Latest / Ctrl AI Profit / Ep. 113 | Anthropic Just Turned a Profit — And the AI Race Changed Overnight
Transcript
- 0:20SPEAKER_01: Anthropic just turned a profit.
- 0:22SPEAKER_01: Let that sink in.
- 0:23SPEAKER_01: The AI company that built Claude, the one everyone said was racing toward infinite losses, they just posted their first profitable quarter ever.
- 0:31SPEAKER_01: And it changes everything about how this industry works.
- 0:34SPEAKER_00: We're talking roughly$10.9 billion in revenue for Q2 of this year, with about$559 million in operating profit.
- 0:43SPEAKER_00: That is Anthropic's first ever quarterly profit since they were founded.
- 0:47SPEAKER_00: And the number that really matters, they more than doubled their revenue from the previous quarter.
- 0:52SPEAKER_01: Doubled in one quarter.
- 0:54SPEAKER_01: Let's put that in perspective for the business owners listening.
- 0:57SPEAKER_01: If your business doubled revenue quarter over quarter, you'd be popping champagne and calling Forbes.
- 1:03SPEAKER_01: Anthropic went from$4.8 billion in Q1 to nearly$11 billion in Q2.
- 1:11SPEAKER_01: That is not growth.
- 1:12SPEAKER_01: That is a rocket ship finding its second stage ignition.
- 1:17SPEAKER_00: And here's why this matters beyond just being a big number.
- 1:20SPEAKER_00: Anthropic was supposed to reach profitability around 2028.
- 1:24SPEAKER_00: They hit it two years early.
- 1:26SPEAKER_00: That tells you the enterprise AI market is maturing way faster than anyone predicted.
- 1:31SPEAKER_01: Two years early, that's not a rounding error.
- 1:34SPEAKER_01: That's a paradigm shift.
- 1:36SPEAKER_01: When a company beats its own financial projections by 24 months, something fundamental has changed in the market.
- 1:43SPEAKER_01: And that something is what every business owner needs to understand.
- 1:48SPEAKER_00: And the something is enterprise adoption.
- 1:50SPEAKER_00: Specifically, it's companies moving AI from the pilot phase, you know, the let's try this in a sandbox phase, into full production deployment.
- 2:00SPEAKER_00: That transition happened faster than anyone expected, and Anthropic was positioned perfectly to capture it.
- 2:07SPEAKER_01: Because they built for business, not for virality.
- 2:11SPEAKER_00: So what's driving this?
- 2:13SPEAKER_00: Because it's not consumer subscriptions, it's not people signing up for Claude Pro at 20 bucks a month.
- 2:18SPEAKER_01: That's what I thought it would be consumer growth, chatbot adoption, viral moments.
- 2:24SPEAKER_01: I was wrong.
- 2:25SPEAKER_00: You were wrong, and I'm going to enjoy saying that.
- 2:29SPEAKER_00: The primary driver is Claude Code.
- 2:32SPEAKER_00: They're a Gentic coding assistant for enterprise.
- 2:35SPEAKER_00: Claude Code alone is generating about$2.5 billion in annualized revenue.
- 2:41SPEAKER_00: And here's the kicker.
- 2:42SPEAKER_00: It did not exist 18 months ago.
- 2:45SPEAKER_01: A product that didn't exist a year and a half ago is now generating$2.5 billion a year.
- 2:53SPEAKER_01: If you're a small business owner, I want you to hear that number and really sit with it.
- 2:58SPEAKER_01: The AI tools market isn't theoretical anymore.
- 3:00SPEAKER_01: Companies are not just trying AI, they are buying it in volume.
- 3:05SPEAKER_01: Real contracts, real seats, real budgets.
- 3:08SPEAKER_01: This is not pilot programs and free trials.
- 3:11SPEAKER_01: This is production infrastructure.
- 3:14SPEAKER_00: The enterprise data backs this up.
- 3:16SPEAKER_00: The number of anthropic customers spending over a million dollars a year doubled from around 500 to over 1,000 in just a few months.
- 3:24SPEAKER_00: That's not hobbyists.
- 3:26SPEAKER_00: That's not experimentation.
- 3:27SPEAKER_00: That's JP Morgan, Goldman Sachs, Price Waterhouse Cooper's, Bristol Meyer Squib, companies rolling out clawed code across thousands of developer seats and paying real money for it.
- 3:38SPEAKER_01: And let me tell you what those companies are actually buying.
- 3:41SPEAKER_01: They're not buying a chatbot.
- 3:43SPEAKER_01: They're buying a tool that can refactor code across an entire code base, debug production issues, write tests, and review pull requests all at the same time.
- 3:52SPEAKER_01: A junior engineer costs them six figures a year.
- 3:55SPEAKER_01: Cloud code costs them a fraction of that and works 24-7.
- 4:00SPEAKER_01: The ROI is not theoretical, it's measurable and immediate.
- 4:05SPEAKER_00: And this is where I want to add some context that most coverage is missing.
- 4:10SPEAKER_00: Cloud code isn't just a coding assistant in the old sense, you know, the autocomplete thing that finishes your function.
- 4:16SPEAKER_00: Cloud code is agentic.
- 4:18SPEAKER_00: It can plan a multi-step change, execute it across dozens of files, run tests, and then fix what it broke.
- 4:25SPEAKER_00: It's more like having a junior developer who never sleeps and never asks for a raise.
- 4:30SPEAKER_01: I've seen businesses spend months debating whether to try AI tools.
- 4:35SPEAKER_01: Meanwhile, companies like Duolingo, Cisco Meraki, and Harvey are already running Cloud Code in production, shipping faster and cutting costs.
- 4:45SPEAKER_01: The decision isn't whether AI coding tools work.
- 4:48SPEAKER_01: The decision is whether you want your competitors to have them while you're still debating.
- 4:54SPEAKER_00: That's exactly right.
- 4:55SPEAKER_00: The work that would take a junior engineer hours for review, refactoring, and debugging now costs a few dollars in token spend.
- 5:03SPEAKER_00: When you can show a CFO that equation, hours of engineering time replaced by pennies of compute, the procurement conversation becomes very short.
- 5:12SPEAKER_01: And here's where it gets really interesting for our listeners.
- 5:15SPEAKER_01: Roughly 80% of Anthropic's revenue comes from API and enterprise contracts, not consumer chatbots, not people arguing with AI on Twitter.
- 5:26SPEAKER_01: It's businesses paying for AI to do real work.
- 5:29SPEAKER_01: And that distinction is the entire story.
- 5:32SPEAKER_00: It is, because this is where the AI race just fundamentally changed.
- 5:36SPEAKER_00: Let me give you the comparison that should make every business owner sit up straight.
- 5:55SPEAKER_01: Wait, say that again.
- 5:57SPEAKER_01: For every dollar they made, they lost more than a dollar on top of it?
- 6:01SPEAKER_00: That is correct.
- 6:02SPEAKER_00: For every dollar of revenue, OpenAI lost an additional$1.22.
- 6:09SPEAKER_00: Their projected operating loss for this year is around$14 billion.
- 6:13SPEAKER_00: Their inference costs alone, just the compute to keep ChatGPT running for all those free users, are estimated at over$8 billion in climbing.
- 6:22SPEAKER_00: They are not projected to reach cash flow positive until around 2030.
- 6:26SPEAKER_01: So let me make sure I have this right.
- 6:28SPEAKER_01: Anthropic, the smaller company, just turned a profit.
- 6:32SPEAKER_01: OpenAI, the bigger company with the household name, is burning through billions.
- 6:37SPEAKER_01: That is not a small story.
- 6:39SPEAKER_01: That is the story of this moment in AI.
- 6:43SPEAKER_00: And it comes down to a fundamentally different business model.
- 6:46SPEAKER_00: OpenAI is running a massive consumer operation, hundreds of millions of free users, all burning compute.
- 6:53SPEAKER_00: They're subsidizing access to grow market share, hoping that advertising in premium tiers will eventually catch up.
- 6:59SPEAKER_00: Anthropic went enterprise first.
- 7:02SPEAKER_00: They sell to businesses that pay real money for real productivity gains.
- 7:05SPEAKER_00: The unit economics are just better.
- 7:08SPEAKER_01: Let me translate that for the local business owner listening right now.
- 7:11SPEAKER_01: When the company selling AI to other businesses turns a profit, but the company giving AI away for free is hemorrhaging cash, that tells you which model actually works.
- 7:24SPEAKER_01: The enterprise model.
- 7:26SPEAKER_01: The model where businesses pay for outcomes, not for novelty.
- 7:30SPEAKER_00: There's also a psychological shift happening here.
- 7:33SPEAKER_00: When the most safety-conscious AI lab, the one that literally named itself after principles of responsible AI, is the first to turn a profit.
- 7:42SPEAKER_00: It challenges the narrative that you have to cut corners to make money in AI.
- 7:46SPEAKER_00: Anthropic proved that doing AI carefully and selling it to enterprises is not just morally defensible, it's financially superior.
- 7:55SPEAKER_01: That's a great point.
- 7:56SPEAKER_01: The company that everyone said was too cautious, too slow, too principled, they're the ones who figured out how to make AI profitable.
- 8:04SPEAKER_01: Not the move fast break things crew, not the free everything adds later crew, the careful ones.
- 8:10SPEAKER_01: That should tell you something about what the market actually values.
- 8:14SPEAKER_00: And there's a second layer here that I think is even more important for business owners.
- 8:18SPEAKER_00: Anthropic's compute cost per dollar of revenue dropped from 71 cents to 56 cents in one quarter.
- 8:25SPEAKER_00: Their inference gross margin went from about 38% to over 70%.
- 8:30SPEAKER_00: That means they're not just making more money, they're getting dramatically more efficient at delivering the same product.
- 8:36SPEAKER_01: That's a 15 cent improvement per revenue dollar.
- 8:38SPEAKER_01: On$10.9 billion in revenue, that's$1.6 billion in additional margin from efficiency alone in a single quarter.
- 8:47SPEAKER_01: So they're not just growing, they're growing smarter, getting more output from the same compute.
- 8:52SPEAKER_01: That's the kind of trend that makes investors very confident and competitors very nervous.
- 8:58SPEAKER_00: It also means something concrete for pricing.
- 9:01SPEAKER_00: When your margins are improving that fast, you have room to lower prices, which puts pressure on everyone else.
- 9:08SPEAKER_00: Open AI can't lower prices.
- 9:10SPEAKER_00: They're already losing money on every query.
- 9:12SPEAKER_00: Anthropic can afford to be aggressive on pricing because their cost structure is improving quarter over quarter.
- 9:18SPEAKER_01: And for small businesses, that price pressure flows downstream.
- 9:23SPEAKER_01: When the AI companies that supply your SaaS tools get more efficient, those savings eventually reach you.
- 9:30SPEAKER_01: Maybe not tomorrow.
- 9:31SPEAKER_01: But the direction is clear.
- 9:32SPEAKER_01: AI capability is getting cheaper at the same time, it's getting more powerful.
- 9:38SPEAKER_01: That's a rare combination.
- 9:40SPEAKER_00: There's one more number I want to put on the table.
- 9:42SPEAKER_00: Anthropic's annualized revenue run rate based on Q2 is about$43.6 billion.
- 9:49SPEAKER_00: That puts them in the same revenue neighborhood as companies like Salesforce and ServiceNow.
- 9:54SPEAKER_00: Except Anthropic got there in a fraction of the time.
- 9:57SPEAKER_00: Salesforce took over two decades to reach that scale.
- 10:00SPEAKER_00: Anthropic did it in under three years.
- 10:02SPEAKER_01: So here's what I want every small business owner to take away from this episode.
- 10:07SPEAKER_01: Five things.
- 10:08SPEAKER_01: First, AI adoption in business is no longer optional.
- 10:12SPEAKER_01: When companies like JP Morgan and PwC are spending millions on AI coding tools, they're not experimenting.
- 10:20SPEAKER_01: They're building competitive advantage.
- 10:22SPEAKER_01: If you're not even looking at these tools, you're already behind.
- 10:27SPEAKER_00: Second, the AI vendor landscape is shifting.
- 10:31SPEAKER_00: Open AI has the brand name, but Anthropic has the sustainable business model.
- 10:36SPEAKER_00: That matters for you because the company that can actually fund its own development with operating profits is the one that's going to be around in five years.
- 10:44SPEAKER_00: Anthropic just proved they can stand on their own.
- 10:47SPEAKER_00: They're not dependent on the next funding round to keep the lights on.
- 10:50SPEAKER_01: Third, Claude Code didn't exist 18 months ago.
- 10:54SPEAKER_01: $2.5 billion in annualized revenue from a product that's not even two years old.
- 10:59SPEAKER_01: That's not just fast growth.
- 11:00SPEAKER_01: That's a signal that the market for AI tools is expanding so quickly that new products can reach massive scale in months, not years.
- 11:10SPEAKER_00: Fourth, think about what that means for the tools coming next.
- 11:13SPEAKER_00: If Cloud Code went from zero to$2.5 billion that fast, what's coming in the next 18 months?
- 11:19SPEAKER_00: AI agents for accounting, for legal research, for HR, for customer service?
- 11:25SPEAKER_00: All of those verticals are going to see the same kind of explosive adoption curve.
- 11:30SPEAKER_00: The next Cloud Code is being built right now, and it might be for your industry.
- 11:34SPEAKER_01: I keep thinking about what this means for the local businesses I work with.
- 11:38SPEAKER_01: You're not JP Morgan.
- 11:39SPEAKER_01: You don't have a thousand developer seats, but here's the thing: the same technology that JP Morgan is paying millions for is available to you for 20 bucks a month through Cloud Pro.
- 11:50SPEAKER_01: The democratization isn't coming.
- 11:52SPEAKER_01: It's already here.
- 11:53SPEAKER_01: The enterprise proves it works.
- 11:56SPEAKER_01: The consumer tier makes it accessible.
- 12:00SPEAKER_01: You just have to use it.
- 12:01SPEAKER_00: And that's really the point.
- 12:03SPEAKER_00: Anthropics profitability isn't just a financial milestone for tech investors.
- 12:08SPEAKER_00: It's a proof point that AI tools deliver enough value that businesses willingly pay premium prices for them.
- 12:14SPEAKER_00: When half a billion dollars in quarterly profit comes from enterprise customers who could have chosen free alternatives, that tells you the free alternatives aren't good enough.
- 12:24SPEAKER_00: Paid AI is worth the money.
- 12:27SPEAKER_00: That's the message.
- 12:28SPEAKER_01: Fifth, and this is the one that should make you smile.
- 12:31SPEAKER_01: The price pressure is coming your way in a good direction.
- 12:34SPEAKER_01: When the profitable AI company can afford to lower prices and the unprofitable one can't, the profitable one wins the long game.
- 12:42SPEAKER_01: Expect better tools at lower prices over the next year, specifically from Anthropic and its partners.
- 12:49SPEAKER_00: And there's something else happening underneath this that I think business owners should understand.
- 12:54SPEAKER_00: The AI industry is bifurcating.
- 12:56SPEAKER_00: On one side, you have companies like Anthropic that are proving you can sell AI at a profit by focusing on value and enterprise outcomes.
- 13:03SPEAKER_00: On the other side, you have companies burning billions to give away free products and hoping that scale eventually solves the economics.
- 13:10SPEAKER_00: History is pretty clear about which model wins.
- 13:13SPEAKER_01: Every major technology platform, AWS, Salesforce, Shopify, went through this same inflection point.
- 13:21SPEAKER_01: The ones that proved they could make money survived and dominated.
- 13:24SPEAKER_01: The ones that kept burning cash, hoping for a miracle, eventually became cautionary tales.
- 13:29SPEAKER_01: Anthropic just passed that test.
- 13:31SPEAKER_01: OpenAI hasn't yet.
- 13:33SPEAKER_00: But there's a catch, and I think it's important to be honest about it.
- 13:36SPEAKER_00: Anthropic themselves have warned investors that scheduled increases in compute infrastructure costs later this year could push them back into the red.
- 13:45SPEAKER_00: This might not be sustained profitability yet.
- 13:47SPEAKER_00: It might be a profitable quarter, not a profitable era, at least not permanently.
- 13:52SPEAKER_01: That's fair, and that's actually the most honest take I've heard.
- 13:57SPEAKER_01: One profitable quarter doesn't mean the war is over, but it does mean something very specific.
- 14:02SPEAKER_01: It proves the model works.
- 14:04SPEAKER_01: The business model of selling AI to enterprises, focusing on productivity outcomes, charging for value instead of giving it away, that model just produced half a billion dollars in profit in a single quarter.
- 14:18SPEAKER_01: Even if costs go up later, that proof of concept is massive.
- 14:22SPEAKER_00: The other caveat I'd add is that Anthropic's profit figure excludes stock-based compensation.
- 14:28SPEAKER_00: So the real economic picture is a bit more nuanced than the headline suggests.
- 14:32SPEAKER_00: But even with that caveat, the direction of travel is unmistakable.
- 14:36SPEAKER_00: They're moving from loss to profit, and they're doing it faster than their own projections.
- 14:41SPEAKER_01: Here's what I know from running businesses for four decades.
- 14:44SPEAKER_01: When a company proves its model works, even for one quarter, it changes the psychology of everyone in that market.
- 14:51SPEAKER_01: Investors pour more money in, competitors scramble to copy the approach, and customers start treating you like you're permanent instead of provisional.
- 15:02SPEAKER_01: That last part is huge.
- 15:04SPEAKER_01: Businesses won't build critical infrastructure on top of a vendor they think might disappear.
- 15:09SPEAKER_01: Anthropic just gave every enterprise buyer a reason to trust them.
- 15:14SPEAKER_00: There's a comparison I want to make that I think really crystallizes this.
- 15:17SPEAKER_00: Remember when Amazon Web Services started turning a profit?
- 15:21SPEAKER_00: For years, people asked whether cloud computing was a real business or just a science experiment.
- 15:26SPEAKER_00: Then AWS posted real margins.
- 15:29SPEAKER_00: And suddenly every enterprise moved their infrastructure to the cloud.
- 15:34SPEAKER_00: Anthropic's profit moment is the AWS profit moment for AI.
- 15:38SPEAKER_00: It's the signal that says, this is infrastructure now, not innovation theater.
- 15:42SPEAKER_00: And that trust creates a flywheel.
- 15:45SPEAKER_00: More trust means more enterprise contracts.
- 15:47SPEAKER_00: More contracts means more revenue.
- 15:50SPEAKER_00: More revenue means more efficiency investment.
- 15:52SPEAKER_00: More efficiency means better margins.
- 15:54SPEAKER_00: Better margins means lower prices.
- 15:57SPEAKER_00: Lower prices means more adoption.
- 15:59SPEAKER_00: Anthropic just entered that flywheel.
- 16:02SPEAKER_00: And once you're in it, it's very hard to stop.
- 16:05SPEAKER_01: So let me wrap this up with the bottom line for you.
- 16:07SPEAKER_01: Anthropic turned a profit, the first major AI lab to do it.
- 16:11SPEAKER_01: They did it by selling real AI tools to real businesses, not by giving away free chatbots and hoping ads will save them.
- 16:18SPEAKER_01: Their revenue more than doubled in one quarter.
- 16:21SPEAKER_01: Their margins are improving, their enterprise customer base doubled in months, and their fastest growing product didn't exist a year and a half ago.
- 16:30SPEAKER_00: The AI race just changed overnight.
- 16:33SPEAKER_00: Not because someone released a new model, not because someone built a new feature, but because someone proved that selling AI to businesses is actually profitable.
- 16:43SPEAKER_00: That's the signal beneath the noise.
- 16:45SPEAKER_00: And if you're a business owner, that signal is aimed directly at you.
- 16:49SPEAKER_01: The tools are real, the money is real, the profit proves it.
- 16:52SPEAKER_01: Now the question is, are you using them yet?
- 16:55SPEAKER_00: And if the answer is no, ask yourself, what are you waiting for?
- 16:59SPEAKER_00: The companies that are paying anthropic millions of dollars a year aren't doing it because AI is trendy.
- 17:05SPEAKER_00: They're doing it because it makes them money.
- 17:07SPEAKER_00: The math is done.
- 17:08SPEAKER_00: The proof is in.
- 17:10SPEAKER_00: The only question left is whether you're on the right side of it.
- 17:13SPEAKER_01: That's the show.
- 17:14SPEAKER_01: We'll see you next time on Control AI Profit.