Claude for Small Business Ships | Altman Testifies He's 'Truthful'  🔗 **Source:** [Claude for Small Business Ships | Altman Testifies He's 'Truthful'](https://share.snipd.com/episode/1ed0be9f-32da-4f54-b670-9d6e97d94caa) ## Highlights - **Hyperscalers Build Onsite Power To Beat Grid Limits** - Hyperscalers are adding on-site generation because grid upgrades lag and regulators push local power for data centers. - Jaeden Schafer cites XAI adding 19 gas turbines (over 500 MW) at Colossus 2 and related NAACP litigation as an example. Transcript: Jaeden Schafer So XAI has added 19 more gas turbines at their Colossus 2 data center. So remember, Colossus 1 is where they're currently, they've now leased that over to Anthropic. Anthropic is using that. Then they have Colossus 2, which is currently the Southern Environmental Law Center is kind of suing them. They had got some emails from the Mississippi Department of Environmental Quality, and those emails basically showed that between March and May, XAI installed 19 new additional Gas turbines at Colossus 2, data center in Mississippi. That site currently has 46 turbines. That's over 500 megawatts of new gas generation that they've added since mid-March alone. This is what I think a lot of people are. This is what gets in the news a lot. Anyways, the NAACP filed a lawsuit against XAI in April over air quality and environmental justice concerns, right? The NACP is not usually concerned about environmental issues and not usually concerned about suing companies for like, you know, how they're generators, right? They're usually trying to protect people of color. And that's kind of what their organization does, right? So this is kind of weird, in my opinion, specifically, they're saying that there are, you know, in this area, there's a disproportionate amount of people of color that are being affected By the gas-powered generators, which are giving poor air quality. Anyways, I think it's really weird for the lawsuit to be coming there from the NAACP. It feels very political in a way, considering Elon Musk had ties to the Trump administration, all of that. So for me personally, I think that's not great. But I mean, maybe they have some some grounds, right? Perhaps eight of the 19 new turbines that they've just added over there that added more than 200 megawatts, they were installed after the lawsuit was first filed. So XAI is basically saying, they're like, look, yeah, we see you're suing us, we're going to keep building anyways. I think this is interesting, because basically every hyperscaler right now has to face the same choice. You can either wait for the grid to upgrade, we have, you know, a massive energy, we have massive energy problems. And I think in the recent Trump administration, they passed a bunch of laws where and they're basically telling a lot of these data centers, look, you should build a power generation On site to power your data center. Personally, I actually think that is the right direction because I hate it when one way or another, these data centers get some sort of handout from the government, which happens a lot In the form of subsidies. And who's paying for the subsidies? It's the people that live there. So I've seen my electrical prices go up a ton after Arizona gavezona gave out um you know for my my homes in arizona after arizona gave out a bunch of subsidies to data centers and other Places um i don't know if that's directly tied or correlated but when i see subsidies going out my electric price doubles in four years i'm you know not going to be thrilled about it so With all of that happening i know like a lot of people throw shade at xai because they're like oh they're gas turbines yada yada but like at the end of the day, I'm really happy that they Are generating their own power on site. So at least people's electrical costs don't go up. As far as, you know, creating smog and other things, I would hope that they have some sort of good carbon capture programs. And if they don't, I mean, that's something you should probably litigate or get figured out. But I don't think intrinsically having gas fire generators is the end of the world, considering that's how a huge portion of America is powered. I think people just don't know that. But anyways, at the end of the day, a lot of these hyperscalers are going to have to find solutions for more gigawatts of power. We litigated nuclear to oblivion, and now we're trying to litigate gas. And at the end of the day, like we need electricity from some source and solar and wind is not going to do it for every every place and for every data center. So I hope that we get this problem figured out. ([Time 0:01:53](https://share.snipd.com/snip/2c584483-9a98-4cb8-98f8-41c41dae07cb)) - **Weigh Grid Independence Against Environmental Impact** - Anticipate tradeoffs: on-site generation avoids grid reliance but raises environmental and community concerns to litigate. - Jaeden Schafer urges ensuring carbon capture or mitigation if companies install gas turbines to limit air quality impact. Transcript: Jaeden Schafer And who's paying for the subsidies? It's the people that live there. So I've seen my electrical prices go up a ton after Arizona gavezona gave out um you know for my my homes in arizona after arizona gave out a bunch of subsidies to data centers and other Places um i don't know if that's directly tied or correlated but when i see subsidies going out my electric price doubles in four years i'm you know not going to be thrilled about it so With all of that happening i know like a lot of people throw shade at xai because they're like oh they're gas turbines yada yada but like at the end of the day, I'm really happy that they Are generating their own power on site. So at least people's electrical costs don't go up. As far as, you know, creating smog and other things, I would hope that they have some sort of good carbon capture programs. And if they don't, I mean, that's something you should probably litigate or get figured out. But I don't think intrinsically having gas fire generators is the end of the world, considering that's how a huge portion of America is powered. I think people just don't know that. But anyways, at the end of the day, a lot of these hyperscalers are going to have to find solutions for more gigawatts of power. We litigated nuclear to oblivion, and now we're trying to litigate gas. And at the end of the day, like we need electricity from some source and solar and wind is not going to do it for every every place and for every data center. ([Time 0:04:20](https://share.snipd.com/snip/5919b7b6-6933-4304-91fe-eeaabe59ab86)) - **Build Or Use Privacy Scrubbers For Model Inputs** - Expect a new privacy tooling layer to emerge that filters or blocks brokered personal data before it reaches models. - Jaeden Schafer points to product responses like Meta's Incognito chat and likely third-party scrubbing services. Transcript: Jaeden Schafer Okay, next up, we have AI chatbots that are leaking real phone numbers. And this isn't just one chatbot. This is all of them. This is Google. This is OpenAI. And this is Anthropic. There is a company, it's a security firm called DeleteMe. And they basically help people scrub their personal info off of data broker sites. And they published a report basically showing how many people were complaining about their data showing up specifically in AI related chats. So someone's going to go chat, you know, prompt ChatGPT and say, what is Jaden Schaefer's phone number? What is his address? And they would get my information back of all of the complaints that people have kind of submitted to delete me. 55% are coming from chat GPT 20% from Gemini and 15% from Claude and then 10% come from other tools. But I think basically people are asking chat bots, you know, addresses, their phone numbers, their, you know, family information, and they're, you know, they're getting actual real Responses back. So where is all of that data in particular coming from? It's coming from data brokers in California, 31 of the 578 registered data brokers self reported that they're selling or sharing consumer data with generative AI developers in the Last 12 months. And so basically, that's the supply chain of where all this information is getting into the AI models, that data is bought, it's then fed into the models and the models spit it back out, If you know how to prompt it the right way. So I think there's going to be a whole new category born out of this kind of like privacy tooling layer around some of these foundational models. I think people are going to I mean, just like in healthcare and so many other industries, this is something that is not great. So I think we're going to start seeing a lot of products to help solve this problem. And even if you think about like what Zuckerberg is doing, you know, he just, he just like came out with end-to encrypted meta AI mode, a new, you know, thing called Incognito chat. It's kind of the same thing, but it's, I guess, a different response to this. So I think a lot of the market is going to start focusing on making sure that this is not a problem anymore. ([Time 0:05:38](https://share.snipd.com/snip/4ca44739-0cc3-4a58-a970-24570b117b96)) - **Chatbots Are Leaking Phone Numbers From Data Brokers** - AI chatbots from Google, OpenAI, and Anthropic are returning real personal phone numbers sourced from data brokers. - DeleteMe found 55% of complaints from ChatGPT, 20% Gemini, 15% Claude, and 31 of 578 brokers self-reported sharing data with AI developers. Transcript: Jaeden Schafer Okay, next up, we have AI chatbots that are leaking real phone numbers. And this isn't just one chatbot. This is all of them. This is Google. This is OpenAI. And this is Anthropic. There is a company, it's a security firm called DeleteMe. And they basically help people scrub their personal info off of data broker sites. And they published a report basically showing how many people were complaining about their data showing up specifically in AI related chats. So someone's going to go chat, you know, prompt ChatGPT and say, what is Jaden Schaefer's phone number? What is his address? And they would get my information back of all of the complaints that people have kind of submitted to delete me. 55% are coming from chat GPT 20% from Gemini and 15% from Claude and then 10% come from other tools. But I think basically people are asking chat bots, you know, addresses, their phone numbers, their, you know, family information, and they're, you know, they're getting actual real Responses back. So where is all of that data in particular coming from? It's coming from data brokers in California, 31 of the 578 registered data brokers self reported that they're selling or sharing consumer data with generative AI developers in the Last 12 months. And so basically, that's the supply chain of where all this information is getting into the AI models, that data is bought, it's then fed into the models and the models spit it back out, If you know how to prompt it the right way. So I think there's going to be a whole new category born out of this kind of like privacy tooling layer around some of these foundational models. I think people are going to I mean, just like in healthcare and so many other industries, this is something that is not great. So I think we're going to start seeing a lot of products to help solve this problem. And even if you think about like what Zuckerberg is doing, you know, he just, he just like came out with end-to encrypted meta AI mode, a new, you know, thing called Incognito chat. It's kind of the same thing, but it's, I guess, a different response to this. So I think a lot of the market is going to start focusing on making sure that this is not a problem anymore. ([Time 0:05:38](https://share.snipd.com/snip/0695c193-211f-4a1d-b8ec-f9daac5825a3)) - **Sutskever Document Could Haunt Sam Altman** - The Elon Musk vs Sam Altman trial is airing internal disputes, including a 52-page Ilya Sutskever document alleging Altman lied. - Jaeden Schafer highlights the document was used in cross-examination and may cause long-term reputational damage. Transcript: Jaeden Schafer Okay, next up, Sam Altman just spent four hours on the stand in his trial, Elon Musk versus Sam Altman. And in this whole lawsuit, Elon Musk, I guess for a little bit of context, spent about three days on the stand earlier in the trial. And I felt bad because you kind of had like Elon Musk was the first one on the stand. He got absolutely roasted. And then everyone's like, oh, my gosh, he's toast. Then it feels kind of like a boxing match, right? Now we got Sam Altman on the stand. Now we're going to roast him. Anyways, this goes back and forth. I'm not particularly on anyone's side with all of this, although I do feel like it's kind of shady for a non-profit to become a for-profit company. So I guess maybe that's a bit of a bias there. And it's not even particularly Sam Altman or OpenAI. I just wouldn't want to see that with any company. People are saying got off relatively light this time, but the cross-examination was super brutal. So Elon Musk's lawyer is Stephen Molo, and he opened up by asking Sam Altman directly whether he was, quote, completely trustworthy. Then Molo pulled out a 52-page document that was compiled by Ilya Suskever, which is Sam Altman's former co-founder and also the former chief scientist at OpenAI. And this is something that Suscovor has described basically as documenting, quote, a consistent pattern of lying by Sam Altman. So the lawyer pulls out this document by his former co-founder and goes through this consistent pattern of lying and every single thing in it. So this is a real document. It's about 52 pages long and it's from Suscovor when he tried to fire Sam Altman back in November of 2023. And I guess this document was sort of the justification that Ilya had for trying to make this decision. Of course, you know, Sam Altman came back. Ilya Suscovur left. A lot of the founders and co-founders of OpenAI left. There's a lot of drama in all of that. But I think right now, bringing this up, I don't know if it's really going to change much of the verdict of the case. We have, of course, $150 billion Elon Musk is asking for in damages from OpenAI, and he said that he's going to donate those damages back into OpenAI's nonprofit. So he's like, look, pay me $150 billion and then turn it back into a nonprofit. I'll donate the money back to the nonprofit and kick out, you know, kick out Sam Altman. What I don't think is going to happen is I don't think Elon's actually going to get $150 billion. I don't know if it's going to have to convert back to a nonprofit. But regardless, I think the reputational damage of having this Ilya Susquefort document, because it's kind of private and to the board when they're firing Sam, but having this read Into the record trial, I think that is something that is going to follow Sam Altman for years because now it's, you know, official and public. So that is going to be something Sam Altman is going to be dealing with for a long time. ([Time 0:07:32](https://share.snipd.com/snip/064272c7-b492-4c09-87a9-25ad52e5ec56)) - **Anthropic Surged To Lead Paid Business AI Spend** - Anthropic overtook OpenAI in paid business adoption per Ramp's expense data, reaching 34.4% vs OpenAI's 32.3%. - Jaeden Schafer notes Anthropic rose from 9% a year ago to gain 26 percentage points in corporate spend share. Transcript: Jaeden Schafer The first one is that Ramp, which is the corporate spending platform, they just released their May AI index yesterday. And for the first time ever, Anthropic is the top vendor for paid business AI. There's 34.4% of businesses that in Ramp's sample of businesses that are paying for Anthropic service and of OpenAI, 32.3%. So Anthropic is officially passing OpenAI for the top AI model that businesses are paying for. This is what I think is pretty wild. 12 months ago, Anthropic was at 9%. So they're going from 9% in one year. They've now added 26 points of paid business adoption. And I think overall, AI spending, the whole market only grew 9%. So they are way up. I don't think they just, you know, took share. I think they basically kind of redrew this whole map. And I actually love these reports coming from a company like Ramp, because they basically Ramp processes expense data for over 50,000 client companies. ([Time 0:11:01](https://share.snipd.com/snip/7e2454b1-13e8-485e-be93-9cc84a9b9660)) - **Using Claude CoWorker To Automate QuickBooks Categorization** - Anthropic launched Claude for Small Business inside Claude CoWorker with integrations like QuickBooks, Canva, DocuSign, HubSpot, and PayPal. - Jaeden Schafer used Claude CoWorker to auto-categorize his QuickBooks expenses and reduce accountant work. Transcript: Jaeden Schafer Okay, the next thing for Anthropik that we have to talk about is that Anthropik has just launched something called Claude for Small Business. It's basically a toggle inside of the Cowork agent platform. And Cowork is the product that basically I use it every single day. I don't even use Claude code. I just use Claude Cowork. It can take control of your computer. It can browse the web for you. It can manage all the files on your computer. It can do tons of workflows. I have automations. I have skills where I teach you how to do something and it can do them every day. So anyways, the small business bundle is now going to add bookkeeping, business insights, ad generation, and all this is going to be directly tied into integrations with other third-party Software. So for example, QuickBooks, Canva, DocuSign, HubSpot, PayPal, those are some of the apps that I think small businesses actually use every single day. And this is what Anthropic is tying in. Now, what's wild is that the market is about 36 million US small businesses, 44% of US GDP, which is almost half of the private sector workforce. It's all in SMBs. And so Anthropoc is trying to target this heavily. I mean, personally, I think this is a brilliant move. It's a massive market. Making this easier for those customers is a great play. I personally the other day was just using QuickBooks to go through all of my expenses and categorize them inside of using, sorry, I was using Claude Cowork inside of my QuickBooks to Categorize all of my expenses. And it's really cool because if it didn't know what a specific expense was, it would go to Google, search the name of the company, figure out what they did, go back and then know how to accurately Expense that out. And I actually was able to tell my accountant who I usually pay to do that, like, hey, look, I mean, keep doing my taxes and a few other things for me, but like for expense categorization, I don't really need that anymore. I think a lot of small businesses are going to see a lot of really great ways to save money and also great ways to be more productive with a lot of these tools. Anthropic is absolutely going after a great segment. ([Time 0:12:25](https://share.snipd.com/snip/62e48719-414c-4d1c-a041-c6faed6aea64))