Latest / Deep Values: Value Investing Deep Dives / Anthropic Business Deep Value Analysis: Valuation, Growth, and Risks
Transcript
- 0:00Welcome back to the Deep. Dive we're.
- 0:02Here to cut through the noise and get straight to what
- 0:03matters. And today we're diving into
- 0:06something really interesting. Anthropic.
- 0:07It's this AI startup growing incredibly fast, but it's
- 0:10private, so you know, a lot happens behind closed doors.
- 0:13Now you send us this Anthropic financial research report, which
- 0:16is fantastic. Thank you.
- 0:17Our job today? Pull out the key insights,
- 0:20figure out where Anthropic really sits in this booming AI
- 0:22landscape, and see what we can learn from a company showing
- 0:25just, well, explosive growth even without all the usual
- 0:29public financial details. It's a bit of a puzzle, right?
- 0:32How do you make sense of it? OK, let's unpack.
- 0:34This and when I say. Growth, I really mean it.
- 0:36The report kicks off with some frankly eye popping numbers.
- 0:40We're talking annualized sales jumping from maybe $1 billion
- 0:43late last year, 2024. To.
- 0:45Potentially $3 billion by what may 2025 and some estimates are
- 0:51even higher, maybe 4 billion, even $5 billion annualized by
- 0:54this summer. That's.
- 0:56Well, that's not just fast growth.
- 0:57It's like warp speed compared to most tech companies.
- 1:00It really is, and it tells you a lot about just how hungry
- 1:03businesses are for advanced AI right now.
- 1:06Anthropic's catching a huge wave.
- 1:08Basically, their core business is selling AI services.
- 1:11Specifically, they're clawed large language models, LLMS.
- 1:14They sell mostly to enterprises and developers.
- 1:19So how does that work in practice?
- 1:20Like what are the main ways they make money?
- 1:22It's pretty focused. The main revenue streams are API
- 1:26access that's letting other businesses plug Anthropics AI
- 1:30into their own software and then subscriptions for using the
- 1:32models. And often this happens through
- 1:34cloud platforms. You know AWS, Google Cloud makes
- 1:37it easier for companies to integrate, right?
- 1:39And the report lists some pretty big names using it.
- 1:42It's not just startups tinkering, is it?
- 1:43We see the European Parliament, Pfizer, United Airlines, Thomson
- 1:47Reuters, these are major organizations.
- 1:49Exactly. Plus software companies like
- 1:51Cursor and GitHub Copilot. The report even says coding
- 1:55platforms have brought in hundreds of millions for them.
- 1:58Yeah, that's a crucial point. It shows these aren't just small
- 2:00tests. These are deep integrations and
- 2:03the variety of clients, government, pharma, airline
- 2:06software, it tells you Claude is pretty versatile, pretty robust.
- 2:10And what about the split between business and, say, regular
- 2:13consumers using a chat bot? That's really telling.
- 2:18In the report. It seems something like 85% of
- 2:21their revenue, it comes from those business customers, the
- 2:24API users, the consumer side, you know, people subscribing to
- 2:27a chat bot that's much smaller, maybe around 15%.
- 2:30Wow, 85% enterprise. Yeah.
- 2:32And it also mentions a big jump in large deals, like they
- 2:35tripled the number of eight and nine figure contracts signed
- 2:38this year compared to all of last year.
- 2:40So it's not thousands of small customers, it looks like a
- 2:42smaller base of really big committed enterprise clients
- 2:45driving most of the that. Concentration is fascinating.
- 2:48It suggests businesses are really going all in on AI now,
- 2:52not just testing. It's a strong signal about
- 2:54Anthropics perceived value and maybe where the whole enterprise
- 2:58software market is heading. OK, here's where it gets really
- 3:00interesting. What does this tell us about the
- 3:02wider adoption curve? It suggests we're past the early
- 3:06adopter phase. For many large organizations,
- 3:10they see a clear ROI, or at least a critical strategic need
- 3:14for this level of AI. But even a rocket ship like this
- 3:17faces headwinds. Let's pivot to the headwinds and
- 3:20risks section of the report. No company has a perfectly clear
- 3:24sky ahead. Absolutely not, and the short
- 3:27term picture has some serious challenges.
- 3:29Competition mainly. The report specifically calls
- 3:32out Open AI launching GPT 5, apparently with much lower
- 3:36pricing. That puts immediate pressure on
- 3:38Entropic. How so?
- 3:40Because Anthropic is seen as more premium price.
- 3:42Exactly, if Open AI offers something comparable or good
- 3:46enough for certain tasks at a lower cost, enterprise customers
- 3:49are definitely going to look at their contracts again.
- 3:50You could see a pricing war starting which would squeeze
- 3:53Anthropics growth and definitely their margins.
- 3:56Makes sense. What other short term risks does
- 3:59the report mention? I saw something about economic
- 4:01cycles potentially hitting tech budgets and that customer
- 4:04concentration risk we touched on.
- 4:06Yes, both are significant. If the economy tightens, those
- 4:09big AI projects might get scaled back or companies will negotiate
- 4:13harder on price and the concentration.
- 4:15The report suggests just two coding partners make up about
- 4:181/4 of their revenue. That's, well, that's a
- 4:20vulnerability. 1/4 from just two partners.
- 4:23Wow, yeah. It's great while it lasts, shows
- 4:25deep partnerships, but if one leaves or cuts back, that's a
- 4:28big hit. OK.
- 4:29So those are near term, What about longer term risks?
- 4:33The report outlines several. One is the rise of open source
- 4:36AI models. They're getting better fast.
- 4:39If open source becomes good enough and more customizable,
- 4:42some companies might just build their own solutions instead of
- 4:45paying Anthropic. Interesting.
- 4:47Then there's AI regulation. It's coming globally.
- 4:50Anthropic being a public benefit company, which means they
- 4:54legally have to balance profit with societal good.
- 4:57That could be an advantage. Maybe builds trust?
- 5:00Or a disadvantage? Could it slow them down?
- 5:02Potentially yes, more compliance, maybe slower
- 5:06development if safety checks take longer than competitors.
- 5:09It's a double edged sword. Plus you can't forget the tech
- 5:12giants Google, Microsoft. They have deep pockets and can
- 5:16bundle their own AI into everything.
- 5:18That's huge competitive pressure.
- 5:20And the talent war, I assume. Oh, absolutely.
- 5:22It's a global arms race for AI researchers.
- 5:25Losing key people could really hurt their innovation.
- 5:28And finally, there are macro things like just getting enough
- 5:31computing power, GPUs, Tpus, and the energy required.
- 5:34Costs could go up, or regulations could limit how much
- 5:37power they can even use for training these massive models.
- 5:40It's a complex web of challenges, intense competition,
- 5:44regulatory uncertainty, resource constraints.
- 5:47What's fascinating here is how they navigate all this while
- 5:49trying to maintain that incredible growth.
- 5:51It really is. It forces them to be incredibly
- 5:54strategic about innovation, pricing, partnerships,
- 5:57everything really. They have to keep proving their
- 5:59value proposition is worth the premium and maybe that their
- 6:01safety focus is a long term win. All right, we've seen the growth
- 6:04in the potential storms. Let's look inside the engine
- 6:06room. What powers this thing we need
- 6:08to talk about costs, management, governance, the internal
- 6:12machinery, OK. Well for any major AI player the
- 6:16biggest costs are pretty clear compute and people training and
- 6:21running models like Claude needs massive amounts of specialized
- 6:24computing power. Think huge server farms full of
- 6:27GP US or TP US and that uses just enormous amounts of
- 6:32electricity. The report mentioned a big
- 6:34Amazon deal. Right, a multibillion dollar
- 6:37investment deal. Part of that involved Anthropic
- 6:39committing to use Amazon's own specialized AI chips.
- 6:42That's crucial, locking in access to that compute power.
- 6:45So compute infrastructure is #1 what's #2?
- 6:47People, the cost of hiring and keeping top AI researchers and
- 6:50engineers is incredibly high. There's just fierce global
- 6:54competition for that talent. So R&D personnel, cloud compute,
- 6:57those are the big ticket items. Makes sense.
- 6:59It's fundamentally a capital intensive game.
- 7:01Who's steering the ship? The report mentions the
- 7:03founders. Yes, Dario Amode as CEO and
- 7:07Daniella Amode as President. They're siblings and
- 7:10interestingly, both came from Open AI.
- 7:12And they recently added Reed Hastings to the board, the
- 7:15Netflix Co founder. That feels significant.
- 7:18Very significant. Hastings brings, you know,
- 7:20immense experience in scaling disruptive tech, navigating
- 7:23competition, building subscription businesses.
- 7:26His presence is a major vote of confidence.
- 7:29It suggests they're thinking seriously about long term
- 7:32strategy, market positioning, maybe even future public
- 7:35offerings or major strategic plays.
- 7:38But beyond the big names, the report emphasizes their unusual
- 7:41structure. Right, The public benefit
- 7:43company thing. Exactly.
- 7:44That's not just PR, it's a legal structure.
- 7:47There's this long term benefit trust, the LTBT, made-up of
- 7:50independent trustees. Their job is specifically to
- 7:53make sure the company stays true to its mission, particularly
- 7:56around AI, safety and ethical development, alongside making
- 7:59money. And they've acted on that.
- 8:01Reportedly, yes. The report reminds us they
- 8:03actually held back an early version of cloud for more safety
- 8:06testing. They chose safety over speed,
- 8:08which is pretty telling. That safety focus, baked right
- 8:11into the company DNA, seems like a real differentiator in an
- 8:15industry often criticized for moving too fast.
- 8:18Does the report suggest any downsides to that approach?
- 8:21Maybe it slows them down? It's a trade off for sure.
- 8:23Extra safety steps, more governance.
- 8:26It can mean slower development cycles compared to competitors
- 8:29who are purely profit driven. But they're betting that this
- 8:32builds a long term trust, maybe appeals to certain customers or
- 8:34regulators. It's a strategic choice.
- 8:37And who's funding all this high cost R&D and ethical oversight?
- 8:40The shareholders. Some very big names.
- 8:43The report highlights Amazon committing up to $4 billion,
- 8:46Google committing $2 billion, plus major VC firms like
- 8:49Lightspeed. All that capital is what
- 8:52supported that massive $61.5 billion valuation reported in
- 8:56early 2025. Wow, so the money is definitely
- 8:59there. If we connect this to the bigger
- 9:00picture, that huge influx of cash plus the mission driven
- 9:03leadership, that's what lets them afford the massive R&D and
- 9:06compute bills. It's essential fuel for
- 9:08competing in this AI arms race. Without it, they just couldn't
- 9:11keep up. It gives them the runway they
- 9:12need. OK, we've looked at Anthropic
- 9:14through the lens of a fast growing tech company, but the
- 9:17report includes something different.
- 9:19Buffettstyle checklist. How does a cutting edge private
- 9:23AI firm stack U against traditional value investing
- 9:26principles? Huh.
- 9:27Well, it's a fascinating exercise, but challenging.
- 9:31Buffett looks for things like PE ratios, consistent earnings,
- 9:34free cash flow, book value metrics that mostly don't apply
- 9:38or aren't even available for a private company like Anthropic
- 9:41that's pouring everything into growth, not profit right now.
- 9:44So the valuation is the first obvious mismatch.
- 9:46Absolutely stark. That reported $61.5 billion
- 9:50valuation works out to roughly 44 times their trailing
- 9:53annualized revenue. Buffett usually looks for PE
- 9:56ratios under 15, maybe priced to book under 1.5.
- 9:59Anthropic is playing a completely different game.
- 10:01It's clearly priced for massive future growth, a huge premium.
- 10:05Exactly. The market is betting on future
- 10:07disruption, not current earnings.
- 10:09What other Buffett criteria does anthropic likely miss?
- 10:12Free cash flow is almost certainly negative.
- 10:14They're investing so heavily in R&D compute.
- 10:18Buffett loves companies that generate lots of cash and
- 10:21profitability metrics, margins, return on equity, return on
- 10:25invested capital. We can't measure them
- 10:27accurately, but they're likely low or negative at this stage.
- 10:30Again, the opposite of a mature Buffett style company.
- 10:33Seems like a total non starter for value investing then.
- 10:36Are there any areas when they might align even loosely?
- 10:39Maybe balance sheet safety with billions raised from Amazon,
- 10:43Google and that reported $2.5 billion credit line, they likely
- 10:48have a very strong capital base and probably not much
- 10:51traditional debt relative to their funding.
- 10:54Buffett likes low debt, strong finances.
- 10:56So if we could see the details, they might actually look OK on
- 10:59that specific point, but it's speculative.
- 11:01OK, so mostly mismatch. This raises an important
- 11:04question. If those classic metrics don't
- 11:06work, how do we evaluate a company like this?
- 11:08It's clearly a tech startup growth narrative, not value
- 11:11investing. What should we be looking at
- 11:13instead? That's the key question, isn't
- 11:15it? If the old rules don't apply,
- 11:17what are the new ones for assessing value in this AI
- 11:20space? You have to look at different
- 11:22indicators, things like the strength of their technology,
- 11:25how unique and defensible are their AI models, talent, Can
- 11:29they attract and keep the best researchers, customer metrics,
- 11:32How sticky are those big enterprise contracts?
- 11:35What's the potential lifetime value, market share in key areas
- 11:39and the quality of their strategic partnerships like that
- 11:41Amazon deal for compute? Those become the proxies for
- 11:45long term value. So to wrap up the Buffett
- 11:48comparison, by his traditional standards, Anthropic doesn't fit
- 11:52the mold of a great company today.
- 11:54High valuation, negative cash flow.
- 11:56It's all about future potential. Right, which is exactly what
- 11:59you'd expect for a company at this stage in this kind of
- 12:01revolutionary industry. And that brings us towards the
- 12:03end of our deep dive on Anthropic for today.
- 12:06We've seen the absolutely explosive growth driven by
- 12:09enterprises betting big on AI. We've also touched on the very
- 12:13real risks, intense competition, regulatory hurdles, resource
- 12:17needs. We looked inside of their unique
- 12:19structure as a public benefit company, and we saw how trying
- 12:23to apply traditional financial analysis like Buffett's
- 12:26highlights just how different the valuation game is for these
- 12:29kinds of companies. Now, a quick reminder, all the
- 12:31analysis we discussed is based on publicly available sources.
- 12:34Interpretations can differ, things can change.
- 12:37This is all for educational purposes.
- 12:38It's definitely not investment advice.
- 12:40Always do your own research. Absolutely.
- 12:42Due diligence is key. So here's the final thought for
- 12:45you to consider. And Thriotic's journey isn't
- 12:47just about one company. It feels like a snapshot of the
- 12:50entire AI industry right now. This deep dive into a private AI
- 12:54leader, the incredible speed, the big risks, that sky high
- 12:58valuation, what does it tell you about where technology is
- 13:01heading and maybe how we even define value when industries are
- 13:05being reshaped this quickly definitely gives us a lot to
- 13:07think about. Thanks for joining us on a deep
- 13:09dive. We'll catch you next time.