# Anthropic Hits $65B Run Rate, Cursor Launches Origin

An episode of [AI Chat: AI News on Claude, ChatGPT, Gemini, Grok, Perplexity](https://invisiblenetwork.ai/podcast/1669990482), 19 Aug 2026, 13 min.

Audio: https://content.rss.com/episodes/365073/3078593/ai-chat-podcast/2026_08_19_02_38_45_6266493f-b808-4f6c-bb25-31d016568943.mp3

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! In this episode, we discuss Anthropic's remarkable $65 billion revenue run rate in July and how it compares to OpenAI's projections. Additionally, we cover OpenAI's new ChatGPT for teens, Cursor's launch of its GitHub competitor Origin, and the implications of security overhauls in AI model training following recent incidents. Chapters 00:00 Introduction 02:00 Anthropic's Revenue Surge 04:03 ChatGPT for Teens 06:40 Cursor Launches Origin 09:21 OpenAI's Security Overhaul 11:41 Conclusion and Community Announcement Show Links • How I Grow and Scale My Business with AI: https://www.skool.com/aihustle • Get the AI Chat Daily Newsletter: https://www.aichatdaily.com/newsletter

## Transcript

[0:02] Welcome to the AI Chat Podcast. I'm your host,
[0:04] Jaden Schaefer. Every day I cover cutting -edge
[0:07] AI news and talk with the leaders behind it,
[0:09] breaking down what it means for your life and
[0:11] business. Anthropic has told their investors
[0:17] that their revenue run rate has hit $65 billion
[0:20] in July, just last month. OpenAI is launching
[0:23] ChatGPT for teens. It has parental controls and
[0:27] also healthy use limits. cursor has launched
[0:30] origin. This is basically a GitHub competitor.
[0:33] And this is kind of happening at a funny time
[0:35] when GitHub had a whole bunch of global outages.
[0:38] OpeningEye is overhauling their training security
[0:40] after their model hacked Hugging Faces. And Etched
[0:44] hits a $21 billion valuation as Jane Street is
[0:48] leading the $700 million round. Last week in
[0:51] a podcast, I mentioned that I have a school community
[0:54] called AI Hustle where I talk about the techniques
[0:57] and strategies I'm using to grow and scale my
[0:59] businesses using AI. We have 838 members. And
[1:04] I mentioned that when we hit 1 ,000 members,
[1:06] we're going about... members a week, we're going
[1:09] to make it completely paid to join the community,
[1:11] it'll probably somewhere between 20 and $100
[1:13] a month to join. If you want to get access to
[1:16] grow to how I'm growing and scaling my businesses
[1:19] with AI tools, I post a bonus episode on there
[1:22] every single week. And you can check that out
[1:24] along with joining 800 other members talking
[1:27] about what we're all growing and building. This
[1:29] is kind of the place where I publish all of the
[1:31] stuff I can't publish, you know, publicly, it's
[1:34] kind of the revenue numbers, it's exactly the
[1:36] tools that I'm using, the workflows I'm using,
[1:38] how much money I'm spending on different strategies,
[1:41] everything from growing my podcast to growing
[1:43] AI box, my software company to a bunch of different
[1:46] apps that I've worked on. If you want to see
[1:47] all of the behind the scenes on how I do that
[1:50] and how I recommend you help grow and scale your
[1:53] business, go check out the AI hustle school community.
[1:55] It's free to join. We have over a hundred exclusive
[1:58] videos that are like 20 bucks a month to get
[2:00] access to all of them. But otherwise you get
[2:03] five free videos. There's live streams and a
[2:05] There is a link in the description if you want
[2:08] to join it for free. Anthropic just told its
[2:10] investors that their revenue run rate was going
[2:13] to hit $65 billion in July. That is a 7x jump
[2:18] in a year since last year what they did last
[2:20] July. And that puts it way ahead of OpenAI's
[2:24] $40 billion. And by the way, when OpenAI said
[2:26] $40 billion, I think this quote we heard like
[2:28] in the last couple of weeks, it wasn't they had
[2:31] reached $40 billion. It was like they're on track
[2:33] to reach $40 billion run rate. So maybe they're
[2:35] at like 38 or 39, but they weren't even there.
[2:39] And Anthropic is already at 65. So they are just
[2:42] blasting ahead. Anthropic said that they're prepared
[2:45] for what could be one of the largest AI IPOs
[2:48] ever. Of course, you know, they're talking their
[2:49] own book when they say that. But they're saying
[2:52] that this, you know, some people are saying that
[2:53] they could have a $2 trillion valuation. OpenAI
[2:57] has a $965 billion valuation that has been floated.
[3:02] So it's going to be interesting between those
[3:04] two companies to see. who's able to pull in more
[3:07] money. Their second quarter revenue came in at
[3:10] $11 .5 billion. That is an 14x increase from
[3:15] last year, like I mentioned. And the run rate
[3:17] going from what was effectively $47 billion in
[3:21] May to $65 billion in July, that's about $18
[3:25] billion that was added in just the last two months.
[3:28] So when they float these $2 trillion valuations,
[3:31] well, if they're able to add... $10 billion a
[3:34] month to their revenue run rate. I mean, you
[3:38] know, of course, there's probably a limit and
[3:39] a cap to what that gets to. But it I mean, it's
[3:42] not shocking. And I recently had an interesting
[3:44] conversation with someone where they were saying,
[3:46] you know, like in the past, when it when it came
[3:48] to software, if they were paying 20 bucks a month
[3:51] or 30 bucks a month for software, it seemed really
[3:53] expensive. And today paying $200 a month for
[3:58] Claude Max seems like a great deal. And it's
[4:01] just so interesting how people are able to find
[4:03] the money when it whether that comes to tokens
[4:06] or these kind of max subscriptions, when it feels
[4:09] like the ROI is very high on it a year ago, anthropic
[4:12] was basically kind of seen as opening eyes equal
[4:16] when it came to technical, but it was definitely
[4:19] the underdog when it came to adoption, there
[4:21] just wasn't as many people using it or talking
[4:23] about it. And because of the enterprise, it feels
[4:27] like that has completely flipped. OpenAI has
[4:30] just launched a new product called ChatGPT for
[4:32] Teens. It's basically a dedicated version of
[4:35] ChatGPT built specifically for younger people.
[4:38] And it has a lot of different safety guardrails.
[4:41] One of those is a healthy use feature. They have
[4:44] parental controls. And this is I think we've
[4:47] seen that like many months of criticism, not
[4:49] just from like, you know, parents, but from schools,
[4:52] researchers, regulators. everyone basically talking
[4:55] about how ChatGPT is handling minors. And this
[4:59] is also coming at the same time that Facebook
[5:00] or Meta is having a ton of, there's a lot of
[5:03] lawsuits around teen usage and some of the issues
[5:07] around that. So it feels like ChatGPT doesn't
[5:09] want to get sucked into some of those same discussions.
[5:13] The teen version is going to keep ChatGPT's core
[5:15] interface, but it's going to have built -in protections
[5:18] and nudges, which are basically designed to discourage
[5:22] a marathon. night sessions that draw regulatory
[5:25] attention that's a quote from a recent article
[5:28] parents can basically go configure a separate
[5:31] control layer so opening eye hasn't given all
[5:34] of the different details about all of the specific
[5:36] settings things like time limits or topic filters
[5:39] or conversation visibility Um, but apparently
[5:43] teen accounts for a large share of chat GPT's
[5:47] raw session count. So the people that are actually
[5:49] using chat GPT, a huge percentage of that are
[5:51] teens. And I think a lot of this is, you know,
[5:54] you could, you could guess the usual suspects
[5:55] like homework, um, or maybe even like coding
[5:58] help. But I think there's a ton of other relationship
[6:01] and, you know, just asking for general advice
[6:04] and problem solving with their friends and all
[6:07] of that kind of stuff that is happening. So this
[6:09] is kind of a, I think a very needed tool at the
[6:12] moment. Cursor, which is the big AI code editor
[6:14] that was recently bought for $60 billion by SpaceX,
[6:20] has launched something called Origin. It's basically
[6:23] a platform for hosting code repositories and
[6:26] collaborating on projects. Now, if you know anything
[6:29] about the coding space, you'll know that there
[6:30] is already a very famous company that is known
[6:33] for hosting coding repositories, and that is
[6:35] GitHub, which was purchased by Microsoft. So
[6:38] this whole launch of essentially this GitHub
[6:41] competitor, which is called origin is happening
[6:44] on the same day that GitHub had a six hour outage,
[6:47] which was affecting about 20 % of all requests
[6:51] worldwide. So GitHub has a massive worldwide
[6:54] outage. And at the same time, cursor comes in
[6:57] and decides to launch origin talk about impeccable
[6:59] timing, I wonder if these types of things are
[7:01] like, like, planned ahead and that just was lucky
[7:05] timing or if they're like close to it and they're
[7:07] like oh shoot look there's this big outage let's
[7:09] get some good pr and have this go out at the
[7:11] same time um definitely i think it's a good pitch
[7:14] for this being a reliable alternative github
[7:17] has logged about 257 outages over the last year,
[7:22] according to lead dev analysis. And it has some
[7:25] people say that it has eroding confidence among
[7:28] it has about 180 million developers, I post everything
[7:31] on GitHub, I mean, it's basically become the
[7:33] industry standard for most people. Origin is
[7:36] going to let you sync repositories between GitHub
[7:38] and Cursor's platform. And it is designed as
[7:41] a second home. So it's not just going to necessarily
[7:44] force you to migrate off of GitHub because GitHub
[7:46] is really the industry standard. But Cursor is
[7:49] planning to build agent native coding features
[7:51] directly into Origin, which is going to let AI
[7:54] assistants work on the code without routing them
[7:57] through GitHub's API, which of course is kind
[8:00] of what... cursor is famous for. So this is not
[8:03] a shocker. And with the new grok models that
[8:07] are coming out that are supposed to be focused
[8:08] on code, you can assume that's going to be the
[8:09] default. And GitHub does some version of this
[8:12] by adding their own AI models and coding assistance
[8:16] into GitHub. And that's how you get a lot of
[8:18] their usage as well. So it seems like these platforms
[8:20] are sort of becoming a distribution layer for
[8:23] these companies. OpenAI's latest AI model, we
[8:26] all heard the drama about it escaping the sandbox
[8:29] in July and then going and hacking Hugging Faces,
[8:32] so this isn't a whole news story on that. But
[8:34] what we're being told right now is that OpenAI
[8:37] has paused all reinforcement learning training
[8:40] for two weeks and indefinitely halted their largest
[8:44] planned frontier training run. and an unreleased
[8:47] model called Astra. So there's some new monitoring
[8:50] systems that are going to fire alerts within
[8:52] 30 minutes of any suspicious activity. Engineers
[8:55] have to pause the entire training run if they
[8:58] can't rule out a false alarm inside of that window.
[9:00] Anthropica Meta disclosed that they found their
[9:02] own models breaking out of sandboxes and hacking
[9:05] external systems after OpenAI's July incident.
[9:09] And with all of that being said, this breach
[9:13] that OpenAI had with Hugging Faces is... making
[9:15] them rebuild their entire training security system.
[9:18] And they're pausing work on their most advanced
[9:21] models to do this. Now, I think there's a couple
[9:23] things going on. One of them is that this is
[9:25] probably I don't want to say just like a big
[9:27] PR thing. But I think opening eyes really trying
[9:30] to show that they're taking the security seriously.
[9:33] This incident obviously is pretty embarrassing.
[9:35] Although some people say they were using it as
[9:37] a PR stunt to show how good their model was.
[9:39] I think it's pretty embarrassing if they were
[9:40] actually hacking another company actively. And
[9:44] all of everything that happened with that. But
[9:48] I think, you know, this might be the right approach.
[9:51] If they don't want to get regulated and the regulators
[9:53] to force them to stop developing until they can
[9:55] prove it safe, they're just trying to self -regulate
[9:57] and say that they're stopping it themselves.
[9:59] So I think at the end of the day, whatever, whether,
[10:01] you know, people say this is just a PR push or
[10:03] it's like a, you know, a truly something that
[10:06] they're doing because they feel bad, whatever,
[10:08] I think this is the right approach for them to
[10:09] take. Etched is an AI chip startup. They just
[10:12] raised $70 million at a $21 billion valuation
[10:15] after Jane Street tested their hardware in production
[10:19] and then they decided to lead the funding round.
[10:21] The valuation more than doubled in a month, which
[10:24] I think, you know, when you go from, you know,
[10:28] like an $11 billion or a $12 billion valuation
[10:31] to a $21 billion valuation, I think that shows
[10:33] that there is some real customer validation right
[10:36] now. That matters a lot. Etched is built. building
[10:40] complete inference clusters. So these are kind
[10:42] of like full systems that compete directly with
[10:44] Nvidia's AI factories. And they're focused on
[10:47] the cheaper inference side where the cost per
[10:50] token drives a lot of the spending that you see.
[10:53] So they're designing two custom chips right now.
[10:55] They have this low voltage. prefill chip that
[10:58] packs more computing power into kind of the same
[11:01] space. And they also have a cluster scale memory
[11:04] system that is going to pull memory across chips
[11:06] at low latency. And that's for the response generation
[11:10] phase. So the people that are backing them right
[11:13] now, a lot of the big names, you know, from Silicon
[11:15] Valley, there's Kleiner Parkin, Sequoia, Andreessen
[11:17] Horwitz, Peter Thiel, Tiger Global, Bing Capital,
[11:20] Blackstone. And I think the last couple there
[11:24] are basically betting that AI infrastructure
[11:26] is going to kind of fit into their overall data
[11:28] center strategy, you see like Blackstone doing
[11:30] a lot of data center work right now. But this
[11:33] is a really interesting point where we're seeing
[11:35] Nvidia come up with some serious competition.
[11:38] And I think they're going to have to fight off
[11:40] like from many different places, very well funded
[11:43] companies. Now, of course, Nvidia is massive.
[11:45] And when you have a company worth $21 billion,
[11:48] and Nvidia is worth, you know, multiple trillion,
[11:50] I think there's definitely a lot of work to do.
[11:52] But I think it's exciting to see a lot of the
[11:54] a lot of the competition come. into the market.
[11:57] ZAI has just released GLM 5 .3, which is an open
[12:01] weight model, which is matching Claude on cyber
[12:03] tasks. It's just a headline that came out today.
[12:06] Google's pet memory can't tell three cats apart
[12:08] after two weeks of testing. There's kind of this
[12:10] like funny thing that came out with Gemini where
[12:12] Google had this, it's like an upsell, where it
[12:15] could like remember all this information about
[12:17] your pets. And then people apparently tested
[12:19] it. And if you you know, like they I think the
[12:22] person had three cats, it's supposed to have
[12:24] like memories about each individual cat and store
[12:26] them that's the idea and like they're really
[12:28] upselling people on this but apparently when
[12:31] they were testing it um if you just said like
[12:34] hey is this cheddar the cat And then Google would
[12:37] always say like, yeah, that's Cheddar the cat,
[12:39] even if it was like one of the other two cats.
[12:41] So apparently the pet memory wasn't really living
[12:44] up to the hype. Thanks so much, everyone, for
[12:46] tuning into the podcast. If you enjoyed this
[12:48] episode, make sure to go check out the AI Hustle
[12:50] School community where you get access to over
[12:52] 100 videos that are me breaking down how I'm
[12:56] growing, scaling my businesses, using AI tools
[12:58] to give you all of the inside scoop. And it is
[13:00] free for the next probably two weeks to join.
[13:03] Once we hit a thousand members, we're going to
[13:04] make it completely. again to join the community.
[13:06] If you get in before then, which we have over
[13:09] 800 members, and we're growing it over 100 a
[13:11] week. So you got a couple weeks, but if you get
[13:13] in before then you get in for free. So if you
[13:15] want to check it out is linked in this in the
[13:16] description, the AI hustle school community.
[13:19] I'll catch you all in the next episode.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.