31
  token sits on a server untouched for two years,
• 14:33
  you are entirely exposed. So the actionable advice
• 14:36
  here for you listening right now is to go audit
• 14:38
  your systems today. Look at which third -party
• 14:41
  applications have OAuth access to your Salesforce,
• 14:44
  your Microsoft 365, your Google Workspace. Yes,
• 14:48
  please do that. If you see an integration that
• 14:50
  is stale or an app your team hasn't actively
• 14:53
  used since 2024, revoke the permission immediately.
• 14:56
  Cut the connection. Take the valet key back.
• 14:59
  It is tedious administrative work, I know. But
• 15:02
  as the clue breach clearly demonstrates, it is
• 15:05
  arguably the most critical defense mechanism
• 15:07
  you have. It really paints a picture of an incredibly
• 15:09
  fragile landscape. I mean, megacorporations are
• 15:12
  buying up the foundational tools. Governments
• 15:15
  are shutting down the frontier models overnight.
• 15:17
  And forgotten tokens are exposing the world's
• 15:19
  top security firms. It's a lot to take in. It
• 15:21
  is. So taking a step back, how is the rest of
• 15:24
  the world reacting to this on a macro level?
• 15:26
  Well, the contrast on the global stage is what
• 15:28
  really stands out on the Kinsoft update. At the
• 15:31
  G7 summit in France, world leaders held an unprecedented
• 15:35
  working session. They actually brought in the
• 15:37
  top executives of OpenAI, Anthropic, and Google
• 15:40
  DeepMind. Okay, so the absolute biggest players.
• 15:44
  Yes. Getting all of those major AI figures in
• 15:47
  the same room with the G7 leaders is a... highly
• 15:51
  unusual and major diplomatic event. Right. You'd
• 15:54
  expect some massive ironclad regulations to come
• 15:58
  out of that. But the source says they landed
• 16:00
  on a package of largely voluntary AI commitments
• 16:03
  focused on child safety online. Yes. Mostly voluntary.
• 16:07
  Wait a minute. We just spent 10 minutes talking
• 16:09
  about. The U .S. Commerce Department forcefully
• 16:11
  shutting down anthropics models globally because
• 16:14
  they treat them like literal cyber weapons. Right.
• 16:16
  And now the G7, the most powerful international
• 16:18
  table in the world, is rolling out voluntary
• 16:22
  non -binding standards for child safety. That
• 16:25
  feels like a massive disconnect in how governments
• 16:27
  are treating this. Are they treating this as
• 16:28
  a hard, immediate threat or just making soft
• 16:31
  suggestions? You've zeroed in on the exact contradiction
• 16:34
  that defines this current era of tech regulation.
• 16:36
  We are watching a split reality play out. out
• 16:39
  in real time. On one side, when it comes to international
• 16:42
  diplomacy and trying to write a global rulebook,
• 16:45
  you have these soft emerging standards. Governments
• 16:49
  at the G7 level are moving slowly, relying on
• 16:51
  voluntary commitments because achieving international
• 16:54
  consensus on hard law for a technology evolving
• 16:58
  this rapidly is practically impossible. Right.
• 17:00
  Diplomacy is a slow machine. Extremely slow.
• 17:03
  So they settle for handshakes and diplomatic
• 17:05
  press releases while they try to figure it out.
• 17:07
  Meanwhile, the tech itself isn't waiting around
• 17:09
  for them. Exactly. On the other side of the equation,
• 17:12
  the physical reality of this technology is moving
• 17:15
  at light speed, and it is anything but soft.
• 17:17
  To run these frontier models, you need immense
• 17:20
  computing power. You need hardware. Heavy hardware.
• 17:23
  The source mentions that right around the time
• 17:25
  the G7 was agreeing to voluntary standards, SoftBank
• 17:29
  pledged around 45 billion euros to build AI data
• 17:32
  centers in France. 45 billion euros just for
• 17:36
  data centers? Yes. This is what we call the capex
• 17:38
  arms race capital expenditure. While the global
• 17:41
  regulatory rules remain soft and voluntary, the
• 17:44
  financial and structural reality of the AI industry
• 17:47
  is cementing itself into the ground. 45 billion
• 17:50
  euros isn't buying software. It is buying massive
• 17:54
  tracts of land, securing water rights for cooling
• 17:56
  systems, laying thousands of miles of high -voltage
• 17:59
  power grids, and installing specialized silicon
• 18:02
  chips. It's incredibly physical. Yes. The governments
• 18:06
  might be moving slowly with regulations, but
• 18:08
  the infrastructure required to run these virtual
• 18:10
  tools is becoming a permanent, massive footprint
• 18:13
  on the physical world. It really brings into
• 18:16
  focus how complicated your world is getting if
• 18:18
  you are a business leader trying to build on
• 18:21
  top of these systems. It's a minefield. It is.
• 18:23
  To summarize the massive arc we've traveled today
• 18:25
  for you, the listener, the actual tools your
• 18:28
  developers are using to build your company's
• 18:30
  future are rapidly shrinking into the hands of
• 18:32
  a few mega giants like SpaceX and Microsoft,
• 18:34
  and that is severely limiting your leverage.
• 18:37
  Yes. At the exact same time, governments are
• 18:39
  realizing they have the power to instantly pull
• 18:42
  the plug on those load -bearing AI models. under
• 18:45
  the banner of national security. The kill switch.
• 18:47
  The kill switch, right. And yet... But despite
• 18:50
  all this high level corporate consolidation and
• 18:52
  sovereign power, your entire massive expensive
• 18:56
  operation can still be completely brought down
• 18:59
  by a single forgotten digital valet key left
• 19:03
  open by a third party vendor. It just demands
• 19:05
  a level of continuous vigilance and strategic
• 19:07
  planning from business leaders that simply wasn't
• 19:10
  required a decade ago when software lived on
• 19:13
  a disk on your desk. No, not at all. So our call
• 19:16
  to action for you enthusiastically reminds you
• 19:18
  to stay patched. Stay deeply skeptical of what
• 19:20
  you're plugging into your network and go check
• 19:23
  those OAuth tokens. Digital hygiene is just as
• 19:26
  important as the strength of your firewalls.
• 19:28
  Absolutely. And to leave you with one final thing
• 19:30
  to mull over. Building on all the source material.
• 19:33
  He's due. We've seen how the entire digital infrastructure,
• 19:35
  from the code your developers write to the 45
• 19:38
  billion euro data centers powering it, is consolidating
• 19:42
  into just three or four mega corporations. Yeah.
• 19:44
  If these entities now hold the keys to tools
• 19:47
  that can single -handedly trigger global national
• 19:49
  security emergencies, are we moving toward a
• 19:52
  future where these tech giants essentially have
• 19:54
  to act and be treated as nation states themselves?
• 19:58
  Man. When the company's laying the digital tracks
• 20:01
  have the capital and the power of a country,
• 20:03
  you really have to wonder what kind of world
• 20:05
  we are all renting space in. Thanks for joining
• 20:08
  us on this deep dive. Keep questioning the systems
• 20:11
  you rely on, and we will catch you next time.