Latest / Elon Musk Podcast / Anthropic Hits One Trillion Overtaking OpenAI

Transcript

Anthropic Hits One Trillion Overtaking OpenAI · Elon Musk Podcast

• 0:00
  Anthropic's implied valuation on private secondary markets,
• 0:03
  specifically on platforms like Forge Global, has officially
• 0:06
  touched $1 trillion, overtaking Open AI.
• 0:10
  Right. And that valuation, it comes
• 0:13
  entirely from secondary market trading.
• 0:16
  OK. So for anyone trying to grasp
• 0:18
  the mechanics here, existing shareholders are basically
• 0:21
  selling their private stakes to new, usually institutional,
• 0:25
  buyers. Yeah.
• 0:26
  So it's not a direct investment into the company.
• 0:28
  Exactly. This is fundamentally different
• 0:30
  from a primary funding round where the company itself
• 0:33
  actually issues new shares to raise capital.
• 0:35
  We're going to look at how a company focused purely on
• 0:37
  business software completely outpaced the creator of the most
• 0:41
  famous consumer artificial intelligence allocation, right?
• 0:44
  And we will also exlore the unique mechanics driving these
• 0:47
  private trades, plus the really unconventional ways everyday
• 0:50
  investors are trying to get a piece of the action.
• 0:53
  So how does a business with a fraction of the consumer user
• 0:55
  base justify a higher market value and generate more revenue
• 0:58
  than its biggest rival? Well, Anthropic shares are
• 1:01
  trading at that implied $1 trillion on Forge Global right
• 1:05
  now. Meanwhile, Open AI is trading
• 1:08
  lower, basically lingering near its last official primary
• 1:12
  funding valuation. The demand for Anthropic is, I
• 1:16
  mean it is so intense that buyers are literally offering
• 1:20
  physical real estate as collateral just to secure
• 1:23
  shares. Wow, real estate.
• 1:24
  Yeah, you have these massive institutional investors and
• 1:28
  highly capitalized family offices, and they are just
• 1:32
  scrambling to acquire equity from absolutely anyone willing
• 1:35
  to sell. Right, they just want in.
• 1:36
  Exactly, they are leveraging hard physical assets to make the
• 1:41
  math work for these secondary brokers.
• 1:43
  It is a completely aggressive posture from the buying side.
• 1:46
  And that you know, that creates a severe supply and demand
• 1:49
  imbalance in the market. You have to understand the
• 1:51
  psychology and the financial reality of the employees and
• 1:54
  early investors who are actually holding these shares.
• 1:56
  Yeah, because they're on the inside.
• 1:58
  Exactly. They look at their internal
• 1:59
  dashboards, they see the underlying revenue growth
• 2:02
  trajectory, and because they see that data, they fundamentally
• 2:05
  refuse to sell. I mean, why would you, right?
• 2:07
  Right. So that creates a market with
• 2:09
  virtually 0 active sellers, but an endless line of buyers
• 2:14
  holding giant checks. When a broker does manage to
• 2:17
  find like a a block of shares, they are inundated with
• 2:21
  competing offers almost instantly.
• 2:23
  The bidding wars are frantic. But then you look at open AI
• 2:27
  secondary market and it's experiencing the exact opposite
• 2:30
  sentiment. Oh totally.
• 2:32
  They have seen a ratio of five sellers to everyone buyer. 5:00
• 2:36
  to 1:00. Yeah, 5 sellers for every buyer.
• 2:39
  That creates a very tepid market for their shares.
• 2:42
  Bids are coming in below their last official valuation, but
• 2:45
  Entropic shares are purchased within hours of being listed on
• 2:48
  these exact same platforms. You really to wonder about the
• 2:50
  mechanics of how these platforms even function when the demand is
• 2:53
  that skewed. Think about a rare art auction.
• 2:57
  Imagine the few owners of pieces by a highly coveted artist
• 3:01
  simply refused to part with. Them right, they just lock them
• 3:03
  in a vault. Exactly, they will not sell
• 3:05
  under any circumstances. That refusal artificially Dr.
• 3:08
  the perceived price up for the tiny fraction of pieces that do,
• 3:11
  you know, eventually become available.
• 3:13
  Because everyone wants the one piece that's out there.
• 3:15
  Yeah, the scarcity itself becomes the primary driver of
• 3:18
  the premium. When there was practically no
• 3:21
  floating supply of an asset, the marginal buyer, the one person
• 3:26
  willing to pay the absolute highest price, they set the new
• 3:29
  price floor for the entire asset class.
• 3:31
  I am curious though, how much of this is based on real
• 3:35
  technological superiority versus just extreme fear of missing
• 3:39
  out? That's the big question.
• 3:40
  Right, because you have venture capital funds who feel they
• 3:43
  absolutely need exposure to the hottest artificial intelligence
• 3:46
  company just to satisfy their own limited partners.
• 3:49
  The motivation and certainly blends both genuine
• 3:51
  technological belief and intense financial pressure, but the
• 3:55
  confidence of this behavior is completely undeniable.
• 3:58
  This frenzy really alters the perception of leadership in the
• 4:02
  artificial intelligence race. It shifts the focus.
• 4:04
  Exactly. The private market is heavily
• 4:07
  rewarding enterprise momentum over consumer brand recognition.
• 4:12
  Institutional capital is sending a clear signal that deep
• 4:15
  business to business integration is fundamentally more valuable
• 4:19
  than having a household name chatbot that everyday people use
• 4:22
  for trivia. I think we need to step back for
• 4:24
  a second. We are talking about private A
• 4:27
  liquid shares here, not public stock.
• 4:29
  You can just trade from your phone.
• 4:30
  True, but looking at the numbers, Anthropic surpassed
• 4:34
  Open AI and annualized revenue run rate hitting $30 billion
• 4:38
  compared to Open AI $24 billion. That revenue flip happened
• 4:42
  incredibly fast, and the underlying economics explain
• 4:45
  exactly why Anthropic monetizes at roughly $211.00 per user.
• 4:50
  Wait 211. Yeah, per user.
• 4:52
  Compare that to Open AI's $25 per user.
• 4:56
  And beyond that, Anthropic boasts over 500 individual
• 4:59
  customers who are spending more than $1 million annually.
• 5:02
  Oh. Wow.
• 5:02
  So it's heavily enterprise. Exactly.
• 5:04
  They are securing massive long term enterprise contracts.
• 5:09
  Those contracts guarantee consistent capital inflows
• 5:12
  rather than just hoping consumers renew a monthly
• 5:15
  subscription. Wait, back up.
• 5:17
  How does a software company jump from 9 billion to 30 billion in
• 5:20
  annualized revenue in a single quarter?
• 5:23
  That math is wild. That gross is driven almost
• 5:27
  entirely by high value enterprise API contracts and
• 5:31
  intense developer adoption. They're not relying on
• 5:34
  individuals paying $20.00 a month to ask a chat bot to write
• 5:37
  a recipe or, you know, draft a polite e-mail to their boss.
• 5:40
  Right, the consumer stuff. Yeah, they are integrating their
• 5:42
  systems directly into the back end workflows.
• 5:45
  The Fortune 10. Imagine you were running a
• 5:47
  global logistics company. You do not buy a chat bot, you
• 5:51
  purchase guaranteed server throughput.
• 5:53
  Because you need reliability. Exactly.
• 5:55
  You pay per token to process massive, endless volumes of
• 5:59
  proprietary corporate data securely within your own
• 6:01
  firewalls. Every single time your internal
• 6:04
  software makes a routing decision, it calls their API and
• 6:07
  they charge you a fraction of a cent.
• 6:08
  Multiply that by billions of operations.
• 6:11
  Picture the difference between selling a fleet of commercial
• 6:13
  jets to International Airlines versus selling millions of
• 6:17
  bicycles to individual commuters.
• 6:19
  That's a great way to put it, right?
• 6:20
  The absolute volume of individual users matters
• 6:23
  significantly less than the contract size and the stickiness
• 6:26
  of the product within a corporate infrastructure.
• 6:29
  A business user utilizing an API for their core supply chain
• 6:32
  operations has a significantly higher lifetime value and a much
• 6:37
  lower churn rate than a casual consumer testing out
• 6:39
  conversational software. This proves consumer virality is
• 6:43
  absolutely not the only path to massive software revenue.
• 6:47
  It fully validates the business to business enterprise model as
• 6:50
  the most lucrative route in this entire sector.
• 6:52
  Yeah, definitely. Getting to 100 million users the
• 6:55
  fastest was a phenomenal headline for consumer apps.
• 6:58
  But deeply embedding your models into the software that actually
• 7:01
  runs global finance, healthcare systems and logistics networks?
• 7:05
  Well, that is what generates $30 billion in highly reliable
• 7:08
  recurring revenue. And you know, a specific
• 7:11
  developer tool called Claude Code is generating 2 1/2 billion
• 7:15
  dollars in annualized revenue all by itself.
• 7:17
  Just that one tool. Just that one.
• 7:20
  That ejectic coding tool is now authoring a significant
• 7:23
  percentage of all public code commits on GitHub, with internal
• 7:27
  projections showing it could handle 1/5 of all commits
• 7:29
  globally. 1/5 That is unbelievable.
• 7:33
  We are looking at a single developer product that scaled
• 7:35
  from zero to multibillion dollar revenue in a matter of months.
• 7:40
  Entire engineering teams at major corporations are utilizing
• 7:43
  this tool to automate their routine coding tasks, completely
• 7:47
  debug complex legacy systems, and heavily accelerate their
• 7:51
  entire software deployment pipeline.
• 7:53
  See, I look at that and see a massive concentration risk.
• 7:56
  Really. How so?
• 7:57
  While relying so heavily on a single product category like
• 8:00
  coding assistance makes the overall company highly
• 8:03
  vulnerable, if well funded competitors release highly
• 8:07
  capable open source alternatives, that entire 2 1/2
• 8:10
  billion dollar revenue stream could evaporate overnight.
• 8:13
  You think they would switch that fast?
• 8:14
  Absolutely. An enterprise client will
• 8:16
  happily switch to a free or heavily discounted alternative
• 8:20
  if the performance gap between the tools narrows.
• 8:23
  Code generation is actively becoming commoditized.
• 8:26
  I disagree completely with that assessment.
• 8:28
  Developer tools have the highest switching costs in the entire
• 8:32
  software industry. But if it's cheaper.
• 8:34
  Doesn't matter. Once you have a development team
• 8:36
  integrate an agent decoder into their daily deployment cycle, it
• 8:40
  becomes entirely entrenched. You establish your core security
• 8:43
  protocols around this specific tool.
• 8:45
  You spend months trading your software engineers on its highly
• 8:48
  specific nuances and prompting structures.
• 8:51
  I guess the training does take time.
• 8:53
  It takes a lot of time. It creates the deepest possible
• 8:55
  entrenchment in enterprise workflows.
• 8:58
  Removing that tool disrupts your entire engineering pipeline and
• 9:02
  completely kills internal productivity.
• 9:04
  No chief technology officer will risk halting their product road
• 9:07
  map just to save a few $1,000,000 on software licenses.
• 9:10
  That dynamic absolutely shifts the focus of modern software
• 9:14
  development. Then developer focused tools are
• 9:17
  actively establishing themselves as the most profitable layer of
• 9:21
  the new software stack. Exactly.
• 9:23
  This opens up entirely new categories of enterprise
• 9:26
  spending. You are seeing chief technology
• 9:29
  officers aggressively reallocating massive budgets
• 9:32
  from other departments just to equip their engineers with these
• 9:36
  specific capabilities. The return on investment in pure
• 9:39
  labor efficiency is undeniable. If a tool makes your engineering
• 9:42
  team 20% faster, you pay whatever the vendor ask.
• 9:46
  And here's another wild Stat Anthropic project spending four
• 9:49
  times less on model training than open AI.
• 9:51
  Four times less. Yeah, Open AI is projected to
• 9:54
  spend over $120 billion on compute over the coming years,
• 9:58
  while Anthropic projects their spend at around 30 billion.
• 10:01
  That is a massive gap. It is, and because of this
• 10:04
  disciplined spending and highly focused engineering strategy,
• 10:08
  anthropic projects reaching actual profitability much
• 10:11
  sooner. They are optimizing their model
• 10:13
  architectures to achieve top tier reasoning capabilities
• 10:17
  without requiring the same brute force computational power that
• 10:20
  their competitors rely on. Hold on, does spending less
• 10:24
  naturally mean their models will eventually fall behind in
• 10:26
  performance? Not necessarily.
• 10:29
  I mean, the entire industry currently assumes pure size and
• 10:32
  scale are the only mathematical ways to achieve advanced
• 10:36
  intelligence. Cutting costs on training runs
• 10:39
  seems like a guaranteed recipe for losing the capability race
• 10:42
  in the long run. Anthropic is aggressively
• 10:44
  testing the assumption that the company spending the absolute
• 10:46
  most on training automatically wins the market.
• 10:49
  So they think there's a smarter way.
• 10:51
  Right. They are betting heavily that
• 10:53
  algorithmic efficiency, high quality synthetic data
• 10:56
  generation, and highly specialized model routing matter
• 11:00
  significantly more than just raw compute volume.
• 11:04
  They believe they can match or exceed competitor performance
• 11:07
  through smarter, highly targeted engineering rather than just
• 11:12
  buying more graphics processing units and plugging them into the
• 11:15
  wall. I am still highly skeptical of
• 11:17
  the math on the other side of that equation.
• 11:20
  You have to wonder if these companies are just throwing
• 11:22
  billions of dollars into a bottomless pit.
• 11:25
  It definitely looks like that sometimes.
• 11:26
  Building an artificial intelligence model right now
• 11:29
  feels exactly like building a massive, highly expensive
• 11:32
  highway system before knowing exactly how how many cars will
• 11:35
  actually use it. You commit 10s of billions of
• 11:37
  dollars to massive data centers, cooling systems, and specialized
• 11:41
  silicon chips just hoping the resulting model produces enough
• 11:44
  commercial value to justify the massive capital expenditure.
• 11:47
  That changes how institutional investors evaluate these
• 11:50
  companies entirely, though. Financial success is
• 11:53
  increasingly measured by who can generate the most revenue per
• 11:56
  single dollar of training spend, rather than who simply builds
• 12:01
  the biggest underlying model with the most parameters.
• 12:04
  It's all about efficiency now. Think about the pure math of it.
• 12:07
  If you spend $100 billion just to train a single model, your
• 12:11
  revenue requirements to reach free cash flow are
• 12:14
  astronomically high. Capital efficiency is rapidly
• 12:17
  becoming the defining metric for long term survival in this
• 12:20
  specific sector. Let's talk about where that
• 12:22
  money is coming from. Major technology giants provide
• 12:26
  enormous funding to artificial intelligence startups, who then
• 12:29
  turn around and use that exact same money to buy cloud
• 12:32
  computing services from those exact same tech giants.
• 12:35
  It's quite the loop. Amazon and Google have made
• 12:37
  multibillion dollar investments directly into Empropic.
• 12:40
  And that creates a highly constrained circular financing
• 12:43
  loop. Amazon invest billions of
• 12:45
  dollars into the startups bank account, but Anthropic
• 12:48
  explicitly commits to using Amazon Web Services for its
• 12:51
  massive compute needs to train and serve its models.
• 12:55
  So the money just goes right back.
• 12:56
  Yeah, the capital effectively never leaves the corporate
• 12:59
  ecosystem. It is simply reclassified from
• 13:02
  an investment asset on the balance sheet to cloud computing
• 13:06
  revenue on the income statement. Big Tech gives them money and
• 13:09
  they hand that money right back to big tech to rent servers.
• 13:13
  It's that simple. It operates exactly like a
• 13:15
  commercial landlord, giving a tenant a large cash loan
• 13:19
  specifically so the tenant can pay the monthly rent directly
• 13:22
  back to the landlord. That sounds incredibly fragile.
• 13:25
  You have to strongly question the long term health of this
• 13:28
  financial dynamic. If the underlying enterprise
• 13:31
  return on investment does not materialize for the actual end
• 13:34
  user buying the software, this entire closed loop could
• 13:37
  collapse entirely. The major cloud providers are
• 13:40
  heavily subsidizing the artificial intelligence boom
• 13:43
  specifically to drive their own infrastructure growth metrics
• 13:46
  for Wall Street. This arrangement completely
• 13:49
  limits the true independence of these artificial intelligence
• 13:52
  startups, but I guess it absolutely ensures their short
• 13:55
  term survival. They couldn't survive without
• 13:57
  it. It guarantees immediate access
• 13:59
  to crucial computing infrastructure.
• 14:02
  Without these highly strategic partnerships, securing the
• 14:05
  necessary gigawatts of electrical power and hundreds of
• 14:08
  thousands of specialized chips would be mathematically
• 14:11
  impossible for an independent startup.
• 14:13
  They are trading complete autonomy for access to the only
• 14:16
  machines capable of running their software.
• 14:18
  Now here's a really strange angle.
• 14:20
  Zoom Video Communications owns a stake in Anthropic that could be
• 14:24
  worth up to $10 billion. Wait, Zoom, the video call
• 14:28
  company. Yeah, Zoom, they made an early
• 14:31
  $51 million investment directly into Anthropic.
• 14:34
  Even factoring in the heavy share dilution from subsequent
• 14:37
  massive funding rounds, a 1% stake in a trillion dollar
• 14:40
  company mathematically equals $10 billion.
• 14:43
  That is wild. Buying Zoom stock right now
• 14:47
  operates as a brilliant, highly asymmetric bet on Anthropics
• 14:50
  future success. Zoom is currently sitting on
• 14:53
  nearly $8 billion in cash and short term investments with
• 14:56
  absolutely zero debt. So it's like a backdoor.
• 14:59
  Exactly. If you buy Zoom equity, you are
• 15:02
  effectively getting Zoom Score video conferencing business
• 15:06
  completely for free, simply given their cash reserves and
• 15:10
  the enormous paper value of that artificial intelligence equity I
• 15:13
  have. To push back on the idea that
• 15:14
  this is a safe investment vehicle, though, yeah, Zoom
• 15:17
  Score business faces extreme competition and steadily
• 15:20
  declining growth. Sure, but the cash buffer?
• 15:23
  But think about it. The permanent shift in hybrid
• 15:26
  work models combined with intense pressure from bundled
• 15:29
  enterprise software solutions like Microsoft Teams puts heavy
• 15:33
  downward pressure on their enterprise retention metrics.
• 15:36
  That is a fairpoint that makes it a highly risky vehicle just
• 15:39
  to hold an artificial intelligence lottery ticket.
• 15:41
  If Zoom's enterprise market share arose further, the equity
• 15:45
  value of the core business will drag down the entire stock price
• 15:48
  completely, regardless of what the underlying anthropic stake
• 15:52
  is actually worth. I see what you mean, but this
• 15:54
  opens up a highly creative strategy for everyday retail
• 15:56
  investors to gain exposure to private market darlings.
• 16:00
  They do not need the legal accreditation or the massive
• 16:03
  capital required to buy private shares directly.
• 16:05
  They just buy Zoom instead, right?
• 16:07
  Retail investors are actively hunting for publicly traded
• 16:10
  proxies to capture the financial upside of companies that remain
• 16:14
  strictly locked behind private venture capital doors.
• 16:18
  They are looking for any backdoor entry into the wealth
• 16:21
  creation event. And some are going even further.
• 16:24
  Tokenized Anthropic shares are actively trading on
• 16:27
  cryptocurrency platforms, implying an $850 billion
• 16:31
  valuation. The crypto angle is fascinating.
• 16:34
  These highly experimental decentralized pre stocks trade
• 16:37
  on networks like Solana, theoretically allowing everyday
• 16:40
  retail participants to speculate directly on the private market
• 16:43
  valuation. But there is a critical
• 16:45
  structural flaw in that specific trading system.
• 16:49
  A retail trader made over $1 million in pure paper profit
• 16:53
  trading these tokens, but they cannot actually cash out due to
• 16:57
  a severe lack of market liquidity.
• 16:58
  Wait so they have $1,000,000 on screen but can't touch it?
• 17:01
  Exactly when they actively attempted to simulate selling
• 17:04
  their position across various decentralized exchanges, the
• 17:08
  financial slippage was massive. There were simply no actual
• 17:11
  buyers sitting on the other side of the trade at that specific
• 17:14
  price point. Wait, let me understand this
• 17:16
  fully. People are purchasing digital
• 17:18
  tokens that represent private shares, but they do not actually
• 17:22
  own the underlying shares at all.
• 17:24
  Right. They are purely bearer digital
• 17:26
  assets. They carry absolutely no legal
• 17:28
  voting rights. They provide 0 direct ownership
• 17:31
  in the company itself. So what are?
• 17:33
  They they are highly complex synthetic instruments explicitly
• 17:36
  designed just to track the price of the underlying private asset.
• 17:40
  They are managed by highly obscure special purpose vehicles
• 17:44
  that hold the actual legal paperwork offline.
• 17:46
  You are basically holding a winning lottery ticket in a
• 17:49
  jurisdiction where you cannot legally claim the financial
• 17:52
  prize. The numbers look phenomenal on a
• 17:54
  digital screen and the implied valuation creates highly
• 17:57
  clickable headlines, but the actual real world financial
• 18:01
  value is effectively 0 if you cannot mathematically convert
• 18:04
  those digital tokens back into usable Fiat currency.
• 18:07
  Yeah, the liquidity trap is absolute.
• 18:10
  This completely limits the actual value of the synthetic
• 18:13
  assets. It heavily highlights the
• 18:15
  extreme, highly risky lengths everyday retail investors will
• 18:19
  go to for any exposure to the artificial intelligence
• 18:22
  financial boom. People just want in so badly.
• 18:25
  The pure desperation to participate in this specific
• 18:28
  wealth creation event is driving retail capital into highly
• 18:31
  experimental and intensely illiquid financial structures.
• 18:35
  People are buying complex derivatives just to feel like
• 18:38
  they are participating. And all this pressure is
• 18:41
  building up to something both Open AI and Anthropic are
• 18:45
  aggressively preparing for initial public offerings heavily
• 18:48
  driven by competitive market dynamics and mounting employee
• 18:52
  stock options. There is an intense internal
• 18:54
  pressure cooker building inside these exact companies.
• 18:58
  Thousands of highly paid engineers and researchers are
• 19:00
  holding illiquid stock options that strictly vest over a four
• 19:03
  year period, the standard one year Cliff.
• 19:06
  And they want their money. Of course they do.
• 19:08
  These individuals have generated billions in theoretical paper
• 19:12
  value. They naturally want to realize
• 19:14
  the actual financial gains of their incredibly hard work to
• 19:17
  buy homes and permanently secure their personal financial
• 19:20
  futures. Investment banks predict these
• 19:22
  eventual public market offerings will heavily exceed $60 billion
• 19:27
  in raised capital. The scramble among major global
• 19:30
  financial institutions to secure the rights to underwrite these
• 19:33
  listings is intensely competitive.
• 19:36
  I fully believe the public markets will eagerly absorb
• 19:39
  these specific offerings simply due to the pure revenue growth
• 19:42
  velocity we discussed earlier. You think so?
• 19:44
  Yeah. When massive mutual funds,
• 19:47
  global pension funds and everyday retail investors
• 19:49
  finally secure a direct mechanism to buy the companies
• 19:52
  actively defining the next computing platform, the capital
• 19:55
  inflow will be absolutely staggering.
• 19:58
  The public market is entirely starved for pureplay artificial
• 20:01
  intelligence infrastructure stocks.
• 20:03
  I completely disagree with that outlook because public markets
• 20:06
  strictly demand clear, mathematically sound paths to
• 20:10
  free cash flow. Wall Street analysts will
• 20:13
  absolutely not stomach revenue multiples of 27 times or higher
• 20:17
  for companies that are actively burning billions of dollars
• 20:20
  every single month on raw server costs.
• 20:23
  But the growth is unprecedented. The transition from private
• 20:27
  venture markets to public markets involves a highly brutal
• 20:30
  shift in financial valuation metrics.
• 20:32
  Private venture capitalists value exponential top line user
• 20:36
  growth above all else. Public market analysts strictly
• 20:40
  demand highly sustainable operating margins and incredibly
• 20:43
  clear profitability timelines. This sets up the absolute
• 20:47
  ultimate test for the entire artificial intelligence
• 20:49
  industry. Then the harsh transition from
• 20:52
  private venture capital logic to strict public market scrutiny
• 20:56
  will completely force these specific companies to legally
• 20:59
  prove their underlying business models are highly sustainable.
• 21:02
  Not to show the math works. They will have to mathematically
• 21:04
  demonstrate that they can sell digital intelligence to
• 21:06
  enterprise clients at a strictly higher margin than it natively
• 21:10
  cost to generate it on their hardware.
• 21:12
  So tying this all together, the competition between artificial
• 21:15
  intelligence leaders has entirely shifted from consumer
• 21:19
  popularity contest to a ruthless focus on enterprise revenue
• 21:22
  efficiency, strategic cloud partnerships, and intense
• 21:26
  private market maneuvering. If the physical hardware and
• 21:29
  daily compute costs continue to rise exponentially, will these
• 21:33
  trillion dollar valuations hold up when the public markets
• 21:36
  finally demand actual sustainable profit margins?
• 21:39
  If you're not subscribed yet, take a second and hit follow on
• 21:42
  whatever app you're using. It helps us keep making this.
• 21:45
  We appreciate you being here. Also, check out our YouTube
• 21:47
  channel for more business and tech updates.
• 21:49
  There's a link in the description.