This year, we have seen the two biggest cybersecurity acquisitions ever made: Google Cloud’s acquisition of Wiz, for $32bn USD, and Palo Alto Networks’ $25bn deal to acquire CyberArk. And it’s only August!
Cybersecurity investment continues to grow at a staggering pace. According to Return On Security, cybersecurity funding is up 202% in Q2/3 2025 compared to the same period last year, with M&A deals up 92%.
What’s driving these trends? What are the hottest areas to invest in cybersecurity today? Who are the hottest companies to watch, and who could be next in line for a $20bn USD+ acquisition?
At Black Hat USA 2025, we spoke with Deepak Jeevankumar, Managing Director at Dell Technologies Capital, to cover these questions, and much more. You can read our full interview below. We’ll also release this as an episode of the Expert Insights Podcast after the show – subscribe here so you don’t miss it!
It’d be great to start off just with a bit of an introduction to yourself, Deepak—how you got into the cybersecurity space, your career journey, and your role as MD at Dell Technologies Capital.
I work for the investment arm for Dell Technologies called Dell Technologies Capital. I’m one of the partners there; I’ve been there since 2017, and investing in cyber for about a decade. Before that, I was in another VC firm called General Catalyst Partners.
Cyber is one of those things where there’s a fertile ground for innovation and investment because the attackers are getting more sophisticated and cleverer every day. The attack surface is increasing; with every new technology trend, a new attack surface gets formed. The current vendors are always lagging because attackers have an asymmetrical advantage. So, it’s a very fertile ground for innovation.
I’ve been investing in cyber for the last decade in multiple areas like endpoint security, network security, identity security, governance risk and compliance, and cloud security. As a firm, we have invested in some of the giants in cyber like Zscaler, Netskope Cylance, Redlock, and Twistlock—which are the core of Palo Alto’s Prisma Cloud offering—and Humio, which is the basis of CrowdStrike’s next generation SIEM offering. I’m looking forward to figuring out the next ones to invest in. We have a bunch of upcoming players like Halcyon on ransomware; Twine, and Entro in Identity Security; and Endor Labs, Descope, and Cequence in application security. And we are super excited to figure out what Black Hat holds for us.
Cybersecurity is very cyclical—you have zero trust and there’s a big spike in that, and then the next thing comes along. What’s the bird’s eye view of the trends you’ve seen over the last 10 years, from that VC perspective?
I would say there are two kinds of trends: evergreen trends and cyclical trends. I think zero trust has now become evergreen—it’s all permeating; it’s in everything.
The flavor of the year has been AI security, security for AI, and securing AI. There’s a lot we can talk about there.
Identity has been the forgotten part of cyber for a long time. You have big cyber companies that have been focused on everything but identity; Palo Alto was focused on networking and cloud, Zscaler on networking, CrowdStrike on endpoint, but you didn’t have any of the big ones focused on identity.
CyberArk was a bigger one that was focused on identity, and now it’s going to go to Palo Alto very soon. I think there’s going be huge resurgence in identity because it’s kind of the forgotten perimeter of cybersecurity. And with AI, identity becomes more important.
Interesting. Why’d you see that becoming the case?
So, let’s talk about what identity is and what the different kinds of identities are. Identity is connected to an entity—it could be a human entity or a machine entity. Every AI entity is a machine entity.
The problem with human entities versus machine entities or AI entities is that, firstly, the quantity of AI entities and machine entities is like 100x or 1,000x more than human entities. The second thing is they have a much shorter half-life. In an organization, an employee identity is when an employee joins, then the employee leaves with the same identity. But if you think about machine identities, AI or agent identities, or LLM identities in the future, these have very short half-lives; they could exist today, they could not exist tomorrow, or it could be a different version with no continuity between these different identities. So, it’s very ephemeral.
The entire identity ecosystem was built around the concept of human identities, which are not ephemeral, and which last a much longer period of time.
Plus, machine identities access the network through different ways that we’re not used to: a browser, APIs. And the system is just not used to humans and machines working together.
So, does there need to be an evolution?
I would say a revolution, not just an evolution. It has to be like a completely new way of looking at it.
The other thing is that the entire IT infrastructure stack has been built for programs or applications making deterministic decisions based on sets of rules. But artificial intelligence changes the game. The results from generative AI are usually not deterministic; they’re probabilistic and stochastic.
And when you have an entire infrastructure stack built on the fundamental assumption of determinism, but your new technologies are not deterministic—they’re probabilistic and stochastic—lot of fundamental things change.
Because of all of this, the challenges of identity ecosystem in the AI world are going to be far more complex.
And of course, there are all the things that are well written about, like deepfakes, phishing, and so on. There are also other things that have not yet been written about, like AI insider threat.
Prompt injections and that sort of thing?
Much more than that. It could be like an autonomous agent doing things that a human insider threat actor would not even imagine. And they could cover the tracks much more than human threat actors could. So, I think the next decade will see a rise of AI insider threat in addition to all the other problems AI is going to create for us—and that’s very much an identity problem.
Are there any vendors you’re seeing that are maybe looking at this as an issue? Is this an area that you see hotting up?
Definitely. Like all the VC firms, we’ve been tracking all these problems. If we look at problems created by new technology trends, AI is a huge bucket, cloud was a huge bucket and has been for the past decade, and there are also problems that have always existed but have not been solved well.
One problem that has always existed but hasn’t been solved well is ransomware. We invested in a company called Halcyon, which is probably the first cyber vendor to say that ransomware is the number one priority of problems they want to solve, and it’s the only thing they’re going to solve. If you think about all the other big, amazing cyber companies, nobody can claim that ransomware is the number one problem they want to solve—it’s like one of the ten things they do. If ransomware is such a huge problem, why is it that no major cyber company can claim that to be the only problem they’re going to solve? There’s a huge gap in the market, and we like the Halcyon technology, from visibility protection to recovery and resilience. So, we invested in them like a couple of years ago and they’re doing really well. And that’s like: existing problem; no good vendors.
In the identity space, we invested in a machine and AI identity company called Entro Security. They catalog and track all the different identities, help you figure out what is stale, what is not stale, and help you rotate all identities—whether they’re passwords, API keys, agent access, LLM access, and all of that.
Another of our identity security companies, Twine, has been using AI to improve and automate identity workflows. They’re actually one of the finalists at the Black Hat startup competition and they were a finalist at the Innovation Sandbox as well. I’m on the board of Twine.
So, there are some problems we think that need to be solved; that need a fresh view.
AI is going to both help and hurt. It’ll hurt the attack surface more, but it’s going to help the vendors to be more effective. So, I think those technologies and founders who can understand the power of stochastic and probabilistic systems and use it really well will succeed.
Fascinating. There’s always a debate about whether it’s better to have a point solution, as you mentioned, focusing on one issue, or a broader platform. A couple of years ago people were talking about a move to consolidation and how companies were minimizing the number of tools they had. Do you think the number of AI tools that are launching will change that? Are companies looking at these AI tools as quick implementations that don’t have massively long deployment cycles? Is there a space for the loads of AI players coming onto the market?
Yes. There are loads of AI players who are creating products, from foundation models to vertical apps to horizontal apps. There are also lots of AI security companies who are trying to solve problems created by AI.
I don’t see any short-term consolidation with AI security. I think it’s going to take a while because AI itself has not consolidated well. If you think about the cloud world, it took five or six years, maybe even longer, for the ecosystem to collapse around three cloud vendors, and then Oracle became the fourth cloud vendor.
So, in the AI world, yes: OpenAI, Anthropic, and Gemini are rising fast. Let’s see what Meta does with all the changes, but you may have four major AI players around which ecosystems are going to emerge. Some will be closed, some will be open, some will be partially open and closed.
In terms of AI security, it goes from model red teaming to API control, MCP servers, A to A, and so on. For example, one of the problems created in the context of AI is you’ve got APIs from the old world and MCP servers in the new world. How do they communicate with each other? It’s an end-to-end problem.
One of our companies called Sequence introduced a product called a “secure AI gateway”, which connects all MCP servers on all APIs to one point and then it’s easy for application developers, cyber vendors, and cyber professionals to track what’s happening between the APIs and MCPs.
So, there’s going to be a bridge that needs to happen between the old world and the new world, and I’m super excited about all of this. There’s going to be a period of a lot of noise and there will be a consolidation, just like what we saw in the world of cloud security. It took probably a decade for the consolidation to happen in the cloud space. I would say this will also take a decade.
I also think that the splintering of the ecosystem is good, because you have multiple sources of innovation and then the strongest wins. It’s a Darwinian evolution. But I think it makes it hard for CISOs, because they have to decide who they want to take a bet on. I totally empathize with that problem, and what I would tell CISOs is, you cannot afford to take a bet on just one. You need to have multiple bets, because the problems are still not well-known, and figure out how to work with multiple vendors in an effective manner. Use all the AI tools that are available to make things efficient for your workflows as well, really understand who you want to introduce into your workflows and who you want to use for context and actions. And based on that, you could create a new ecosystem for yourself around AI security. But you can’t stay idle. You have to take multiple bets.
For founders and people launching the companies, the competition must be intense. How do you differentiate between a company that’s got a vision that you think can be helpful for CISOs, and the ones that might have promise there, but it’s never going to take off?
It’s a fast-evolving ecosystem, so I would take a bet on founders who understand the ecosystem is evolving fast and who can really move with that.
Because the ecosystem is moving fast, there’s also another question: how do I take these security products and make them usable? I think the ones that are going to succeed are the ones that can actually instrument the products into the customer’s ecosystem, because the traditional system integrators don’t understand AI yet. They will, but not today.
It’s not about selling a product; it’s about solving a customer problem. Because they’re in such early stages, I would encourage founders not to just sell products, but to sell a service around your product, or sell a service and then sell your product. It’s totally against the conventional wisdom of venture capital and the startup journey, but because the problems are so unknown and they’re all fast changing, unless you are a very trusted consultant for a customer, you won’t know what they’re facing.
You have to be on the bleeding edge. And the only way you do that is by staying with them as a consultant to feel the problems they’re facing.
Think about an extreme example, Palantir. They have huge pricing power because they say, “Hey, we are going to control the outcomes and we’re going to deliver something end-to-end.” I think the same thing can be done in cybersecurity, especially in the context of AI security.
So, I would encourage founders and CISOs to think about the ecosystem as not just a mix of products, but also who can deliver outcomes for you.
What’s AI adoption looking like within organizations? Is there a lot of skepticism, or is there excitement in terms of wanting to adopt these tools and see if they can improve those outcomes?
There’s both excitement and skepticism. The excitement comes because of the potential of what is possible: code generation, natural language generation, et cetera. The skepticism comes because of one of the things I mentioned before: the probabilistic nature of AI output. In other words, can you assure me I’m going to get the same output every time?
We in the IT world have been trained to expect deterministic outputs. We are not being trained to expect probabilistic outputs. But we don’t have a choice—we have to evolve to expect these probabilistic outputs. It’s almost like when the world moved from analog to digital, 50 or 60 years ago. Now, the IT world is now moving from determinism to a probabilistic nature.
The other skepticism also comes from the pace of change. If you take the very narrow segment of AI code generation, every six months there is a hot new product in the market.
What are some of the most exciting companies that you’re looking at in this space?
Let’s classify the companies as big platform players and startups. One great thing about cybersecurity now is the big platform players are doing really well and innovating fast at scale, be it Zscaler, Palo Alto, CrowdStrike, the Microsoft Security Division, Google’s Chronicle, and so on.
I think that’s a good and a bad thing for startups. The good thing is that you have a stable ecosystem around which you can build. The bad thing is that their sales and go-to-market power is very high from a startup perspective.
Startups should expect these big platform players to continue to be really strong in the near future. From a startup perspective, there have been a bunch of really big companies getting created.
For example, one of the fastest growing companies is Endor Labs. Their founder is a second time founder; he had created Redlock, which became Palo Alto Prisma Cloud. Endor Labs is securing code—whether it’s AI- or human-generated code, both of which are growing really fast.
Then if you look at the whole ecosystem, there are a bunch of companies that are using AI in the Security Operations Center. Now, what is unclear to me is how much better they are compared to the previous generations.
We’ve seen some big funding rounds just in the last couple of weeks, haven’t we?
Exactly. We’re tracking that space around AI-based pentesting. Pentesting is one of the big consulting markets in cyber and there’s an ability to make it a product-based market. So, that’s a market ripe for disruption. We’re also tracking a bunch of AI security plays. We are an investor in a company called Sync; they’re like the next generation of protect AI. They provide end-to-end visibility, remediation, recovery—all of that in one platform.
And I think there’s a lot more to do in AI security, including around deepfakes. We haven’t invested anything in deepfake detection, but there are a few amazing startups there, so we’re tracking that space.
We’re also tracking what else can be done with Palo Alto acquiring CyberArk. CyberArk is an amazing company; they control the market when it comes to privileged access management. So, what does this mean? What opportunity does it open up for startups? I think in the intersection of privileged access management, cloud, and AI, a lot of innovation is starting to happen.
Palo Alto and CyberArk is a major acquisition—the second biggest ever in cybersecurity after Wiz. Do you have any predictions on who you think could be next in line, or any sectors where you think an acquisition could be brewing?
I think we’ll see more in identity. SailPoint is a publicly traded company now, and they are the market leader in identity governance and administration. At some point, they may be up for grabs for the right price. I have no insider information or anything; this is just me guessing wildly!
There could also be more consolidation in cloud security. It’s not done yet, but Google is buying WIZ, and there are other players that are doing really well in cloud security.
I think there are going to be a lot of acquisitions in AI security. At some point, a platform like WIZ will appear in AI security. I don’t know what form it’s going to take, but at some point it will appear.
Cisco has also seen a huge resurgence in cybersecurity. They’ve made a lot of acquisitions recently, and many of them are doing really well. I would not discount Cisco out of cybersecurity.
So, it’s going to be a lot of fun to see how the endpoint players, network players, and cloud players move into the AI world.
I believe that this is also an opportunity for some of the cyber players to get into adjacent areas of IT, like observability, logging, and so on. Just like how Datadog has been getting into cyber.
I’m looking for business model innovations in cybersecurity. We briefly touched upon the fact that this is not just about product; you need to sell both a combination of product and services to deliver an outcome for a customer. I’m super excited for founders who are thinking about and delivering that, because that’s one way they can rise above the noise.
Thank you to Deepak Jeevankumar for taking part in this interview.
For more information on AI security, check out some of the following articles:
