In this interview from the AGNT Podcast, recorded at the MCP Dev Summit in Times Square, Raphaelle d'Ornano joins theCUBE + NYSE Wired's Gemma Allen to break down the week's most market-moving developments in AI — from the accidental leak of Anthropic's 512,000-line Claude Code repository to the misread implications of Google's TurboQuant research. D'Ornano unpacks what the leaked codebase actually reveals: a sophisticated orchestration graph with Claude Code as the entry point for capturing user intent, and an indexing mechanism called the Pointer that solves context entropy by fetching information directly from external sources without inflating the context window.
The conversation also explores the widening gap between Wall Street's interpretation of AI breakthroughs and their technical reality. D'Ornano explains how Google's TurboQuant research — an inference efficiency improvement rooted in papers from a year prior — triggered a sharp selloff in memory stocks despite HBM supply constraints remaining fully intact. She outlines her investment philosophy around "architectural resilience," evaluating how companies across the AI stack are positioned against rapid innovation cycles. The discussion turns to OpenAI's 35X revenue valuation and the unexpected difficulty of its latest funding round, raising sharper questions about retail investors gaining index-fund exposure to money-losing AI companies as Anthropic, OpenAI and xAI prepare for landmark IPOs. From the mismatch between market perception and technical nuance to the democratization risks of AI investing, d'Ornano provides a frank assessment of the industry's most consequential fault lines.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: AI Agent Conference. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: AI Agent Conference.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Register For theCUBE + NYSE Wired: AI Agent Conference
Please fill out the information below. You will recieve an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for theCUBE + NYSE Wired: AI Agent Conference.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: AI Agent Conference. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: AI Agent Conference.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Sign in to gain access to theCUBE + NYSE Wired: AI Agent Conference
Please sign in with LinkedIn to continue to theCUBE + NYSE Wired: AI Agent Conference. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
AGNT Podcast Ep. 7 with Gemma Allen & Raphaëlle d'Ornano
In this interview from the AGNT Podcast, recorded at the MCP Dev Summit in Times Square, Raphaelle d'Ornano joins theCUBE + NYSE Wired's Gemma Allen to break down the week's most market-moving developments in AI — from the accidental leak of Anthropic's 512,000-line Claude Code repository to the misread implications of Google's TurboQuant research. D'Ornano unpacks what the leaked codebase actually reveals: a sophisticated orchestration graph with Claude Code as the entry point for capturing user intent, and an indexing mechanism called the Pointer that solves context entropy by fetching information directly from external sources without inflating the context window.
The conversation also explores the widening gap between Wall Street's interpretation of AI breakthroughs and their technical reality. D'Ornano explains how Google's TurboQuant research — an inference efficiency improvement rooted in papers from a year prior — triggered a sharp selloff in memory stocks despite HBM supply constraints remaining fully intact. She outlines her investment philosophy around "architectural resilience," evaluating how companies across the AI stack are positioned against rapid innovation cycles. The discussion turns to OpenAI's 35X revenue valuation and the unexpected difficulty of its latest funding round, raising sharper questions about retail investors gaining index-fund exposure to money-losing AI companies as Anthropic, OpenAI and xAI prepare for landmark IPOs. From the mismatch between market perception and technical nuance to the democratization risks of AI investing, d'Ornano provides a frank assessment of the industry's most consequential fault lines.
AGNT Podcast Ep. 7 with Gemma Allen & Raphaëlle d'Ornano
Gemma Allen
Host, theCUBE + NYSE WiredtheCUBE
HOST
Raphaelle d'Ornano
Founder & CEODecoding Discontinuity
search
Gemma Allen
>> Welcome to AGNT, the podcast for enterprise tech meets the authentic era. I'm Gemma Allen, joined by my co-host Raphaelle d'Ornano, broadcasting from the New York Stock Exchange. And every episode, we unpack how intelligent systems are reshaping companies, markets, and the way real work gets done. From Fortune 500 boardrooms to breakout upstarts, we're digging into the strategies, technologies, and people defining the next chapter of AI. Let's get into it. Raphaelle, welcome back to the NYSE. What a crazy few weeks you were in Europe, right?
Raphaelle d'Ornano
>> Right.
Gemma Allen
>> Had a good time?
Raphaelle d'Ornano
>> Had an amazing time. Sunny Europe was very nice.
Gemma Allen
>> Well, hopefully we'll have a sunny New York soon. Any day now. So the world of tech, the world of agentic, it has been a busy and chaotic couple of weeks from the media perspective. Lots of interesting headlines. Let's first talk about a very interesting headline today, which is the Altman and Musk case is kicking off in San Francisco. I saw a headline yesterday that said the judge mentioned that some of the jurors were having a hard time pretending not to have a bias against Musk. What are your thoughts?
Raphaelle d'Ornano
>> Interesting. Well, I mean, I put this in the perspective of we're having these three major IPOs that are going to rock the public markets this year. Again, Anthropic should be coming first. OpenAI, I'm not sure. We have, of course, xAI. I mean, I think all of this should be seen into that broader picture where it's one more problem for OpenAI that is having a lot of problems these times, you can say. And I mean, of course it's going to be a very dirty fight between Musk and Altman because I don't think they're very good friends, right? But I mean, I look at this from what are the implications for both companies as they approach the public markets. And again, it's just one more huge point that could have a big impact. The amounts we're talking about are like that small amounts.
Gemma Allen
>> Well, speaking of OpenAI reaching the public markets, I wrote a piece for Forbes maybe a week and a half ago about Sarah Friar, right? On the at least rumored beef for misalignment between her and Altman and how quickly they should go public. And I mean, the rumors are in the internal kind of media ring that she thinks they need to take a step back, slow things down, and he has a much faster plan in mind. I mean, she's a very credible and experienced and tenured CFO. That woman brings a lot to the table. What are your thoughts on that whole divide in terms of him racing against the clock and her saying, "Hey, slow the breaks a second. We have a lot to figure out here?" Especially as it relates to how these folks here at Wall Street will respond to spend and to revenue versus spend.
Raphaelle d'Ornano
>> Well, I think that the reason why Altman is racing to the IPO is you have a story of narrative here that is super interesting. The company that goes first is the company that is going to set the narrative of what is the moat for a foundational model company, what matters, what to track, and kind of what does the species look like. And so if Anthropic goes first, of course it has huge implications because then OpenAI is compared versus Anthropic and not vice versa. So I think from Sam Altman's perspective, he knows this very well. I mean, and he wants to rush into this. Of course, Sarah, I mean, I think she's right in the sense that these models are very complicated to understand. They require a ton of nuance. And I think I've said this repeatedly, like Anthropic, xAI, MiniMax, DeepSeek, whatever, those are all foundation model companies, but none of them has a similar business model. Some are B2B, some are B2C, some are like specialized on coding, some not as much. I mean, it's not one category, it's multiple categories. So that matters a lot. And it is not understood by the public markets because, I mean, it's the first time we have these companies going to IPO. So there needs to be like a clear narrative that is built on top of a credible financial story and the credible financial story of you don't know how you're going to monetize all of that while at the same time spending trillions in compute might be a bit of a problem for the public markets at a point where we're having a lot of nervosity around the spend. I mean, today we have the earnings from the big hyperscalers, right? The semis today are ... I mean, the semis have been on a terror over the past weeks and that is performing like crazy, but there is this like very strong worry around the corner where when the Wall Street Journal says, "Oh, OpenAI missed their numbers. You have all of the stocks linked to OpenAI that tumble." We don't even know why, but people were very scared. So I think what is at stake is that it's going to be important, the message that comes out, who says it first, and the numbers that back it, I think it's going to be a very messy story.
Gemma Allen
>> Well, another interesting story this week is DeepSeek in terms of what that has, that whole open source China versus the US, how much of a threat is this? Is DeepSeek two companies like OpenAI and Anthropic? Are we going to see more DeepSeek embedded models here in Fortune 500s on US soil? How quickly is that going to happen? What are the realities around that, right? What would it actually mean for companies to be able to deploy DeepSeek?
I mean, the commentary is that right now, up until this point, actual deployment of DeepSeek and enterprises is below 2%. In fact, they're saying that less than 2% of Fortune 500 have even touched the technology. Perhaps that's because of the pros and cons of open source and the perspective of what it would actually mean to ... You have a choice, send your data to China and it comes back to you or to actually have open weights in your own environment, right? And the infrastructure spend on that is colossal. And I mean, if the hyperscalers can't figure it out, how could we expect a Fortune 500 to figure it out in the enterprise space? What are your thoughts? How real do you think this threat is?
Raphaelle d'Ornano
>> Well, look, I've been following the Chinese model companies for a long time now. And of course, when DeepSeek came out last year, it was like a very important event with, at that time, huge implications on the public markets the day it came out. Everyone was very panicked, which did not happen this time, which is, I mean, not surprising, but interesting, I would say. I think what happened last week was not just DeepSeek. It was MiniMax, Zhipu, like all of those models that came in a flurry on top of the closed models from Anthropic with Cloud Opus 4.7 and GPT 5.5, both that came out very recently. So it's like in one week or like in two weeks, you have seven new models that are all very good. I think what we have seen, which is interesting is that the moat has moved from the pure model capabilities to how these models are orchestrated in the enterprise and how they have built in agentic capabilities. And I think the new ... The takeaway from last week is really, and Kimi K2 was interesting, I would say on that point, it's that they are architected around deploying agentic AI in the enterprise and their moat is what OpenAI and Anthropic were building last year. When in May last year, when we were asking ourselves what makes a moat, we said the moat is not in the technical power, it's in the orchestration. And this was one year ago. And one year later, the Chinese model companies are doing exactly that. So what does that mean for OpenAI and Anthropic? It means that now they have to go one layer up because that's not their moat anymore. And I think the Chinese model companies are at par with OpenAI Anthropic in, again, those orchestration capabilities. So does that mean that Anthropic has to go into applications? Does that mean that OpenAI has to go into applications? It's like those companies are in a race and it has just accelerated their need to go higher, better to be able to differentiate themselves. Remember, Anthropic launched two weeks ago, the managed agent service by which they're billing the runtime on top of the actual APIs, which I think is interesting because it's a new layer of revenue. So it's one evidence of, okay, Anthropic still has this orchestration moat, but what is interesting is not the Chinese companies for the Chinese companies. It's how the Chinese companies are setting the clock for the US closed source companies at the time where they go to IPO and at the time where they have to face this huge compute crunch, which is a problem and which is one of the reasons why at some point there was a pause on the semis in the public markets because everyone was like, "Oh, well, okay, Anthropic has caused this massive SaaS sucker, however we call that, but if Anthropic doesn't have enough compute. Well, is Anthropic goes red?" So we don't know. So I think that what this shows is, to your point, no one is really actively deploying DeepSeek in the enterprise. Though, there are a lot of platforms that allow to deploy those without having to run this on your own servers. And those companies that are in the private markets, Fireworks being one of them, are like on fire in the private market. So I think there's a way to go around that without all of the architectural constraints of deploying that without the hyperscalers. So I think that's being like managed to some extent. But for me, what I really, I would say, enjoy in this debate between the Chinese and the US companies is how it pushes the competition always one step above.
Gemma Allen
>> Well, talking of pushing the competition, we have seen Jensen go on the record and say, "We need friendly relationships with China. We want geopolitical positive relations as much as we can allow." It was also interesting that some of the commentary around DeepSeek also mentioned to Huawei and how that is potentially now, in some respects, maybe encroaching on the NVIDIA moat. You said earlier, OpenAI miss a customer revenue forecast and the market goes crazy, right? People panic. They think anyone who's aligned to open AI, "Oh my God, what does the future hold?" But similarly for NVIDIA, I mean, that company is just on the up, up, up all the time. And there are evidence, I mean, perhaps it's yet to be fully seen that maybe there is Chinese solutions too from the perspective of what NVIDIA have built their moat around, right? So it's an interesting time, I think, in terms of how things are being, I guess at least the narrative has been led and the response has been ingested. Here, I mean, what are your thoughts? Do you think NVIDIA also has some kind of rougher days on the horizon?
Raphaelle d'Ornano
>> Look, I think NVIDIA is a tremendous company with a leader that is, I mean, the most visionary leader, we can say that they just passed a five trillion mark, I mean, which in itself-
Gemma Allen
>> Incredible....
Raphaelle d'Ornano
>> is also a huge event. The fact there's an oligopoly in GPUs is not new and that is only going to accelerate. I think that's been the case for the previous months. And it's come in the picture that NVIDIA doesn't have a monopoly. It's multiple players that are in that market. I think we've also seen a very interesting trend of the ratio of CPUs to GPUs and agentic workloads that is completely going in the favor of more CPUs per GPUs. And AMD released an interesting study on that point recently, which is another change of ... I mean, GPUs are fundamental, but CPUs are very fundamental as you deploy agentic AI. I think to the point on China, Huawei is still not there and there's still a gap. However, the release of the Chinese model companies that we saw last week were a lot around architectural innovation, around, for example, like mixture of experts or around how do you bake into the model architectural efficiency to allow inference to run using less compute. On the training side, we had the manifold constraint that we had talked, which was one of the key points of the DeepSeek model. I think to be honest, we haven't had that breakthrough yet. So the need in terms of huge amounts of GPUs that come from NVIDIA is still very present. And I would say that's a consensus view. I think however, that things are approaching ... Maybe something is around the corner that is going to impact this whole hardware space. I don't know if it exists yet, but my guess would be that something is going to happen in how these models are built, which could change the game very quickly. I mean, there's going to be a ton of demand for NVIDIA. Again, they're not alone anymore. I would say that, that path accelerates. I would say there's going to be more and more choices putting aside the CPU, GPU debate, which is another one. So I would think the volume compensates, but NVIDIA is not the only person in town anymore.
Gemma Allen
>> I mean, another interesting headline this week surrounding the whole debate was while tech is in somewhat of a nuanced position geopolitically, there has been a push in the financial side of the US to see more and more like Chinese investment. These different funds led by companies like Goldman Sachs are looking very actively on China and maybe it's because of inflation. There's a whole load of macroeconomic factors, but I mean, it's evident that there is a huge, huge war on tech arising from these two continents and perhaps people follow the money, right? That just tends to be how things go. So I think it's interesting to understand, okay, while there is a kind of a geopolitical situation over here is somewhat of it maybe like a form of pantomime while markets move in a particular direction as they always do.
Raphaelle d'Ornano
>> I mean, I think to comment on that point, I would talk about the Manus acquisition being off by Meta being blocked by the Chinese government. I think that's kind of been under the radar, but that was pending. I mean, when Meta bought Manus, that was a big deal. I mean, Manus is like one of the first agentic players that really came out. And I mean, to my best understanding, this acquisition has been blocked because the Chinese government has been like, "Oh, you cannot just go to Singapore and do that." I mean, that's a big deal in what is happening and that shows that, I mean, there is geopolitical tension right in that deal.
Gemma Allen
>> I mean, I was in some ways surprised that people were surprised at that response by China. I almost expected it. I assumed, and I read headlines, "Oh, Manus researchers been working alongside Meta researchers in a lab since December." I'm like, "Well, in some respects, so what?" There had to have been some sort of expectation that the Chinese are going to look at this and think, "Would we allow this happen on US soil? Would we?"
Raphaelle d'Ornano
>> I mean, I agree. It's a very fair point.
Gemma Allen
>> It's practical, right? Above all else. So moving on, looking at stocks broadly, and I know we talk a lot here on this show about SaaS stocks in particular. I mean, it has been an interesting first four months for SaaS, okay? Some of these companies have taken huge hammering since the start of the year. We think about companies like ServiceNow, Workday, Monday.com, Salesforce, there is a lot of concern as to like how real or how short-lived these challenges are, right? You think these are going to continue throughout the end of the year? And again, what are your thoughts on this whole debate around short-term capability doesn't necessarily replace longer term context in terms of what these companies have and can deliver for the markets?
Raphaelle d'Ornano
>> Right. So my first point would be that the SaaSpocalypse is not over and it's done because there's a war in around or because there's other problems on like inflation, the job market, whatever, that I mean the fundamental problem of what happens to the terminal value of a software company in this new agentic paradigm, that is not over. That is only starting. We're seeing companies that are actively responding to that. Like for example, Salesforce, I think two weeks ago, announced its biggest restructuring in 27 years since the company existed when they announced Headless 360 by which they're saying, "Well, we exist also through APIs and we don't need the interface." I mean, there's a lot of other announcements that came in that announcement that they made, but that was like a very important milestone. We acknowledge that things are changing and that the agentic era is here, so I think that, that was an interesting one. Comparably, when we look at the ServiceNow earnings where ServiceNow was, I would say hammered by the stock market two weeks ago, last week, sorry, what happened is that they announced incredible results. They actually had very strong growth in bookings, very strong growth in ARR. But for example, they said, "Well, our gross margin is lower," but that is not a surprise. Every single gross margin in SaaS is going to be lower as you incorporate AI and agentic AI in your solution. That's the most obvious thing, but yet investors were very surprised, "Oh my God, the gross margin is going down."
I can tell you right now, the gross margin of every single company in the software universe is going to go down over the next month as agentic AI reshapes the profit and loss statement of every single software company. So I mean, it's kind of unfair for ServiceNow on that, they get punished for like the most obvious thing in finance. Of course, there were other problems, but it was funny to see that the media was picking up that they were disappointed by the gross margins when this is perfectly normal in my view. So what that shows is that there's a ton of nervosity where we don't know what to look for. What is clear is that when ServiceNow says, "Oh, we have a ton of AI revenue," no one cares about that because that is not what the market is looking for. The market is looking for, "What is your terminal value in the age of agentic AI? Are you still around over the next 5 years, the next 10 years? And how do I get the maximum reassurance that this is happening?"
So I think every single SaaS company today, SaaS or software company, I mean, whatever, should be careful about what do they mean? What is their value in this new paradigm in which the rules have changed, in which the P&L has changed? And again, bye-bye 80% gross margins, that is not happening anymore, but that's fine because you also have new network effects. So you have maybe less margins, but much better growth, so it's a trade off. You prefer growth, do you prefer margins, so let's see. It's like this is new. We need new mental models and just saying, "Oh, we have AI revenue," is not going to reassure any investors. So in a nutshell, we are still in the SaaSpocalypse. Companies are pivoting to, we need to acknowledge this new reality. A lot will emerge better, stronger, I hope. Some will not emerge stronger at all or will just not emerge. And by the way, it's the same thing in the private markets. We saw what happened with the company from Thoma Bravo that was given to the lenders. I'm not a pessimistic person, but I think that's just the beginning. And we had that discussion before on the show. The private markets are like, "Oh, the public market companies are getting hammered, that's maybe normal, Monday.com, et cetera, et cetera." I mean, why would the private market companies that are slower, that are smaller most of the times and that often do not have the same growth rates for some of them, why would they be not at risk at only public market companies? It makes no sense. So of course, the private market companies are also going to be in big trouble and maybe to the same proportions as the public market companies, I don't know, but I mean, I think it's going to hurt. And so we are just at the beginning at the first 1% of what is happening in my view.
Gemma Allen
>> So before we close, I think it's interesting too, to bring up this whole topic of horizontal integration, right? And what it has meant in the SaaS space versus what it currently means right now in the agentic space. We have a lot of startups on this show. We speak to startups all day every day, and there are certainly a number of companies that are really cutting their teeth and making an impact, taking a very specific swim lane, like a very clear beat. We're not trying to be everything fintech. We're just trying to be an agentic solution for underwriting. That's one great example, right? And that is a company that is essentially staking a bet saying, "Hey, we are going to be a horizontally integrated part of a stock." They're under the same threat by Claude, by any of the large LLM or frontier models that the SaaS universe is, okay? But perhaps part of this is that the SaaS response to some of these horizontal, high value add parts of their integrated stock offering hasn't really been messaged well, because there has been a lot of value added in very specific and nuanced spaces within set parameters for a very long time that are now not just under threat by frontier models, but also under threat by other horizontally integrated agentic players, right? The response there, even the kind of differentiator there of understanding value versus potential value, it just seems as though the messaging is a little bit weak.
Raphaelle d'Ornano
>> Well, I mean, again, there was this article that was very interesting, written by one of the most prominent venture capital investors, which said, "Well, the new TAM of software is services because now ..." And when you look at the batch from YC Combinator that was announced this week, you see all of these companies that are building agents to serve a certain number of cases across industries, across functions. Like you have your new plumber, your new architect, your new, I don't know, whatever function you take, you imagine one and you have an agent that is going to do that. And so you have the software that says, "I'm selling you not software. I'm selling you a business outcome with those agents." For me, that is not very new. I don't think that's ... Apparently, the market is discovering that there's an expansion for what software can do. I think that we need to take a step back on that. And what software can do is be the secure environment to actually hold all of these agents that are going to be doing plumbing, architecting, translation, you name it. And the function of software is going to be kind of this playground in which you actually host in a secure way in which the agents can talk together in a non-messy way without the huge security breaches. Look at the Cursor incident that happened this weekend. Oh my God. Companies are not going to be deploying agentic AI at scale when they know that all of their databases can be erased in nine seconds. That's not something that is going to accelerate any of this transition. So I actually think that, yes, there is an expansion of TAM in how software goes into services and starts performing business functions. I think that that is not new and that we have been talking about that over the past year. I think what is new is this new role of software in which there's an additional role for these companies that are being destroyed by the public markets is they can be the secure environment on which these agents are held, in which the agents talk together, in which the agents connect to the data and make this a reality in the enterprise and the word matters. And I think that's another avenue of growth that is being open for some of these big software names. So I'm very optimistic around what software can become. I think that the notion of SaaS as in SaaS 2.0 is over because that is not how software is going to be consumed. And again, the profit and loss statement is going to change fundamentally, but there's a new reality that is starting to emerge that will be defined over the next month, which for the companies that actually succeed will be way bigger with a way bigger time, with perhaps better unit economies. And that is starting to be built out, and it's very exciting to see what is happening.
Gemma Allen
>> Well, Raphaelle, I think we need to leave it there for this week. I know you have an exciting couple of weeks ahead too, so look forward to having you back here next week, and I'm sure there'll be lots to discuss in the world of agentic AI and all things defensibility, offensibility, and where exactly we're headed. I'm Gemma Allen joined by Raphaelle d'Ornano. This is AGNT. Thanks for watching.
AGNT Podcast Ep. 7 with Gemma Allen & Raphaëlle d'Ornano
search
Gemma Allen
>> Welcome to AGNT, the podcast for enterprise tech meets the authentic era. I'm Gemma Allen, joined by my co-host Raphaelle d'Ornano, broadcasting from the New York Stock Exchange. And every episode, we unpack how intelligent systems are reshaping companies, markets, and the way real work gets done. From Fortune 500 boardrooms to breakout upstarts, we're digging into the strategies, technologies, and people defining the next chapter of AI. Let's get into it. Raphaelle, welcome back to the NYSE. What a crazy few weeks you were in Europe, right?
Raphaelle d'Ornano
>> Right.
Gemma Allen
>> Had a good time?
Raphaelle d'Ornano
>> Had an amazing time. Sunny Europe was very nice.
Gemma Allen
>> Well, hopefully we'll have a sunny New York soon. Any day now. So the world of tech, the world of agentic, it has been a busy and chaotic couple of weeks from the media perspective. Lots of interesting headlines. Let's first talk about a very interesting headline today, which is the Altman and Musk case is kicking off in San Francisco. I saw a headline yesterday that said the judge mentioned that some of the jurors were having a hard time pretending not to have a bias against Musk. What are your thoughts?
Raphaelle d'Ornano
>> Interesting. Well, I mean, I put this in the perspective of we're having these three major IPOs that are going to rock the public markets this year. Again, Anthropic should be coming first. OpenAI, I'm not sure. We have, of course, xAI. I mean, I think all of this should be seen into that broader picture where it's one more problem for OpenAI that is having a lot of problems these times, you can say. And I mean, of course it's going to be a very dirty fight between Musk and Altman because I don't think they're very good friends, right? But I mean, I look at this from what are the implications for both companies as they approach the public markets. And again, it's just one more huge point that could have a big impact. The amounts we're talking about are like that small amounts.
Gemma Allen
>> Well, speaking of OpenAI reaching the public markets, I wrote a piece for Forbes maybe a week and a half ago about Sarah Friar, right? On the at least rumored beef for misalignment between her and Altman and how quickly they should go public. And I mean, the rumors are in the internal kind of media ring that she thinks they need to take a step back, slow things down, and he has a much faster plan in mind. I mean, she's a very credible and experienced and tenured CFO. That woman brings a lot to the table. What are your thoughts on that whole divide in terms of him racing against the clock and her saying, "Hey, slow the breaks a second. We have a lot to figure out here?" Especially as it relates to how these folks here at Wall Street will respond to spend and to revenue versus spend.
Raphaelle d'Ornano
>> Well, I think that the reason why Altman is racing to the IPO is you have a story of narrative here that is super interesting. The company that goes first is the company that is going to set the narrative of what is the moat for a foundational model company, what matters, what to track, and kind of what does the species look like. And so if Anthropic goes first, of course it has huge implications because then OpenAI is compared versus Anthropic and not vice versa. So I think from Sam Altman's perspective, he knows this very well. I mean, and he wants to rush into this. Of course, Sarah, I mean, I think she's right in the sense that these models are very complicated to understand. They require a ton of nuance. And I think I've said this repeatedly, like Anthropic, xAI, MiniMax, DeepSeek, whatever, those are all foundation model companies, but none of them has a similar business model. Some are B2B, some are B2C, some are like specialized on coding, some not as much. I mean, it's not one category, it's multiple categories. So that matters a lot. And it is not understood by the public markets because, I mean, it's the first time we have these companies going to IPO. So there needs to be like a clear narrative that is built on top of a credible financial story and the credible financial story of you don't know how you're going to monetize all of that while at the same time spending trillions in compute might be a bit of a problem for the public markets at a point where we're having a lot of nervosity around the spend. I mean, today we have the earnings from the big hyperscalers, right? The semis today are ... I mean, the semis have been on a terror over the past weeks and that is performing like crazy, but there is this like very strong worry around the corner where when the Wall Street Journal says, "Oh, OpenAI missed their numbers. You have all of the stocks linked to OpenAI that tumble." We don't even know why, but people were very scared. So I think what is at stake is that it's going to be important, the message that comes out, who says it first, and the numbers that back it, I think it's going to be a very messy story.
Gemma Allen
>> Well, another interesting story this week is DeepSeek in terms of what that has, that whole open source China versus the US, how much of a threat is this? Is DeepSeek two companies like OpenAI and Anthropic? Are we going to see more DeepSeek embedded models here in Fortune 500s on US soil? How quickly is that going to happen? What are the realities around that, right? What would it actually mean for companies to be able to deploy DeepSeek?
I mean, the commentary is that right now, up until this point, actual deployment of DeepSeek and enterprises is below 2%. In fact, they're saying that less than 2% of Fortune 500 have even touched the technology. Perhaps that's because of the pros and cons of open source and the perspective of what it would actually mean to ... You have a choice, send your data to China and it comes back to you or to actually have open weights in your own environment, right? And the infrastructure spend on that is colossal. And I mean, if the hyperscalers can't figure it out, how could we expect a Fortune 500 to figure it out in the enterprise space? What are your thoughts? How real do you think this threat is?
Raphaelle d'Ornano
>> Well, look, I've been following the Chinese model companies for a long time now. And of course, when DeepSeek came out last year, it was like a very important event with, at that time, huge implications on the public markets the day it came out. Everyone was very panicked, which did not happen this time, which is, I mean, not surprising, but interesting, I would say. I think what happened last week was not just DeepSeek. It was MiniMax, Zhipu, like all of those models that came in a flurry on top of the closed models from Anthropic with Cloud Opus 4.7 and GPT 5.5, both that came out very recently. So it's like in one week or like in two weeks, you have seven new models that are all very good. I think what we have seen, which is interesting is that the moat has moved from the pure model capabilities to how these models are orchestrated in the enterprise and how they have built in agentic capabilities. And I think the new ... The takeaway from last week is really, and Kimi K2 was interesting, I would say on that point, it's that they are architected around deploying agentic AI in the enterprise and their moat is what OpenAI and Anthropic were building last year. When in May last year, when we were asking ourselves what makes a moat, we said the moat is not in the technical power, it's in the orchestration. And this was one year ago. And one year later, the Chinese model companies are doing exactly that. So what does that mean for OpenAI and Anthropic? It means that now they have to go one layer up because that's not their moat anymore. And I think the Chinese model companies are at par with OpenAI Anthropic in, again, those orchestration capabilities. So does that mean that Anthropic has to go into applications? Does that mean that OpenAI has to go into applications? It's like those companies are in a race and it has just accelerated their need to go higher, better to be able to differentiate themselves. Remember, Anthropic launched two weeks ago, the managed agent service by which they're billing the runtime on top of the actual APIs, which I think is interesting because it's a new layer of revenue. So it's one evidence of, okay, Anthropic still has this orchestration moat, but what is interesting is not the Chinese companies for the Chinese companies. It's how the Chinese companies are setting the clock for the US closed source companies at the time where they go to IPO and at the time where they have to face this huge compute crunch, which is a problem and which is one of the reasons why at some point there was a pause on the semis in the public markets because everyone was like, "Oh, well, okay, Anthropic has caused this massive SaaS sucker, however we call that, but if Anthropic doesn't have enough compute. Well, is Anthropic goes red?" So we don't know. So I think that what this shows is, to your point, no one is really actively deploying DeepSeek in the enterprise. Though, there are a lot of platforms that allow to deploy those without having to run this on your own servers. And those companies that are in the private markets, Fireworks being one of them, are like on fire in the private market. So I think there's a way to go around that without all of the architectural constraints of deploying that without the hyperscalers. So I think that's being like managed to some extent. But for me, what I really, I would say, enjoy in this debate between the Chinese and the US companies is how it pushes the competition always one step above.
Gemma Allen
>> Well, talking of pushing the competition, we have seen Jensen go on the record and say, "We need friendly relationships with China. We want geopolitical positive relations as much as we can allow." It was also interesting that some of the commentary around DeepSeek also mentioned to Huawei and how that is potentially now, in some respects, maybe encroaching on the NVIDIA moat. You said earlier, OpenAI miss a customer revenue forecast and the market goes crazy, right? People panic. They think anyone who's aligned to open AI, "Oh my God, what does the future hold?" But similarly for NVIDIA, I mean, that company is just on the up, up, up all the time. And there are evidence, I mean, perhaps it's yet to be fully seen that maybe there is Chinese solutions too from the perspective of what NVIDIA have built their moat around, right? So it's an interesting time, I think, in terms of how things are being, I guess at least the narrative has been led and the response has been ingested. Here, I mean, what are your thoughts? Do you think NVIDIA also has some kind of rougher days on the horizon?
Raphaelle d'Ornano
>> Look, I think NVIDIA is a tremendous company with a leader that is, I mean, the most visionary leader, we can say that they just passed a five trillion mark, I mean, which in itself-
Gemma Allen
>> Incredible....
Raphaelle d'Ornano
>> is also a huge event. The fact there's an oligopoly in GPUs is not new and that is only going to accelerate. I think that's been the case for the previous months. And it's come in the picture that NVIDIA doesn't have a monopoly. It's multiple players that are in that market. I think we've also seen a very interesting trend of the ratio of CPUs to GPUs and agentic workloads that is completely going in the favor of more CPUs per GPUs. And AMD released an interesting study on that point recently, which is another change of ... I mean, GPUs are fundamental, but CPUs are very fundamental as you deploy agentic AI. I think to the point on China, Huawei is still not there and there's still a gap. However, the release of the Chinese model companies that we saw last week were a lot around architectural innovation, around, for example, like mixture of experts or around how do you bake into the model architectural efficiency to allow inference to run using less compute. On the training side, we had the manifold constraint that we had talked, which was one of the key points of the DeepSeek model. I think to be honest, we haven't had that breakthrough yet. So the need in terms of huge amounts of GPUs that come from NVIDIA is still very present. And I would say that's a consensus view. I think however, that things are approaching ... Maybe something is around the corner that is going to impact this whole hardware space. I don't know if it exists yet, but my guess would be that something is going to happen in how these models are built, which could change the game very quickly. I mean, there's going to be a ton of demand for NVIDIA. Again, they're not alone anymore. I would say that, that path accelerates. I would say there's going to be more and more choices putting aside the CPU, GPU debate, which is another one. So I would think the volume compensates, but NVIDIA is not the only person in town anymore.
Gemma Allen
>> I mean, another interesting headline this week surrounding the whole debate was while tech is in somewhat of a nuanced position geopolitically, there has been a push in the financial side of the US to see more and more like Chinese investment. These different funds led by companies like Goldman Sachs are looking very actively on China and maybe it's because of inflation. There's a whole load of macroeconomic factors, but I mean, it's evident that there is a huge, huge war on tech arising from these two continents and perhaps people follow the money, right? That just tends to be how things go. So I think it's interesting to understand, okay, while there is a kind of a geopolitical situation over here is somewhat of it maybe like a form of pantomime while markets move in a particular direction as they always do.
Raphaelle d'Ornano
>> I mean, I think to comment on that point, I would talk about the Manus acquisition being off by Meta being blocked by the Chinese government. I think that's kind of been under the radar, but that was pending. I mean, when Meta bought Manus, that was a big deal. I mean, Manus is like one of the first agentic players that really came out. And I mean, to my best understanding, this acquisition has been blocked because the Chinese government has been like, "Oh, you cannot just go to Singapore and do that." I mean, that's a big deal in what is happening and that shows that, I mean, there is geopolitical tension right in that deal.
Gemma Allen
>> I mean, I was in some ways surprised that people were surprised at that response by China. I almost expected it. I assumed, and I read headlines, "Oh, Manus researchers been working alongside Meta researchers in a lab since December." I'm like, "Well, in some respects, so what?" There had to have been some sort of expectation that the Chinese are going to look at this and think, "Would we allow this happen on US soil? Would we?"
Raphaelle d'Ornano
>> I mean, I agree. It's a very fair point.
Gemma Allen
>> It's practical, right? Above all else. So moving on, looking at stocks broadly, and I know we talk a lot here on this show about SaaS stocks in particular. I mean, it has been an interesting first four months for SaaS, okay? Some of these companies have taken huge hammering since the start of the year. We think about companies like ServiceNow, Workday, Monday.com, Salesforce, there is a lot of concern as to like how real or how short-lived these challenges are, right? You think these are going to continue throughout the end of the year? And again, what are your thoughts on this whole debate around short-term capability doesn't necessarily replace longer term context in terms of what these companies have and can deliver for the markets?
Raphaelle d'Ornano
>> Right. So my first point would be that the SaaSpocalypse is not over and it's done because there's a war in around or because there's other problems on like inflation, the job market, whatever, that I mean the fundamental problem of what happens to the terminal value of a software company in this new agentic paradigm, that is not over. That is only starting. We're seeing companies that are actively responding to that. Like for example, Salesforce, I think two weeks ago, announced its biggest restructuring in 27 years since the company existed when they announced Headless 360 by which they're saying, "Well, we exist also through APIs and we don't need the interface." I mean, there's a lot of other announcements that came in that announcement that they made, but that was like a very important milestone. We acknowledge that things are changing and that the agentic era is here, so I think that, that was an interesting one. Comparably, when we look at the ServiceNow earnings where ServiceNow was, I would say hammered by the stock market two weeks ago, last week, sorry, what happened is that they announced incredible results. They actually had very strong growth in bookings, very strong growth in ARR. But for example, they said, "Well, our gross margin is lower," but that is not a surprise. Every single gross margin in SaaS is going to be lower as you incorporate AI and agentic AI in your solution. That's the most obvious thing, but yet investors were very surprised, "Oh my God, the gross margin is going down."
I can tell you right now, the gross margin of every single company in the software universe is going to go down over the next month as agentic AI reshapes the profit and loss statement of every single software company. So I mean, it's kind of unfair for ServiceNow on that, they get punished for like the most obvious thing in finance. Of course, there were other problems, but it was funny to see that the media was picking up that they were disappointed by the gross margins when this is perfectly normal in my view. So what that shows is that there's a ton of nervosity where we don't know what to look for. What is clear is that when ServiceNow says, "Oh, we have a ton of AI revenue," no one cares about that because that is not what the market is looking for. The market is looking for, "What is your terminal value in the age of agentic AI? Are you still around over the next 5 years, the next 10 years? And how do I get the maximum reassurance that this is happening?"
So I think every single SaaS company today, SaaS or software company, I mean, whatever, should be careful about what do they mean? What is their value in this new paradigm in which the rules have changed, in which the P&L has changed? And again, bye-bye 80% gross margins, that is not happening anymore, but that's fine because you also have new network effects. So you have maybe less margins, but much better growth, so it's a trade off. You prefer growth, do you prefer margins, so let's see. It's like this is new. We need new mental models and just saying, "Oh, we have AI revenue," is not going to reassure any investors. So in a nutshell, we are still in the SaaSpocalypse. Companies are pivoting to, we need to acknowledge this new reality. A lot will emerge better, stronger, I hope. Some will not emerge stronger at all or will just not emerge. And by the way, it's the same thing in the private markets. We saw what happened with the company from Thoma Bravo that was given to the lenders. I'm not a pessimistic person, but I think that's just the beginning. And we had that discussion before on the show. The private markets are like, "Oh, the public market companies are getting hammered, that's maybe normal, Monday.com, et cetera, et cetera." I mean, why would the private market companies that are slower, that are smaller most of the times and that often do not have the same growth rates for some of them, why would they be not at risk at only public market companies? It makes no sense. So of course, the private market companies are also going to be in big trouble and maybe to the same proportions as the public market companies, I don't know, but I mean, I think it's going to hurt. And so we are just at the beginning at the first 1% of what is happening in my view.
Gemma Allen
>> So before we close, I think it's interesting too, to bring up this whole topic of horizontal integration, right? And what it has meant in the SaaS space versus what it currently means right now in the agentic space. We have a lot of startups on this show. We speak to startups all day every day, and there are certainly a number of companies that are really cutting their teeth and making an impact, taking a very specific swim lane, like a very clear beat. We're not trying to be everything fintech. We're just trying to be an agentic solution for underwriting. That's one great example, right? And that is a company that is essentially staking a bet saying, "Hey, we are going to be a horizontally integrated part of a stock." They're under the same threat by Claude, by any of the large LLM or frontier models that the SaaS universe is, okay? But perhaps part of this is that the SaaS response to some of these horizontal, high value add parts of their integrated stock offering hasn't really been messaged well, because there has been a lot of value added in very specific and nuanced spaces within set parameters for a very long time that are now not just under threat by frontier models, but also under threat by other horizontally integrated agentic players, right? The response there, even the kind of differentiator there of understanding value versus potential value, it just seems as though the messaging is a little bit weak.
Raphaelle d'Ornano
>> Well, I mean, again, there was this article that was very interesting, written by one of the most prominent venture capital investors, which said, "Well, the new TAM of software is services because now ..." And when you look at the batch from YC Combinator that was announced this week, you see all of these companies that are building agents to serve a certain number of cases across industries, across functions. Like you have your new plumber, your new architect, your new, I don't know, whatever function you take, you imagine one and you have an agent that is going to do that. And so you have the software that says, "I'm selling you not software. I'm selling you a business outcome with those agents." For me, that is not very new. I don't think that's ... Apparently, the market is discovering that there's an expansion for what software can do. I think that we need to take a step back on that. And what software can do is be the secure environment to actually hold all of these agents that are going to be doing plumbing, architecting, translation, you name it. And the function of software is going to be kind of this playground in which you actually host in a secure way in which the agents can talk together in a non-messy way without the huge security breaches. Look at the Cursor incident that happened this weekend. Oh my God. Companies are not going to be deploying agentic AI at scale when they know that all of their databases can be erased in nine seconds. That's not something that is going to accelerate any of this transition. So I actually think that, yes, there is an expansion of TAM in how software goes into services and starts performing business functions. I think that that is not new and that we have been talking about that over the past year. I think what is new is this new role of software in which there's an additional role for these companies that are being destroyed by the public markets is they can be the secure environment on which these agents are held, in which the agents talk together, in which the agents connect to the data and make this a reality in the enterprise and the word matters. And I think that's another avenue of growth that is being open for some of these big software names. So I'm very optimistic around what software can become. I think that the notion of SaaS as in SaaS 2.0 is over because that is not how software is going to be consumed. And again, the profit and loss statement is going to change fundamentally, but there's a new reality that is starting to emerge that will be defined over the next month, which for the companies that actually succeed will be way bigger with a way bigger time, with perhaps better unit economies. And that is starting to be built out, and it's very exciting to see what is happening.
Gemma Allen
>> Well, Raphaelle, I think we need to leave it there for this week. I know you have an exciting couple of weeks ahead too, so look forward to having you back here next week, and I'm sure there'll be lots to discuss in the world of agentic AI and all things defensibility, offensibility, and where exactly we're headed. I'm Gemma Allen joined by Raphaelle d'Ornano. This is AGNT. Thanks for watching.
