Ben Lorica and Evangelos Simoudis on Open Models, AI Coding tools, and the Future of Consulting.
Subscribe: Apple • Spotify • Overcast • Pocket Casts • YouTube • AntennaPod • Podcast Addict • Amazon • RSS.
Ben Lorica sits down with Evangelos Simoudis to unpack why open-weights models are closing the gap with frontier labs and pushing enterprises toward “corporate AI sovereignty.” They then apply an attenuation-funnel lens to AI coding tools — more code, but far less shipped software and usage growth — before turning to “neoconsulting,” the outcomes-based services model reshaping Accenture, Palantir, and the big Indian IT firms.
Interview highlights – key sections from the video version:
Related content:
- A video version of this conversation is available on our YouTube channel.
- Does AI Actually Make Developers More Productive? The Evidence, For and Against
- The Hybrid AI Stack Is Coming for the Pricing Power of OpenAI and Anthropic
- Previous conversations with Evangelos Simoudis can be found here.
- Evangelos Simoudis’ blog
Support our work by subscribing to our newsletter📩
Transcript
Below is a polished and edited transcript.
Ben Lorica. All right, we’re back with my friend Evangelos Simoudis of Synapse Partners. His blog is at corporateinnovation.co. We are recording this on the morning of July 10, 2026. The episode notes, where I will place links to everything we discuss, are on The Data Exchange Media. Hit the subscribe button if you’re on YouTube, and subscribe wherever you get your podcasts. Topic number one is the rise of open models and our belief that open models will ultimately prevail. Evangelos, you start.
Evangelos Simoudis. So, a lot of this is coming out from the research that I’ve been doing on legacy enterprises and the use of AI by them, as well as the work that our firm has been doing with such corporations, and I would say enterprises have embarked on a journey for AI. Most of them, if not all of them, are starting from frontier models. But very quickly, their CFOs are starting to recognize the very high costs that are associated with such models, and also what it means for their own data. So they are, as part of this journey, they’re starting now to ask questions, and in some cases already start utilizing models that they first, models they can modify with their data, and second, models that will allow them to have full corporate sovereignty, for lack of a better term, as well as better control their costs, and obviously certain types of open open models will allow them to do that. Frankly, I also believe that over time, and this is what we are advising these corporations, they should use such models as part of their journey. But ultimately, they will have to build their own models, and those models do not have to be large. In many cases, we’re starting to show them that they don’t need a massive model in order to be able to accomplish certain of their tasks using AI. But that’s a lesson they have to learn over time.
Ben Lorica. All right, I’m going to play devil’s advocate here. First of all, doesn’t this contradict what you keep saying on this podcast, Evangelos, that enterprises don’t have their act together? Why in the world would they use open models? They should start with proprietary endpoints because they’re beginners.
Evangelos Simoudis. Right? And they do. And frankly, if you look at the two requirements this entails, the first is that they need to start getting the right people. And the right people doesn’t mean only AI people. They need to have the culture and the people that will allow them to embark on this journey and bring it to completion. And The second is that they need to start bringing their IT back in-house because if you’re going to start running these models and have what I have called corporate sovereignty, you cannot be doing all this in the cloud, right? The hyperscalers will love you to do that, and obviously that would be a step, but the way I see it evolving over time is we’re going from full cloud, which is what we have today with proprietary frontier or specialist models, to a hybrid deployment where some of the models may start running within the enterprise, and others will continue running in the cloud, whether it is neoclouds or hyperscalers, whatever. And then ultimately, the more capable of these enterprises will need to bring these models in house and run them in their own data centers. And only in this way, they will be able to have what I’ve called in some of my write-ups their AI factories, which they will control, and they will be able to build on that expertise.
Ben Lorica. All right, I’ll offer my first reason, although it is not really a reason so much as a way to set the stage. Open models have become real. If we had this conversation a year ago, we might have said the gap between open-weights models and frontier models from the frontier labs was about 12 months. Now it is perhaps three to six months. You could argue that this is merely a benchmark gap because Chinese open-weights providers are optimizing for benchmarks and using distillation. But benchmarks are ultimately a way to attract attention. People still have to use the models, and based on anecdotal information, including my own usage and that of many people I know, open-weights models have become genuinely useful. You can realistically offload much of the work to them. I do not know exactly how to quantify it, but perhaps 80% of the time you can use open-weights models and reserve frontier models for the remaining 20%. The current reality of open-weights models is one reason for our view.
Evangelos Simoudis. Let me add to what you’re saying. The first point is that we get a lot of questions because everybody realizes that the majority of open models today come from China. In my opinion, this is a very interesting, deliberate, and important strategy that China is employing.
Ben Lorica. The two exceptions are Gemma from Google, which is smaller and perhaps more optimized for edge devices, but should work for many routine enterprise tasks if you tune it, and Nemotron from NVIDIA.
Evangelos Simoudis. Yes, I was going to get to that, but I wanted to make a broader point. We receive many questions about the risks of running models with Chinese provenance. The second point, regarding their capabilities and how they apply to enterprise tasks, is that after two and a half or three years of generative AI, enterprises are finally learning to match a task’s requirements with a model’s capabilities. The more deliberate enterprises, those that truly want to continue on this AI journey, are spending time understanding not only how to automate a specific task or function with AI, but also how to look more holistically at workflows and processes. Once you examine an entire process, ask which steps AI can automate, determine how it can automate them, assess the value, and calculate the cost of deploying an AI-driven process, you become much more aware of the importance of open models, what they can provide, and where they fall short. That is why I see them as an intermediate step toward full corporate sovereignty.
Ben Lorica. Yeah, yeah, yeah. Realistically, what you’re looking at is a hybrid stack, right? So where you probably use open-weights models for a lot of what you do, and then for the remaining 20% for the tasks that really require some deep reasoning or something, you use the models from the Frontier Labs. By the way, for our listeners, when I say Frontier Labs, that’s usually a shortcut for Anthropic, OpenAI, and DeepMind. So, do you have another reason, or should I close up with my last reason?
Evangelos Simoudis. Another reason for using them, you mean?
Ben Lorica. Another reason why open-weights models will prevail, meaning they will take on a lot of the workload.
Evangelos Simoudis. Look, I want to keep emphasizing these two points: cost and corporate sovereignty. To me, these are the paramount. I mean the cost of using the proprietary models and the need for a corporation to keep its its IP, which, for AI purposes, is largely in the data that the corporation has and how it uses that data in the processes that it employs.
Ben Lorica. So for me, my last reason for owning your own AI has to do with tools and infrastructure, and what I’m about to describe is not quite there yet, but will be there. I would say, I don’t know. I don’t want to know. I don’t want to give a time frame, but maybe 12 to 24 months. So by this I mean, so now we have somewhat advanced teams doing pre-training of specialized models. So Datadog is famously built a foundation model for observability data, ranging from parameter counts in the several hundred million to a few billion, I know of healthcare startups that have done the same for very specific workloads in the few hundred million to a couple of billion. So there, there are advanced teams who are able to essentially pre-train specialized models, and I believe the tools that allow more mainstream enterprises to do that will come shortly. And the reason I say that is there’s the infrastructure on the compute side. You’ve heard me talk about the PARK stack, right? So PyTorch, Ray, Kubernetes. But on the data management side, now you have Lance file format for multimodal data. So managing pipelines is going to be much more convenient for enterprises, and then so that’s pre-training. So then on the customizing, so we all know that supervised fine tuning is basically a solved problem now. I mean, I think all you have to do is come up with your labeled data sets of examples of prompts and desired response. So the next level is reinforcement fine tuning, and I have a post coming out. I think next week on Tuesday, in which I describe many many startups that focus on reinforcement learning. Many of them are building tools that are specifically aimed at helping companies do this step of reinforcement fine tuning. And now, so I believe that the tooling will get there for doing both pre training and post training, not of the massive models, but of the more targeted, specialized models, and the reason more companies will do this is for the reasons that evangelists described. But the other reason is really owning your own intelligence, specifically owning that compounding loop, right? So, meaning once you deploy these models in production, you can really observe usage and things areas where the model needs improvement. You can get feedback from your users, and you can get that loop going of improving the model over time, and you own that. So obviously, a lot of what I’ve described right now, the tools are still in the initial stages, but as they say, the writing is on the wall that this will happen. It’s just a matter of time. The democratization of pre-training and post-training, and this ability to extract benefit from this compounding loop.
Evangelos Simoudis. I have written that corporations that are serious about AI, whether they are discovering AI for the first time, rediscovering it, or enhancing what they have already been doing, need to have their own factories. There is no other way. The twist I add is that, to achieve what I keep calling corporate AI sovereignty, they need to engage more than the IT department in building, expanding, and maintaining that factory. This is not only a technology issue. It cannot simply be, ‘Build me something. You are responsible for it because you’re in the IT or AI organization.’ We have been working with a couple of large organizations, and costs are becoming a very big deal. The cost and usage of tokens, and how to account for tokens in budgets, are becoming major issues.
Ben Lorica. And people are becoming more sophisticated about the fact that it is not just about token costs. It is about task completion. For example, Z.ai’s GLM-5.2 is quite inefficient in terms of tokens.
Evangelos Simoudis. But this so that involves or that requires the engagement of the finance organization, right? It requires, as we’ve been saying, bringing the right people into the organization. That includes not only people who understand modeling and understand AI, but people who can handle deployment and educate the end users on this. And that means HR becoming involved in this effort, and then obviously the process owners, right? The the process owners are the ones that will be able to not only help the AI experts to do the right thing and bring the right technology and the right modeling into the into this process, but also they’re the ones who show that there is an ROI from all of these efforts because these are not quick efforts and they’re not cheap efforts. So that, but they are necessary efforts towards that ultimate goal that you and I have been discussing.
Ben Lorica. To quickly close this discussion, the infrastructure for pre-training and post-training is coming. What you described in terms of the financial management of AI usage also requires infrastructure, because attribution is key. If you cannot attribute usage and cost, you cannot manage them. Topic number two is one of the main success stories for generative AI and agents: programming and coding. I became curious after attending an event with representatives from many frontier labs and the people who built several popular AI coding tools. I will not name them because the discussion was off the record, but they were extremely bullish about the space. I like these tools too, so I decided to examine what recent studies actually say. I wrote a comprehensive post that lays out the bullish case and then red-teams it with a counterargument. At a high level, the evidence suggests an attenuation funnel: the effect becomes smaller as you move from activity to outcomes. There is little doubt that developers are writing more code, perhaps 2x or 3x as much. The data in this meta-analysis comes mainly from developer surveys, which can overstate gains because they are self-reported, and from telemetry. But the next question is how much more software is actually shipped. That increase is not 2x or 3x. It may be closer to 30%. If you then follow the funnel to actual software usage through the Apple App Store, Google Play, or other distribution channels, as far as we can tell, little has changed. One could argue that some software users are now agents rather than humans, and I do not know how much that matters. But if one goal of these tools is to produce more mobile apps, they do not yet appear to have moved the needle substantially. We are certainly writing more code. I now write a great deal of throwaway code with these tools, such as one-off scripts to crawl a website or perform data munging. My code output may be 2x or 3x higher, but the increase in shipped software is closer to 30% according to these studies, and end-user usage has not changed much. My analysis includes more data on bugs, security challenges, and other issues, but the attenuation funnel is a useful summary: 2x or 3x more code, roughly 30% more shipped software, and little change in end-user usage.
Evangelos Simoudis. Let me add my perspective, based on our firm’s engagements. First, there was tremendous excitement about what these tools could mean for enterprises and their project backlogs, but that excitement is becoming more measured. Companies are recognizing the technical debt being introduced and what it could mean later. There is also the issue you described, Ben, of how the code is evaluated and which code passes an enterprise’s evaluation criteria. The more sophisticated enterprises we work with are using these tools extensively for prototyping and for understanding what new applications could look like. In my opinion, they are also wise enough to use the tools primarily for new applications rather than trying to modify or enhance existing ones, which would require ingesting a large quantity of enterprise code. We are seeing some efforts to use these tools to convert or refresh the periphery of an older enterprise application from an older language to a newer one, rather than touching its core. In general, the initial excitement is being moderated. I am speaking specifically about enterprises.
Ben Lorica. What do you think of that attenuation funnel? Does it make sense?
Evangelos Simoudis. I think it makes sense. We have not done a quantitative analysis of the work we have been involved in, but I would say it makes sense.
Ben Lorica. And by the way, it actually so here’s one intuition for you listeners, right? So yes, so 2x to 3x more code, but only 30% more shipped software, and in many ways, that makes sense and intuitively because basically people still have to review this code.
Evangelos Simoudis. Yeah, right.
Ben Lorica. The review processes may also be somewhat automated, but people still have to be careful.
Evangelos Simoudis. Let me add two more points very quickly, Ben. First, for perhaps the last eight months, we have seen startups produce version 0.5 of their product using these tools with far fewer people. That is important from a capital-efficiency perspective for startups. That was a revelation. The second point, which is less surprising, is that companies we would characterize as digital natives make much more efficient use of these tools than legacy enterprises do, by and large. This is a broad statement, but digital natives tend to have more modern code and practices, which makes these tools easier to use. Still, with respect to your funnel, I do not believe we will see significantly different results between how a digital company gets value from these tools and how a sophisticated legacy corporation does.
Ben Lorica. So a couple of things on that point, right? So in the in that long study that I published, right? So there was a DevOps research survey, right? So, and they did find that AI functions as an amplifier of your existing team. So, meaning, if you have mature internal platforms, clean workflows, really disciplined engineering practices, you extract outsized benefits. The other thing is that code-based maturity also decides where AI helps or hurts, right? So I guess let me let me kind of also tie this back to well, then what should I do, right? So I think you know because the attenuation funnel starts with that 2x or 3x right more code. The temptation obviously is to reduce headcount, right? So, and I think that might be that make might make sense short term, but over the long term, maybe that could hurt you because you do still need. As evangelists and I have talked about repeatedly here, you still need the more code you produce and write, the more people you need to actually maintain and watch over all of these systems that you are either using or your customers are using, right?
Evangelos Simoudis. Ben, you reminded me of something which I think at least some listeners who work in established companies, corporations may may already be feeling, as you know, there is a the term K-shaped has become very popular in many contexts, but on because of the cost that I talked about a few minutes ago, we’re starting to see a type of a K-shaped budget allocation. So, because of the costs that are associated with some of these AI projects, particularly code generation, we’re starting to see corporations taking money from certain IT projects, including certain new applications, migration of applications, and moving them towards AI projects. Some of it is to plug holes. Some of it is because they are seeing important results, and that may relate both to the modeling discussion. That we had earlier today, but as well as the program and the use of AI tools for, particularly for code generation. So be aware of that. Last year. It was all new money going into AI projects. Now some of these projects are starting to expand and consume a lot more capital. We are starting to see this K-shaped budget allocation.
Ben Lorica. By the way, this ties to our previous discussion around tokenomics, which basically you should actually focus on task completion, not token consumption. The total cost of task task completion. So here you should measure shipped value, not activity. And activity can mean many things, right? So code number of lines of code written, tickets resolved, and things like this. But basically, shipped value is the main metric. By the way, speaking of which, evangelists, one of the things that people talk about in this area is the J curve, which is basically as for people who don’t know, the J-curve describes a phenomenon where you know in the short term there’s a dip in productivity, and before it rises sharply, right? So in this case, people are saying we are still in the downward part of the J-curve, in the following sense, right? So right now people are still understanding how to use these tools. They’re overwhelmed with the new code they have to review and things like that. So then people are speculating that there will then be a J curve where okay, once people stabilize and really understand how this works. Boom! Productivity will explode. But this is obviously hypothetical. We don’t know if that’s going to happen, right?
Evangelos Simoudis. Well, we are. I mean, I have written about this for those the listeners who go to my blog. I have my most recent posts. I mean, I’ve been exploring this concept that I call the AI stakeholder squeeze, and most of the squeeze, which is coming from employees and senior management, starting with the CEO, is exhibited during the dip of the J curve. And what I’m starting to study now, again looking at our customers and more and broader data from several analysis firms is how, depending on the strength of that squeeze and the forces that created them, whether the dip becomes shorter or deeper or shallower. I mean, so it has certain characteristics which vary from company to company. So stay tuned for ongoing analysis. But it’s it’s it’s becoming a lot more interesting than I thought it would be when I first started doing the research.
Ben Lorica. And another thing that I that I was able to uncover is that obviously one of the explanations for why maybe we’re dipping down right now and maybe there’s hope for going back up is that this is an entirely new developer experience, right? So people are people are adjusting to these new tools, and one of the things that people frankly miss, including myself, is the state of flow. It’s hard to have a it’s hard to have a state of flow when you’re chatting with a chatbot or a coding agent. Before, obviously, by state of flow, I mean you’re coding and you’re just you’re just on fire and you’re just able to work through problems, right? So, and one of the reasons is obviously is because you are able to focus and lock in and things like that, but now I think with these tools, you tend to get more distracted. I think your attention is not as
Evangelos Simoudis. Actually, I thought you were going to mention something else: not only do you have to deal with a new regime of tools, but within that regime there are new tools every day, with different workflows. That is also creating stress.
Ben Lorica. But most people, frankly, they lock in, right? I’m going to use Claude Code, or I’m going to use Cursor. In my case, I’m going to use OpenCode, right? So. I think that I think there’s still some shopping around, obviously, but for the most part, I think that would be crazy to try to keep trying something new every week. And obviously, there’s also the notion of governance around these tools, right? So, how do you how do you govern the use of these tools? Because I think you’ve you’ve talked about this in the past. The whole what’s the term where bring your own AI?
Evangelos Simoudis. Yeah.
Ben Lorica. It could be that developers love these tools when they are at home, but are banned from using them at work. They then find a workaround and use them anyway.
Evangelos Simoudis. Essentially, you have dark IT, or shadow IT, with regard to AI. I call it shadow AI.
Ben Lorica. To talk to close the loop on this, I guess we I started out with this attenuation funnel, which was a bit not pessimistic, but you know realistic. But I think overall, I think the my sense is that these are going to be tools that people will embrace and learn how to use, and we will see if so if you follow that attenuation funnel all the way to the App Store apps, will we see increased usage of apps because of these tools? We haven’t seen it, obviously, at this point. Maybe, maybe these, maybe perhaps these tools will allow people to author even more compelling apps that people will download and engage even more. I don’t know. So, all right. So, last topic: what evangelist refers to as neoconsulting. So, go.
Evangelos Simoudis. So again, going back to this point of how do corporations utilize generative AI? AI in general, we’re starting to, and I do not know really whether it was the enterprise that motivated it or the providers of these frontier models and particularly frontier models and associated tools are motivating it, but we’re seeing the emergence of a new type of consulting firm whose primary employee is the so-called forward-deployed engineer, whose goal is to go into a corporation and help the corporation utilize in a better way and faster, and see results faster with the tools that Frontier Lab or an associated partners are providing, so we’ve seen most recently Microsoft allocating two and a half. Obviously, Palantir started that. So
Ben Lorica. We’re not talking about that. Evangelos, in the world of ‘neo,’ there are neoclouds, which are genuinely neoclouds, and neolabs, which are genuinely new labs. With neoconsulting, are there actual new companies, or are these simply new practices or groups within old firms?
Evangelos Simoudis. Microsoft started a new company and put $2.5 billion into it. OpenAI and Anthropic started new efforts, and they took money from private equity firms such as TPG and Blackstone.
Ben Lorica. But there is no new Accenture, Wipro, or Infosys.
Evangelos Simoudis. There are a number of startups, beginning with Distill, which was founded by Palantir alumni, and a few others in our ever-expanding database of AI startups.
Ben Lorica. By the way, startup startups and the word consulting usually don’t mix together.
Evangelos Simoudis. Well, I mean, look, these are you can think of as services companies that combine software expertise or AI software expertise with people who not only may have an MBA, which was the traditional path for. Systems integrators to hire people, but also have a lot deeper expertise of AI and of specific tasks, and they are embedded in an enterprise’s operations in order to bring in the new tooling. What is interesting to me is not only the emergence of this effort. So is there is
Ben Lorica. Is there an easy way to think of neoconsulting as a clearly different category from Accenture or McKinsey? What is the one-sentence description of what makes a consulting company a neoconsulting company?
Evangelos Simoudis. I would I would actually say first and foremost, as far as I’m concerned, is the business model because a lot of these companies’ efforts use outcomes-based compensation, as opposed to time and materials, and that has very big implications to certain services providers, particularly the outsourced providers from countries like India or Indonesia, or you know the lower cost providers, but also it has implications to the more established firms like Accenture, McKinsey, because now before you can recognize that revenue, the client has to see and recognize results and as we know in AI a lot of this is prediction-based right so you’re predicting that something is going to happen it’s not like integrating SAP into your operations where you can see results immediately so how these new consultants structure their contracts so that they can provide value to the customer, but also start seeing revenue from that customer. Recognizable revenue is a challenge that CFOs are working on.
Ben Lorica. So, by definition, do neoconsultants in your framing focus almost entirely on AI?
Evangelos Simoudis. Absolutely. Today, you see them from Anthropic and OpenAI to Microsoft, Palantir, and, obviously, startups. They are all focused on AI.
Ben Lorica. And then you brought up outcomes-based pricing. I bring up two seemingly contradictory terms from the world of consulting. One is being on the bench, right? So meaning you’re a consultant and you’re not on the project. That’s not that’s never good. And then secondly, it’s highly relational in the following sense, right? So if I’m, I don’t know, I’m making this up. I’m J.P. Morgan. I’m working with McKinsey. I want senior partner X, you know, not some unknown person, that so there is branding even among the people inside the company, right? So are those two things gone from this?
Evangelos Simoudis. No, no, no. Actually, they’re not, and I think if you look at the efforts by both OpenAI and Anthropic, they’re bringing, they’re partnering with some of the established both management consulting firms and IT consulting firms, and in fact, so to your second point on the partner, the engagement team is still led by a partner, and the seniority of that partner depends on who is setting up the firm, I guess. But one of the things that we’ve heard from corporations, and we’re facing it as as part of our advisory practice, is that they want the senior partner, what you call the senior partner, to be far more engaged in the delivery of the outcome, as opposed to what was happening before in the past in the in the traditional let’s call it consulting model, the senior partner is responsible for selling the engagement and then providing some supervision and obviously ultimately delivering the outcome, results, or report or whatever. The
Ben Lorica. The main work they are doing is helping these companies become AI companies. In other words, they are not just developing an AI strategy. They are going in and helping them implement AI.
Evangelos Simoudis. I would put it this way, Ben: they are helping clients make the best use of the AI tools they are promoting. An OpenAI neoconsultant is promoting OpenAI tools, and Anthropic and Microsoft are doing the same. First and foremost, they are making sure that the enterprise, as part of this AI journey, makes the best use of the vendor’s tooling.
Ben Lorica. What you have described to me, in the case of OpenAI, Anthropic, and Microsoft, does not sound like neoconsulting firms.
Evangelos Simoudis. And Palantir.
Ben Lorica. So I don’t think of them as neoconsulting firms. They’re basically just services companies for that particular company. So neoconsulting to me would be someone like Accenture or Wipro or Infosys standing up. Up a practice that will mimic what Anthropic and OpenAI are doing, but they’re much more neutral in the fall. In the sense that we’ll help you do what you need to do, but we don’t, we’re not tied to one vendor.
Evangelos Simoudis. Okay, but let’s be clear. Large-scale consulting firms, from Accenture to PwC, Capgemini, and KPMG, have partnerships with large SaaS vendors even outside AI.
Ben Lorica. I’m just using them as an illustration. ‘Neo’ means new. To me, this would be a new consulting company focused on AI that goes into a company and helps with AI without being tied to one vendor.
Evangelos Simoudis. Today, some of them are tied to a vendor.
Ben Lorica. Some of them are the vendors themselves. To me, they are not consulting firms. They are solutions engineers for that firm.
Evangelos Simoudis. Yes, even though you have to appreciate the following is that as they look at a process to automate a process, as we have said throughout this podcast, AI is not 100% of the process, right? So they have to know about other tools to complete the particular task and provide the outcomes that will allow them to get paid. And again, the so to me the new so
Ben Lorica. So they’re they’re still acting as consultants, but when it comes to recommending the AI tools, they only recommend their own.
Evangelos Simoudis. I would I would imagine that I mean, and I think that what you will start to see from let’s say a startup like Distill or Hank 10 Systems, what you’ll start to see is they will differentiate themselves from efforts from Anthropic or from OpenAI or from Microsoft by saying we are much more independent in terms of what AI tools we can introduce to your firm,
Ben Lorica. And they might even use open-weights models, to tie it back to our first topic.
Evangelos Simoudis. Potentially. I mean, you have firms like Umi, for example, that can provide that type of capability, right? If they if they choose to. Anyway, so my point is that the neo part comes from their AI expertise or the expertise of their people, right, which is centered around AI and processes, but second and more importantly, the business model. This outcomes-based business model is completely new, and how it is going to be received is going to be very interesting. I mean, when we talk to some of the companies we collaborate with in India, which provide either engineering services or IT services, they are feeling a lot of pressure to reduce their costs because now they have clients in Europe and in the U.S. Saying that if I use AI, I can do certain things much cheaper. This remains to be seen. And by the way,
Ben Lorica. This new business model, being outcome-based, actually plays directly into the hands of Anthropic OpenAI and Microsoft, right? Because basically, they don’t really care about billable hours. They just want to make sure that at the end of the day, you’re on their AI stack.
Evangelos Simoudis. My personal opinion is that it plays much better in the hands of hyperscalers like Microsoft. AWS, for example, established a similar unit, right? Again, let’s not leave them out because of their balance sheet. In the case of Anthropic, they have to. I mean, they have to get money in order to support these efforts, right?
Ben Lorica. My benchmark is not necessarily the hyperscalers, but the existing consulting companies.
Evangelos Simoudis. Yeah.
Ben Lorica. They focus on billable hours.
Evangelos Simoudis. So if if the business model takes hold, it’s going to have a big impact in the services industry, and it is not clear how far and how much the target customers will embrace it. But I know that more and more are asking for it. Actually, remind me. I’ll give you a quick anecdote. When I was running IBM’s AI division, this is back in the ’90s. We were offering both services and product in my organization. And I remember giving an interview in London to, and as part of the event, I said something like, “We will be willing. We feel so certain about our our capability that we will be willing to do a like a revenue-sharing type of an arrangement with a with a client. This is pre-Watson, by the way. This is ’90s again. Just just again to put it into perspective. As soon as the like the following day, the newspapers print the interview. Actually, quite big newspapers in London. I get a call from IBM CEO. That was the last time I talked about outcomes-based and revenue-sharing type of payment. So what happens material?
Ben Lorica. What happens to the existing consulting companies?
Evangelos Simoudis. I think again, that’s what I say. If if this thing, so the existing consulting companies have two challenges. The first challenge is retraining as many people as possible to be able to understand the needs and to provide the AI advice and capability that enterprises need. That may lead to a lot of reshuffling, maybe even some reductions, but most reshuffling. But if the business model takes hold, then I think we’re going to have a significant impact, or could have a significant impact on their financials, right? On on how they recognize revenue. I mean, a company like Accenture, 400,000 employees plus. Yeah,
Ben Lorica. The reason I ask is that, in the past, billable hours meant wanting more hours to bill. Now you do not need as many hours. You want to finish the job as quickly as possible for the customer. The ideal scenario is a lean team that can do it.
Evangelos Simoudis. So Here is how to think about it. So first and foremost, customers are asking us; they’re asking competitors and all that, and we’re, as I said, a microcosm of what’s happening in the broader consulting ecosystem, but customers are asking for lower hourly rates because they know that all of us are using these tools in the course of performing a task. So they’re saying if you’re using Gemini Anthropic or whatever, you cannot be charging me X. You have to charge me 30% of X, 50% of X. You know some some smaller amount. So that impacts consulting services, regardless of whether you’re doing AI consulting or meat-and-potatoes consulting, right? You know, you’re integrating, as I said, SAP or Workday or something like that. The second part is what do you do with AI projects, and now you have to both deal with a different type of consultant, right, with a different kind of expertise in order to be able to compete.
Ben Lorica. Also, because you are not motivated by hours, what you really want is a lean and productive team.
Evangelos Simoudis. Right? So now those teams need to be managed differently, need to consist of different kinds of people, and then at the end they have they need to have a different business model. So you may end up seeing the legacy firm. Like an Accenture, like a KPMG, having a two-track business model, a different model for time that is more time and materials for the more quote-unquote traditional tasks, but still at reduced rates because of the use of AI in completing those tasks, and then a separate new business model for the tasks that involve the delivery of AI,
Ben Lorica. And for our listeners in India, what happens to the large Indian IT consulting firms?
Evangelos Simoudis. Yes.
Ben Lorica. What happens to them?
Evangelos Simoudis. I think they are already seeing a lot of pressure, For the publicly traded firms, I think you will start seeing missing revenue targets, and also how they hire, who they hire, how many they hire. I think it’s changing. I mean, we’re seeing it because some of them are asking for our advice on how to proceed, but we are we’re seeing we’re starting to see the impact and hear about the impact.
Ben Lorica. And with that, thank you, Evangelos.