Latest / The Accounting Podcast / Intuit's $100m/yr OpenAI Deal & How Tech Companies Juice Profits

Transcript

Intuit's $100m/yr OpenAI Deal & How Tech Companies Juice Profits · The Accounting Podcast

• There may be errors in spelling, grammar, and accuracy in this machine-generated transcript. Blake Oliver: Ai is now a first stop for tax and investing questions. A new survey found that 10% of adults acted on AI tax guidance in just 30 days, while 21% followed AI crypto advice. So adults are acting for tax advice, and they are. They're asking for it and they are acting on it. 10% of them. Here's the scary part. Ai gets complex tax [00:00:30] questions wrong. Up to 50% of the time. David Leary: Coming to you weekly from the OnPay Recording Studio. Blake Oliver: Hello and welcome back to The Accounting Podcast, your weekly roundup of news in the accounting profession. I'm Blake Oliver. David Leary: And I'm David Leary. And yes, it is a welcome back Blake. We we if any of your listeners probably know we've had a little bit of a gap here in our recording. Blake got sick and sick and did not have a voice. And but now we're going to make up for it. We're we're going to record an episode today, [00:01:00] we'll record an episode before Thanksgiving. You're going to get all your accounting news and all your fill. But what happened? How'd you get sick? This sick? Deathly sick. Blake Oliver: As usual. Las Vegas. I blame Las Vegas. Um, I was at the American Society of Cost Segregation Professionals conference delivering a keynote, and I came back with a cold that just turned into the worst thing ever. And so I've been under the weather. You might be able to hear it still a little bit, but, um, I'm recovered [00:01:30] now, so we are. We're back on. We're getting back on track. David Leary: It was so bad that Blake sent me a voicemail so I could hear his voice to prove that he just wasn't wanting to skip. He was actually sick, and he lost his voice. So. Well, so much better. Now, though. Blake Oliver: I don't know why I had to prove it to you, David. We all know those cost seg professionals really go hard when they're in Las Vegas. Um. David Leary: So we we went I did the math. We went seven years and seven months without missing a week. Blake Oliver: Well, we haven't missed a week yet because we're going to catch up. David Leary: It's like that. No injuries [00:02:00] on job sign, right? That's right. An injury on site in this many days. Yeah. Blake Oliver: Well, we're going to, uh, we're going to get going here today. We're going to catch up. We're going to talk about Intuit's OpenAI deal. I want to talk a little bit about my presentation that I did at the conference. It was all about AI for that particular niche within accounting. But there's some material that I think would be interesting to our listeners. We're also going to talk about, um, a, uh, accounting [00:02:30] manipulation, maybe you could call it a fraud, but it's widespread within the tech world right now, especially in the artificial intelligence space. David, you you found that. We'll talk about that. Um, the PCAOB did a little chat with accounting today, and we've got some more info on the rippling versus deal situation with that corporate spy. A little bit of follow up there. Um, but first, let's thank our sponsors, [00:03:00] our sponsors. David Leary: This week we have Onp. Relay and cloud account and Staffing. Forbes and CNBC rank on pay number one for small business payroll on pay really knows how to get payroll done right for every client you serve, no matter how complex their software is, easy to use and backed by outstanding service levels, they handle new client onboarding for free and have experts on call to keep you and your clients on track. The system includes multi-state payroll, local tax filings, integrated HR tools, and more with no hidden fees. [00:03:30] And when you join on Pace Partner program, you get a custom dashboard to easily manage all your clients in one place. Plus, you can gain exclusive perks like revenue sharing or discounts, free payroll for your firm, co-branding opportunities, premium swag, and more on pay helps you run your practice efficiently while providing exceptional payroll that your clients can count on. To learn more about using Onp for your firm and clients that might be farm Pharm startups, restaurants, bars, doctors, nonprofits, gyms, franchisees or dentists. Head over to The Accounting Podcast that [00:04:00] is The Accounting Podcast. Blake Oliver: All right. So I want to share a little bit of my presentation from the conference. This was the Asccp annual conference in Las Vegas. David Leary: A quick because I cost cost gets thrown around a lot. It's all over Twitter. I think the self storage guys like it. Give us a ten second summary of what cost SEG is for our listeners that may not be, you know, in the loop. Blake Oliver: So [00:04:30] you're a real estate investor, David, and you acquire or you build a property, you buy a hotel, you buy a factory or you build one, you have to then depreciate that property over time, right? You're going to depreciate the factory, the hotel, whether you built it or you acquired it. David Leary: So I just take the building and I figure out my depreciation schedule and I just [00:05:00] mark it down. Blake Oliver: Yeah. And the easiest way to do this is just take the purchase price or whatever you paid to build this thing and then pick a you depreciate it over, you know, whatever the the rule is, you know, 39 years or something like that. And cost segregation is about looking at the property and identifying all of the different, um. Pieces of equipment, elements of it, [00:05:30] piping, parking lot, pool. David Leary: So the building might depreciate over 30 years, but the air conditioning that I put on those units might have to depreciate in 5 or 6, because their usefulness isn't exactly. Blake Oliver: They all have different useful lives. All the components of this property have different useful lives. Got it. Um, and so when you do a classic study, you're basically looking at that, that property and you're adding up all the different components, and then you can take that purchase price or the construction [00:06:00] price, and you can then allocate it into the different buckets of depreciation. And some of it you can depreciate way faster like in five years versus 39 years. And it allows you to get a tax benefit upfront because you're spending millions and millions of dollars on this property. You get a big tax benefit in the first year. Okay. You accelerate that depreciation. So it's a huge industry, right. Because basically every time you buy a property or you build something, you want to do this because of the rules. And so, [00:06:30] um, you know, I presented for this, uh, this group, uh, last year, this is my second year doing it. And so it was interesting to go back to my previous presentation and like, reflect on where we are now versus where we are, where we were. David Leary: So so your AI presentation from 2024 versus a year later. Ai presentation. Blake Oliver: That's right. David Leary: Okay. Blake Oliver: And one of the concepts that I focused on last year, in 2024 was this concept of the AI tipping point. I argued that [00:07:00] we at that time were on the precipice of a tipping point with AI adoption in our firms within the next few years that it would begin to accelerate and that we were at this curve, this we're sort of like at the if you if you think of AI adoption or any technology adoption as like a steep upward curve, we were at like the right before the inflection point. And I think we we still are in many ways, because AI hasn't yet been built into a lot of the tools that we use, although it's starting [00:07:30] to be. But in looking back, I did discover that we have passed one tipping point, and that is just basically generative AI adoption among US adults last year in August of 2024. November 2024, we were at about 4,546% adoption. That means 4,546% of US adults had used generative AI like ChatGPT. And in August of 2025, we got up to over [00:08:00] 55%. So we passed the midpoint. The majority of American adults have now used ChatGPT. Interestingly, it's more for non-work than for work. It's only about a third who have used it for work, whereas 45% have used it for non-work stuff. David Leary: I feel like more and more conversations I have, if somebody has a computer or their computer user, they've used ChatGPT. I don't know. I don't think I've talked to somebody that says they haven't used it, or if anybody that's in that bucket [00:08:30] doesn't have a computer, they're not a computer person. Blake Oliver: Now, what percentage of work hours are using generative AI? Last year it was 0.1 percent of work hours. In August of 2025, it was 5.7%. So we're up to a couple hours a week or maybe 30 minutes a day, depending on how you want to look at it. The people are using Gen AI, which is small, but it is substantial, and that's overall. So you can imagine that in areas where it's [00:09:00] really helpful, people are using it even more. And I also reflected on what is new in AI since last year in November. And we had a few big changes. We had the advent of multimodal AI, which for the cost seg industry is big because previously, just a year ago you couldn't upload different file types into the same chat. So I could only give it like Excel files. I couldn't also give it images [00:09:30] at the same time. The same agent or the same instance couldn't handle different file types. I want to upload a PDF. I want to upload an excel file. I want to upload images? I couldn't do that. Now I can. Another big change was AI agents, which have been more hype than reality for most of us. Although I have demonstrated on the show how I did build some AI agents and we have some that are working like that one AI agent David, that you're [00:10:00] really familiar with that takes my fireflies meeting notes and then adds them into calendar events for upcoming meetings automatically. That's an agent. David Leary: Yeah. Blake Oliver: We had the advent or the development of OpenAI's GPT five. That was a big deal because now you don't have to select which model it can intelligently decide if it wants to use a simpler model or a thinking model, depending on the complexity of your task. And then another big one, and I think this is the biggest one, was [00:10:30] this development of connectors. So now we can connect ChatGPT or Claude to our email, to our calendars, to our file storage so we don't have to upload everything. We can just ask it to go search. And I'm using this constantly now with email. Go search through my email and tell me everything about this client. Or go look through Google Drive and give me all the information about X. You can do this now on an individual level. And they're rolling this out for companies enterprises. [00:11:00] So you can search everything in your OneDrive or in your Google Drive across your company. Now. Those are big deals like those those developments are huge. All that in a single year. Amazing. But we still have a key limitation. And I think this is what's really relevant to everyone in accounting. And that has to do with this concept of the time horizon of tasks that different LMS can complete. So what [00:11:30] do I mean by this? David Leary: Okay. We talked about this before in the past. Like what used to take a human 25 minutes An LLM can now do it in 6 minutes or 8 minutes and it keeps decreasing. Right. Blake Oliver: Right. And it's it's sort of two numbers. The reason I like this measurement of like LLM capability is because we need to look at two things, right. It's not just how it's not just what kinds of tasks AI can do, but it's how accurately it can do them. Because it's not 100%. [00:12:00] So. So for instance, this group called Met-rx, I'm just going to say meter. They have this metric called a time horizon of tasks that different llms can complete. Um, and they measure this in time duration. So this chart that I've got on the screen shows different models, and it shows, um, how long [00:12:30] a task a human would take or It on the on the y axis, we see the duration that it would take this task for a human to complete. Going up from 30 minutes to 2 hours. And then on the x axis we see time and we see these LMS advancing. And in the upper right you see GPT five can now do a task that takes a human over two hours. This is a software engineering task. But here's the key point. It takes. It only does that with 50% accuracy. So [00:13:00] that's not very useful to us right. No. Because in accounting I mean even in software engineering it needs to be much more accurate than that. Blake Oliver: So they also measure this at the 80% accuracy level. And we see that GPT five can do a task that takes about 25, 26 minutes for human. And it can do that with 80% accuracy. But 80% is still not good enough for a lot of us. What does it take, or what kinds of tasks can [00:13:30] AI do With near 100% accuracy is really what we care about. And they don't have a chart for this. But in my experience, these tasks are like 2 to 5 minutes. So that's where we're at. Okay. That's the takeaway here is that right now we are at the point where you can use AI to automate a task that takes a human about 2 to 5 minutes, and it can do that almost instantly and reliably, like almost 100% accuracy. And [00:14:00] so if you want to use AI effectively in your organization, you have to identify those kinds of tasks and use AI on those kinds of tasks. And that's what I focus my presentation on. I then looked at what is the typical cost segregation study process, broke that down into 12 steps. And then in each of those steps, tried to identify one way that you could use AI to automate something. David Leary: And it kind of this is interesting because this Your [00:14:30] premise here. In theory, my brain and maps really well to cost seg because I can't. It'd be hard for ChatGPT to earn AI model to go figure out the depreciation for all the parts of the building all at once, but I could just give it a window and the receipt for the window, and it could figure it out for that individual small piece, right? So it actually I like the overlap of this. Blake Oliver: And you're on the right track, David. So in these 12 steps going from business development [00:15:00] to information gathering, proposal preparation all the way to doing the site visit and asset classification report building, technical review, audit support. Right. All these 12 steps I found there were a few tasks that you could do with AI. And I demonstrated these. And I think one of the most interesting examples is the site visit part. You could actually, Uh, take images. [00:15:30] I was able to take images of a site. So this is a hotel, and these are images that I extracted from an actual cost seg report and provided individually to AI. And I asked AI to do a photo by photo analysis and identify the assets. And the attendees were shocked to see that AI could reliably, um. Categorize a storm [00:16:00] drain inlet and give it a useful life. Figure out the depreciation category. Same thing for like pole site lighting, pole lights, patio furniture, barbecue grills. So, you know. David Leary: So I used to use an intern for you. Go to the site, you take a bunch of photos, you come back, you tell your intern, hey, go to this drive and take all these photos and create a spreadsheet of every item of depreciable item. Blake Oliver: Well, no, that's not how they do it. Okay, so you're jumping. So the way they do it is they, [00:16:30] um. Um. A qualified inspector goes to the site, and the inspector then, um, takes detailed written notes on every single item walking around the property, measuring everything, quantifying everything, adding it all up. And then those written notes are given to off square staff, who then enter that into a detailed spreadsheet. So the question is, could [00:17:00] you use AI to speed this up? Could you just take all the pictures, have AI generate the list, and then have the inspector like validate it or something? It's an interesting idea, right? It's an interesting concept, but what's incredible is just how accurate AI is at at adding up. You know what is in the photos. Could you someday walk through a site with like a 360 camera and just snap a bunch of pictures and then do all the measurement later. David Leary: Yeah, it cost SEG as a service. You upload all [00:17:30] the photos, it pumps it back a report and it's signed off. Blake Oliver: Yep. But you know, there's still limitations, right? Like, I tried doing, uh, like, measuring a pool using, uh, like agent mode and ChatGPT. And I gave it, like, a satellite photo and said, you know, measure this pool. And, you know, the first thing it did was it, like, tried to like it measured everything, including the browser, like it couldn't quite figure it out. David Leary: Every tab that you have open. That's pretty funny. Blake Oliver: But then I gave it some more instruction. It got a little better, it got a little better, but you know, it's still just not quite [00:18:00] there. So it's a mixed bag, right? There's like small tasks that AI can reliably do, but then there's bigger tasks that it can't. But here's the thing that's crazy is that this number is doubling every seven months. So every seven months the time horizon gets half what it was. So if AI can do tasks with 100 near 100% accuracy that are two minutes long right now, then in seven months it'll be four, and in 14 months it'll be eight, [00:18:30] and in 21 months it'll be 16. And in 28 months it'll be 32 minutes. Right? So, so rapidly getting better and better. If that trend continues, then in 5 to 10 years will AI be able to do almost the whole thing? That's the big question. But that relies on the trend continuing and we don't know if it will. David Leary: Well, that'll be in your 2026 Dec. You can compare to [00:19:00] your trends here today. Blake Oliver: So David that's my uh, my takeaways from the cost seg conference. Uh, other than my cold that I got, what about you? What is new with, uh, Intuit and their open AI deal? I want to hear about this. David Leary: Yeah. Let's, uh, first, I just want to welcome the live viewers. I see. Boring accountant is here. Give us some coffee emojis. Boring accountant. I don't think I actually know who you are in person, but we are at Intuit Connects conference. Somebody asked me who you were and I said I have no idea who boring accountant is, so maybe you're at the conference, [00:19:30] maybe not. Maybe we've met before, but the the listeners are in demand to want to find out who boring accountant is. So you might have to unmask. Blake Oliver: And you know what, David? Um, before we get into this Intuit story, let's thank relay, our next sponsor. David Leary: Between Blake and myself, we now have three, four, five, maybe six business entities, 20 or so checking accounts, dozens and dozens of virtual cards. It would be impossible for us to manage all of this if we weren't using relays. Our small business bank relay [00:20:00] is truly a part of the tech stack we use to run our businesses. Relay allows Blake and I to each have our own logins. We can grant access to our team and even our accountant without sharing passwords or two factor authentication codes. Relay allows us to grow and scale our banking without ever having to go to a physical branch I released recently added an account to receive inbound merchant services just a few clicks. Had to create a payroll checking account again in just a few clicks, and I instantly had access to my ACH info to give to my payroll provider. With relays virtual [00:20:30] cards, we can issue debit cards to our team around the world for all of their business expenses. I can instantly spin up a new visa debit card and set both daily and monthly spending limits. And when a team member doesn't need their card, I can freeze it until they need to use it again. Relay also has automation features to sweep money automatically from one account to another based on dates, amounts, target balances or percentages. For example, you could take your inbound payments and you could split it daily to your payroll, your sales tax payable, your operating and savings account based on your predefined rules. [00:21:00] To learn more about using relay for your firm and clients, head over to The Accounting Podcast ProAdvisor that is Accounting Today. Blake Oliver: We got a comment here from the Cuckoo Man 117. I'm an intern at a mid-size firm. Got extended an offer for January and international tax. I checked my hours and saw that I spent 10% of work time using Gen AI, and I was the only intern extended and offered a position due to my productivity compared to other [00:21:30] interns. That chart showing the AI usage climbing for work time is on point. David Leary: Wow, that's. We should capture this and you should retweet this out or reshare this. That's really interesting. Um, impress is not the right word, but we should note worthy story. We should really, uh, call that out, because I think that's people are going to be judged against coworkers that are using AI and who are more productive than you. Blake Oliver: That's right. All right. Let's talk about this Intuit OpenAI [00:22:00] deal. What are the details, David. David Leary: Yeah. So OpenAI, I'm sorry. Intuit has announced they are going to pay OpenAI $100 million a year for a multi-year deal. Um, and this is they have a video on YouTube. You could pause and go watch it. We can't play it because there's no actual words. It's just music. So it doesn't make sense for us to play this video. But in the description it says this quote unquote. Soon ChatGPT users will be able to take trusted, secure and accurate financial actions through Intuit powered apps in the ChatGPT experience. [00:22:30] So essentially, TurboTax, QuickBooks, Credit Karma, MailChimp, all that data is now going to be appearing inside of ChatGPT. So you can get your personalized, real time financial guidance without having to leave ChatGPT and open up QuickBooks in theory. Blake Oliver: So this is the connectors. This is using the connectors feature that I just mentioned right. Yeah. Like I can connect my TurboTax now to ChatGPT and just chat with AI about what's in there. David Leary: Now that's what's confusing. [00:23:00] It did not make it. It seemed like this is just kind of there now. Like you're going to use OpenAI or ChatGPT and financial stuff's going to appear like it doesn't feel like it's an opt in. Blake Oliver: Like, well, how would it how would it not be? You'd have to connect your into it. David Leary: Turbotax authorize. In theory, you should you should have to authorize OpenAI to read your QuickBooks data. Right. But that that that hasn't been really said the how isn't kind of in any of these articles. Blake Oliver: So then that's what I'm going to guess is that they're going [00:23:30] to add it to the connectors. And then you'll be able to just like with Gmail, you can connect your data into ChatGPT and work with it there. David Leary: So so instead of adding a third party app onto QuickBooks, you're just adding ChatGPT onto QuickBooks. Kind of that same model. Blake Oliver: Exactly. David Leary: Opening your data. Blake Oliver: Which is kind of genius. Like this. This is this is how ChatGPT becomes the primary place where you work. Or you ask questions because it has access to all the context. That's what's been missing. David Leary: Ai OpenAI is on this, this March, right? This we want to be the all in [00:24:00] one digital utility. And to some extent, this is the history of platforms, right? You watch what people are using your platform for and then you build it inside. You know, we've seen this intuit. This is why people you either build or buy or build or acquire, right or partner, usually it's a partner first. And Intuit did this with Melio. They partner with Melio, then eventually in Bill.com, but then they eventually built their own bill pay. It's kind of a similar model. Um, and I think this tracks to kind of what I've [00:24:30] been thinking as well. So we, we learned this with our own. You built a Nasba CPE model, right? So you have a specialized model that just does one thing. Blake Oliver: No, no. So I built a custom GPT that just focuses on Nasba CPE standards. David Leary: Yeah. And it worked really great. And then the new ChatGPT comes out and I don't have to use that one you built anymore because all the answers are just there. It just has it. Right. So as these models move forward, the specializations going away, because [00:25:00] the model is just going to do everything. And this is an example of this here. You don't have to use QuickBooks anymore if you can just use OpenAI, the chatbot. Blake Oliver: Well, I think this makes a lot more sense than every company trying to build their own chatbot. David Leary: Agree. Blake Oliver: Yeah, because that's what ChatGPT does. The best is they have the best chatbot and you connect your data to it and you're always getting the latest model, the latest updates. If every company has to do it then they're going to be behind. David Leary: Yeah. [00:25:30] David Leary: So and we'll get into it. But Intuit's CFO was interviewed and there's some reasons this is happening. But I do feel like just some thoughts I was also thinking about this that don't follow. I'm not sure this is super thought out by Intuit. If industry leaders, including you, Blake, they've been telling accountants for the last few years, accountants and accounting firms to not share your client data with AI models. And you specifically said probably not OpenAI because, [00:26:00] you know, Sam Altman's maybe not the most trustworthy guy based on, you know, some of his previous dealings, so. I accountants don't panic when clients are connecting ChatGPT. And maybe it's interacting with the data. Maybe it's changing data like our accountants. Is this just cutting the accountants out of the picture? You know, as we move forward with this and the other logic that's crazy that I don't think they thought about. So Intuit just raised prices on developers to pull data from QBO. [00:26:30] Right. So and this in a way, to stop some of these AI plays from sucking all the QBO data. So they're charging per transaction. They pull from QBO. Now it's a complete opposite. And so it's going to pay somebody else to suck that data out. It's just the logic of it is a little bit crazy for me. Blake Oliver: Well it depends on what their deal is, right? So the way that these, these deals work is, is I, as the user of ChatGPT, I authorize the connection. It comes in through the API. Now if I'm a [00:27:00] paying user then I get the data protection where my information isn't going to be used to train the models. The question is, what about all those free users, all those millions and millions and millions of people who are just using the free ChatGPT? They connect their TurboTax account. Now, what is happening to that data? Is that data being used to train the model? What deal did Intuit strike? David Leary: Yeah, that's what's not really clear. Um, and it's funny because I, I guess. Blake Oliver: Actually I would tell you this, David, if if Intuit is paying OpenAI to be able to do this, [00:27:30] then I'm going to guess they they that's their they're paying so that OpenAI will not use the data in that way. That's my guess. Like why else would they be paying for it? David Leary: Well, I can tell you why they're going to pay for it. I'll have a second article. That's a really good question. Blake Oliver: They're paying for it too, because they're using the API to do their AI agents inside of QuickBooks. Right? David Leary: Well, some of that, but the real reason why they're going to pay for it now, actually, before I jump to that, I just want to talk about a tweet I [00:28:00] did. I had a tweet in 2010 and I said, for years I used to say the number one use case in the future will be it's my data, not yours. So Intuit's basically agreeing to take all their data and let OpenAI get access to it. Now, maybe you have to authorize it. Maybe you don't. Like you said, it's you have that should be a feature of this. But it's not really clear that that's that really wasn't a garnish. So the CFO of Intuit, he did get a give an interview and [00:28:30] this is the reason they're doing it. Intuit aims to convert OpenAI's 800 million weekly active users into new customers. So so this is a this is a play for Intuit to gain new customers. Um, he also emphasized that data privacy, customer trust and AI is the core of Intuit's long term strategy. So there's lip service to this theory. Like you said, hey, we're paying. We're going to have some control over the privacy. But we also know [00:29:00] that at one time OpenAI was going to be a nonprofit. Now they're very obviously going to be a profit company. Can you trust this long term? That's a big question. Um, there's one more piece I had here that was interesting. And it just goes back to the all in one strategy of ChatGPT. They're bringing now Shopify, Etsy, PayPal, Zillow, Spotify, Canva and Expedia all inside. So in a weird [00:29:30] way, it's going to be an operating system for lots of different apps you might be using, right? Blake Oliver: You connect all the apps that have all your data to ChatGPT, and then you can simply ask questions and it goes and it finds what it needs. Yeah, this has been the gap and this is how they're plugging it. David Leary: And if you were one of these apps and we see these every week, there's another AI app that pops up that is a niche app that's built to be a chatbot that talks to QuickBooks or personal [00:30:00] Finance Bot. Are these all dead now? Blake Oliver: I think so. David Leary: As a standalone product. Blake Oliver: Because ChatGPT is just going to build that connector and do what you do, but better. All right, David, let's move on to another story that you brought to the show this week. And it has to do with Michael Burry, the famous, uh, Big Short. David Leary: He was the drummer guy in the movie. Right? Blake Oliver: The what? David Leary: He was the drummer guy in the movie, right? The drummer in the movie, [00:30:30] The Big Short. He would play the drums all violently. Blake Oliver: Oh, right. Well, but more importantly, he he called like he was shorting the market for mortgage backed securities. David Leary: But there's a lot of characters. But like, he was very distinctive in that movie. Blake Oliver: Okay. Yeah. I don't remember him playing the drums funny enough, but all right. So he was like the one. He was like one of the few people who saw the mortgage crisis coming And bet against the market and made a lot of money. And he's back with [00:31:00] another contrarian bet. Yeah. His Scion Asset Management fund is betting against some of the most celebrated artificial intelligence companies, including Palantir and Nvidia. So he is he's shorting the market. David Leary: And he's doing this. This goes back to full circle to the your we talked about depreciation for Christ's sake. Right. You want to depreciate more of that building faster. So they're doing that [00:31:30] now with the the chips. They're trying to say that these chips and these servers and the tech they're buying the useful life maybe previously was 5 to 7 years. They're saying maybe it's only 2 to 3 years. And so they're taking depreciation now which makes their income higher. Right. Am I doing the accounting. Blake Oliver: So that's Burry's allegation. Burry is saying that these AI companies that are buying these tech companies, AI companies, whatever that are, buying all these chips from Nvidia are extending the useful life [00:32:00] for depreciation purposes further than it should be, he says. David Leary: Yeah. The other way. Blake Oliver: Yeah, they're only going to be useful for like 2 to 3 years. But the industry as a whole has been moving over the last few years, uh, to five years or more. And this has a huge impact on earnings because the shorter your useful life, the more depreciation expense you have to recognize each year. The longer your [00:32:30] useful life you get, the more you get to spread it out over time. And when companies are investing billions and billions of dollars in AI chips, this can mean billions of expense. So, for example, let's say you have a $10 billion investment in AI chips as a company. Well, if your Or useful life is two years. You're recognizing 5 billion of expense in each year for two years. But let's say you make that five years. Well, now it's only $2 [00:33:00] billion of expense spread over five years. That's a $3 billion difference in the early years in those first two years. So that juices your profits. And he's saying that as an industry, the the tech world has extended the useful life, but it's not realistic. And, um, there's evidence that like the like, we know that these companies have done this meta. Amazon, Google and Microsoft [00:33:30] have all publicly extended the useful lives of servers and network equipment starting in 2022. Until now, alphabet went from 3 to 5 years, which resulted in a $3 billion profit boost, Microsoft 4 to 6 years that added 2.7 billion to their bottom line. Mehta went from 4 to 5 years, gave them a $1.5 billion boost to profit, and Amazon AWS went from 4 to 5 years, and it had a similar impact. So the entire industry has shifted [00:34:00] longer. And the SEC hasn't fought back on this. So that's that's Bernie's allegation is that the profits that we are seeing are in part due to an accounting manipulation and. David Leary: Indirectly overstate Nvidia's sales. And what I mean by that is people are going to be more willing to maybe buy Nvidia chips sooner, since they can stretch [00:34:30] out the expense of that. Blake Oliver: Well, it doesn't matter what chips you're buying. They're doing it for all the equipment, like the whole industry. The whole tech industry has done this together. They all started to increase the useful life of all this equipment, Which seems a little bit odd, right? Why? Why would they all be doing this at the same time? It's because they're making these massive CapEx investments and it's going to have a hit to profit. And [00:35:00] like you say, David, these companies manage earnings. They want to beat the street every quarter. What's the easiest way to do that? Well, let's just let's just extend the useful life a little bit. A few more years, a few more years. We'll stretch out the expense, spread it out over a longer period and we'll look more profitable. David Leary: I mean, so it's like the Big Short when you roll up these bad loans into a different one. Now look, it's good, but it's really 300 bad loans and it's just it starts to stack. It's totally the Big Short over again. Blake Oliver: Right. [00:35:30] So if you are one of those people who thinks that we are in an AI bubble in the market, then this is a strong argument that we are like, this is a I would say this is I mean, Michael Burry, You know, he was really. He was right. He was really right when it came to mortgage backed securities. And I mean, I this seems like reasonable. David Leary: Well, and I think what he's good at, he's figuring out [00:36:00] or looking where somebody's playing games and to some extent, the stacking of these mortgage securities and rolling them up into the the different funds was a game. And this is so when people play financial games there's risk right. So even though this may be technically isn't a fraud the way I think some of the headlines are out there said it was it's I don't even know if the word sketchy is the right word either. But it's a game. People are playing a game and it's going to bite us. Blake Oliver: And it's manipulation. It's earnings management, it's manipulation. It's it's [00:36:30] it's a great example of how you can use accounting to get the number you want. And it's subtle enough where auditors are not going to push back on it, especially when everyone's doing it at the same time. And that's what makes it really hard for regulators to push back on. Is all of this, uh, the there's a lot of discretion. Management has a lot of discretion when it comes to depreciation. And what is the useful life of this equipment. And [00:37:00] auditors and regulators are not going to push back unless it's really unreasonable. And if everybody's doing it, that makes it seem more reasonable. So if you're an investor, this is something to pay attention to. Brewery is obviously paying attention to it, as are we. I'm going to be curious to see what happens in the future. Do these chips end up lasting longer, or are these companies going to have to write them off in a few years? But also, like it [00:37:30] just makes me question our accounting methods like is is does it even make sense to depreciate chips like this over time? Like, like this is a concept from, uh, from railroads and factories. This idea of, like, depreciation over time, like, because tech moves so fast, can we even reliably estimate useful lives of this type of equipment? And what does that say about the reliability of earnings per share if [00:38:00] if we don't even know what the useful life is? David Leary: And it's probably not a straight line either. If you think about as the chip ages when the chips new and you just get it, the the effort and the output and the work you're giving it to do is super high volume, right? It's business critical. And then as the chip gets older, you just have it doing easier tasks, if that makes any sense. So right. It probably shouldn't even be straight line. It probably has to be some sort of diagonal. Blake Oliver: I mean, that's [00:38:30] I guess that's the problem is like we don't even really know. And so then maybe there's other metrics that we should be disclosing like that would be more useful to investors than just trying to calculate profit. For an industry where the technology is so new that there are no profits yet. Really? I mean, the AI, the chip companies are making profits because we got all these tech companies buying the AI chips, but then the the companies that are actually making gen AI applications like [00:39:00] OpenAI and anthropic are not even close to being profitable. So when will the profits flow? Because ultimately, if the profits don't flow, then the chip buying will dry up. David Leary: Yeah. Blake Oliver: All right, moving on. Uh, rippling versus deal. We got some some new, uh. I don't even know what we call it. Just like gossip. New, new dirt [00:39:30] on. David Leary: The dirt was. So you have two competitors, you have rippling. You have deal deals. Founder, CEO and maybe the father of that person hired somebody and made them get a job at rippling. And then they would go into rippling slack channels and copy paste the new deals that rippling was about to sign and give them to the company deal. And it was like corporate espionage. Yes. What's going on? Blake Oliver: Okay, deal. Hired a corporate spy. They got a [00:40:00] job at rippling, and they were giving the allegedly giving deal insight into all these enterprise deals. That deal was then going to try to steal from rippling. And rippling has been really public about this on social media with these allegations, with this lawsuit and just revealed new documents on their blog. Um, some documents got [00:40:30] unsealed by the court and rippling Publish them. They published bank statements showing that showing the flow of money from deal to the corporate spy, allegedly through the chief operating officers wife through her account. So the statements which you can go see on on deals website or on Ripplings website, show that deal transferred funds from its corporate account to Alba Barsha, the wife of COO [00:41:00] Dan Westgarth, who then forwarded the exact amount to Keith O'Brien, the corporate spy alleged corporate spy just 56 seconds later. And this was in Bash's Revolut account. Deal later made payments via cryptocurrency in order to leave no trace, but they got caught on this one. David Leary: So they so they directly pulled the money from the deal bank accounts to pay the the [00:41:30] espionage. They didn't try to mask it some other way or. Blake Oliver: Sent the money from the deal corporate bank accounts to wife of the CEO who then paid the spy. David Leary: And it wasn't even like a fake invoice, or like they paid her for some fake service. Blake Oliver: Well, so they, uh, apparently they coded it as, uh, like a business expense. David Leary: Okay. Blake Oliver: Or. Well, they don't actually know if it was classified as business expense, but rippling is speculating on this in their blog post. So anyway, here's [00:42:00] they actually have like a screenshot of the bank statement that I can show you here. David Leary: It's interesting that this moved beyond just tweets from rippling, uh, founder Conrad, uh, to Parker Conrad. Parker Conrad to they're making blog posts with this information now. It's very, very interesting. Blake Oliver: So there it is. There's the bank statement. So they're they're they're also, uh, making allegations, right? Not [00:42:30] not only is this corporate espionage, but maybe, uh, maybe deal classified this as a business expense, which would be, uh, which would be a no no as well. All right. Where are we going from here, David? David Leary: I have a quick story. New Jersey has, uh, surprisingly passed, uh, the alternative pathways bill, and now it's moving on to the Senate. So apparently in new Jersey, the bill just kind of stalled like nobody was working on it. And then the House unexpectedly just passed [00:43:00] it. Uh, or the state Senate's budget appropriation committee approved it. And now it's going to a full Senate vote. If it becomes enacted, this will be the 24th state this year. So almost half the states now, one more will be 50. 25 of the 50 states have passed alternative pathways. It's shocking. And we're not we're 11 months out and it's happened. Blake Oliver: David, let's thank our third sponsor, cloud accountant Staffing. If you're thinking about increasing your team using [00:43:30] remote team members, but you want to avoid the high pressure sales calls and commitments before you even review possible candidates. Good news you can now review potential candidates without having sales calls or any commitments. Cloud Accountant Staffing has just launched their candidate portal where you can browse resumes, intro videos, assessment and personality test results any time of the day or night. No sales call, no pressure, no commitment. The candidate portal is updated daily and all candidates are available for an interview within 24 hours. You [00:44:00] can browse candidates by both roles and experience. Want a bookkeeper that knows QuickBooks and has experience with nonprofits? You can drill down to find those candidates. How different would your firm look in 2026, and honestly, how different would your life feel if you could instantly add 40, 80, 120 hours of extra capacity every single week to explore the new Cloud Accountant staffing candidate portal? Head over to cloudaccountingpodcast.com. Cast. That's wrong. That's the wrong URL. That's our old name. Head over to The Accounting Podcast. That's [00:44:30] The Accounting Podcast. David Leary: I, I know you had a story. We didn't talk about it yet, but it ties back into why Intuit is doing this deal with OpenAI. I think you said something. People are using AI for advice. Financial advice? Blake Oliver: Yes. So let me see if I can find this story. Here it is. I spotted this in CPA [00:45:00] Practice Advisor. The headline is AI is now a first stop for tax and investing questions. A new survey found that 10% of adults acted on AI tax guidance in just 30 days, while 21% followed AI crypto advice. So adults are acting for tax advice and they are. They are asking for it and they are acting on it 10% of them. Here's the scary part. Ai gets complex tax questions wrong. Up to 50% [00:45:30] of the time, according to the Taxpayer Advocate. Bloomberg tax tested ChatGPT on complex tax scenarios and found it was incorrect 100% of the time. And why do I believe that? Well, because as we discussed at the beginning of the episode, AI is only 100% accurate or near 100% accurate on tasks that take a few minutes. So the longer the task, the less accurate it gets. A complex tax question could take [00:46:00] a human anywhere from 30 minutes to a few hours to research. And we see that ChatGPT is only 80% accurate at tasks to take a human up to 26 minutes and 50% accurate at tasks that take a human two hours or more. So small errors can also compound when you're doing tax research So you get one error at the beginning and it rolls forward. Um, another stat from the survey, 19% of Americans have already lost [00:46:30] over $100 following bad AI advice, but 27% still think AI could give them all the financial guidance they need. So good news for tax pros. You're going to be untangling tax messes for years thanks to bad AI tax advice. David Leary: And the example in the video from Intuit of the new, uh, TurboTax integration with OpenAI. Literally in the text [00:47:00] box, somebody puts in the example, we just got married and bought a house. Our total income for this year is around 175,000. What is our refund going to be? And then the video goes on to show the TurboTax as estimating your 2020 tax refund. So like literally they're they're getting in that that into it's getting into this game. Because if it's happening on OpenAI anyways already today Intuit has to be there, which makes a lot of sense. And maybe that is worth $100 million because they can't let OpenAI [00:47:30] become the new Intuit, so they got to pay to hold that market share. Blake Oliver: The new Pcob acting chair, George Bartik, did an interview with accounting today, and something really stuck out for me from this interview. Apparently, the Pcob, which is overseen audits for 20 years, has never formally defined what audit quality actually means. That's according to [00:48:00] acting Chair George Bartik, and he wants to change that. He says audit quality is like oxygen. You don't notice it until it's gone. And he he he proposes four core concepts for defining what audit quality means. One And integrity and independence. That means freedom from conflicts and maintaining skepticism to is technical competence. That means deep expertise in accounting and auditing standards, three audit performance, rigorous [00:48:30] testing and evidence gathering, and four outcomes and impacts. That means detecting material misstatements and control weaknesses. So his goal is to collaborate with stakeholders to create a unified definition. But I can't believe that after 20 years, the Public Company Accounting Oversight Board doesn't actually have a definition of what audit quality means. Isn't that their job? David Leary: But they're but they measure it and they report on it. Blake Oliver: So this. David Leary: Is. [00:49:00] Blake Oliver: The problem. Yeah. This is the problem is, is if you don't define what it is you're measuring, how can you measure it? David Leary: It feels like this is the week one job. Hey we're going to we have this new organization. We got a form. What are we going to do? How are we going to do it? The how is kind of gone missing. Blake Oliver: Yeah. And and it takes me back to our discussion with Christina Ho. Uh, one of the, uh, board members of the PCAOB who. Explained [00:49:30] that the PCAOB method for selecting audits for inspection is is flawed. At least that's my takeaway from my discussion with her. They don't select the audits randomly. They select them based on risk. And so because the audits are not selected randomly, you can't I don't believe you can actually compare the audit failure rate, the the deficiency rate [00:50:00] among different firms because it wasn't done the same way for them. And it's hard to compare it from year to year. So I feel like this is a key problem If you are a regulator, you have to have clear definitions, and then you have to have a clear way of measuring quality. You have to define audit quality and then measure it in a way that allows you to compare one period to the next. David Leary: Because if they're not reviewing a boring, plain old audit, all [00:50:30] the numbers are going to be skewed, right? You have to randomly select audits to get a true picture of somebody's rate. Blake Oliver: Now meanwhile, David, audit firm enforcement just hit historic lows. The SEC only initiated one enforcement action against auditors in the first nine months of 2025. That's down from 20 actions in the same period back in 2019. This drop in enforcement actions against auditors [00:51:00] happened right after Paul Atkins became the SEC chair in April, and Erica Williams left the PCAOB in July. So we're moving away from aggressive enforcement toward focusing on, uh, fewer cases. David Leary: Yeah. So is this a, uh, a shift of philosophy? And that's why it's less aggressive? Or is it underfunded, understaffed type of cutbacks and things like that? Blake Oliver: It's not clear to me, although the [00:51:30] Trump administration is less interested in, like, tough regulation. Yeah, it seems like so I think this would be like a a change that would be expected. Right. Fewer enforcement actions, uh, in general. All right, David, what's next? David Leary: News stories. Let's see. Sorry. Blake Oliver: No worries. I've got plenty. You want me to go? Yeah. Here's one. David Leary: I'll let you go. Blake Oliver: Here's the, uh, Financial Accounting Standards Board, the, [00:52:00] uh, the maker of GAAP. They figured out that the least standard from 2016. It was way more expensive for companies to implement than they estimated. Now, what's interesting is that they didn't actually quantify the estimate of how expensive it would be. But, um, I [00:52:30] looked into this. I did a little research on this, and I found a CFO dive article that estimated that the average cost of implementing the new standards for lease accounting was about $450,000 per company, and many companies are now paying 7000 to $35,000 annually for software to follow this new guidance. Just this new standard. So [00:53:00] FASB is now admitting that it was more expensive For public companies to do this. Well, I guess it's not just public private companies as well to do this lease accounting standard than they thought, but I couldn't find actually what their estimate was initially. So why do I bring this up? Well, there's a core principle of accounting, which is that the cost to obtain [00:53:30] information should not exceed the value we get from it. And I wonder if anyone at FASB has done an analysis to figure out the value of the additional information provided by the new lease accounting standard versus the cost to implement it, to see if the new standard actually was worth it. David Leary: And this is not like I'm trying to think this lease standard, this is not a law, but this is kind [00:54:00] of that pseudo where they have some authority to make these rules and guidelines, but it's not really a law. Blake Oliver: Well, so the SEC delegates the authority to make accounting standards to define what generally accepted accounting principles are to the Financial Accounting Standards Board, which is a nonprofit. And so they get to decide what the rules are. David Leary: So because I'm thinking like this reminds me of the boy stuff where the [00:54:30] cost of trying to create and track the boy was costing small business owners too much time versus the benefits anybody would get out of it. Correct? Like and then eventually people woke up and revoked the boy. You don't have to do it. It's kind of the same thing. Like, is anybody measuring? You can't just put rules out there and not look back at it six months later. A year later. Right. Blake Oliver: Well, it seems like that's what's happening. And I don't know if there's like the problem. I think the problem with FASB is that they don't really have an incentive [00:55:00] to simplify. They just make more rules. So accounting standards, what. David Leary: Are you going. Blake Oliver: To do? David Leary: That's your job right? Blake Oliver: Well, their job is to make rules. David Leary: More standards. Yes. Blake Oliver: It's not to take them away I guess. But really it should be to keep them simple. Because simple rules are easier to follow. And so we want rules that provide information to investors. But we also want rules that are not expensive to follow. We want to keep the cost low [00:55:30] and keep the value of the information high. And I don't know if that is part of the calculation at FASB. It doesn't seem like it is because they just keep adding rules every year. They never subtract them. Sort of like the tax code, right? The tax code never seems to get simpler. It just gets more and more complex because there's no incentive to make it simpler. No direct incentive. David Leary: I have a fun story we could go out on. Go [00:56:00] for it. Ready for that? Oh! As Blake coughs, we'll we'll get back to. I have a fun story if you want to go out on that. Go for it. So in Peru, remember Peru like, has like, these grand, uh, carvings in the desert is artwork. And, like, maybe spaceships did it. Who knows? Right. They've been there for thousands and thousands and thousands of years. Blake Oliver: Yeah. You can there those there, those lines that you can only see [00:56:30] from the air. David Leary: Peru. Blake Oliver: The Nazca. Is it Nazca? Nah. David Leary: I think ink the lines of Inca. I don't know, I'm gonna blow it. We should have did more research on that. Anyways. In Peru, in southern Peru. Blake Oliver: It's the Nazca lines, the Nazca lines, the Nazca desert in southern Peru. They were created by the Nazca civilization between 500 BC and 500 A.D.. David Leary: Yeah. Or alien spaceships drew the lines, we don't know. Well, they've actually discovered something else in southern Peru. There's a, uh. The site is [00:57:00] called Monte Sempre. They have. There's kind of like a side of a hill. In a field. There's 52, 5200 large holes in a band that's about a mile long. And when they step back and look at it, it looks like this might be a G.L. it might have been an ancient G.L., because it's built off of, um, there's an ancient record record keeping, uh, tool called the quipu, um, that Inca tribes used to use. And it's a, [00:57:30] it's basically a, a cord of knots and colors. And it was basically a ten based decimal system. That's how they would keep track of what you owed me or I owed you and, uh, census tax, stock records, it'd all be done on these courts. And essentially these holes in the ground represent one of these cords. So it's this this ancient system. And what they did is they found pollen in these holes. And the way the pollen was in these holes, it seems like it was put there purposely. And so some sort of massive large [00:58:00] GL that you probably can only see from, from, you know, up in the air, like, like the lines in the artwork, like what is this? Are you. Blake Oliver: Are you are you, are you like theorizing that this is a like some sort of accounting record or like, who's making. David Leary: Well, there's a study. Sorry, there was a study. Yes. I didn't see the holes and make this up. Sorry about that. So there was a study. Blake Oliver: Is this is this show turning into ancient aliens, David? Is that. David Leary: Is that that's what this could be? Is this like an ancient alien GL. Like just [00:58:30] like the images and the artwork. So just, uh, it's just out there. It's really interesting that they discovered these. They studied them. And this is their conclusion is it's maybe, possibly an ancient accounting system. Blake Oliver: Well, David, I think that is a great way to wrap it up. Thank you to everyone who tuned in live. If you have never caught us on YouTube, you can do that. Go to The Accounting Podcast on YouTube. Subscribe. Hit that notification button and you'll get notified when we go live, and you can join us and chat with us as we record the show. Don't [00:59:00] forget, you can earn free CPE for listening to this and past episodes of the podcast. Go to earmarked app in your web browser or download the free app on the App Store. Get all that CPE that you need before year end. It's free to sign up and free to do one a week, and you can subscribe for unlimited. Thanks everyone who joined us live! Tate Omachi Zweifel, male, 22, the QuickBooks chap. Thank you [00:59:30] for chatting with us. Hey Giles, great to see you as well. Mega tasman. We got, uh, Vamsa Reddy, the Kuku man. Boring accountant. Great to see you. See you here next week. Bye, everyone.