Cohesity at AWS re:Invent 2023 – The Six Five on the Road

By Patrick Moorhead - November 30, 2023

On this episode of The Six Five – On The Road, hosts Daniel Newman and Patrick Moorhead welcome Sanjay Poonen, President and CEO at Cohesity for a conversation on Cohesity’s strategic partnership with Amazon Web Services (AWS) by integrating Cohesity Turing with Amazon Bedrock.

Their discussion covers:

  • A look at Cohesity’s Turing integration with Amazon Bedrock
  • How this generative AI solution will allow customers to use the latest AI capabilities to better manage, interact with, and gain insights into their data
  • Cohesity’s new upcoming capabilities for protecting AWS and VMware workloads

Learn more about Cohesity’s partnership with AWS, on the company’s website.

Watch the video here:

Or Listen to the full audio here:


Disclaimer: The Six Five webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.


Patrick Moorhead: The Six Five is on the road at AWS re:Invent, and we are chatting about Cohesity. Dan, it’s been quite an event here. We’ve talked to end customers, we’ve talked to a lot of vendor. A lot of partners we’ve done, a lot of videos we’ve done. And what’s very crystal clear here is that AWS is reliant on a partner network to accomplish what they need to do. And it’s really a maturation of the cloud. Gosh, five years ago, it seemed like AWS was just going to do everything, and then it started opening itself up to partners. It realized that not only for the good of the end customer, but also for the good of business, working with others was a good thing. And we’ve seen a lot of that here.

Daniel Newman: Yeah, it’s definitely great to be here, Pat. It’s been wall to wall from the moment we got here. The energy is always really high. This has become, I think, one of the executives we talked to. So, this isn’t just the AWS conference anymore, this is the B2B tech conference.

Patrick Moorhead: Yes.

Daniel Newman: Everybody that’s anybody in this space is here. This is where it’s happening. And it does feel a bit that way. And to your point, the partner ecosystem, massive growth, the story around hybrid, it’s got a big silicon flare. And of course, yesterday was Cyber Monday. So, we got to talk a little bit about cyber resilience and security, and I think we could even do that now maybe.

Patrick Moorhead: And I think we’ve got to talk about AI too, but it’s my pleasure to introduce Sanjay Poonen, president, CEO of Cohesity. Sanjay, how are you my friend?

Sanjay Poonen: Thank you, Patrick and Dan. Congratulations on… I got the backstory before we got on air about Six Five. So, you guys have built an incredible brand. And excited that you’re here. I’ve been at AWS conferences since, believe it or not, guys, 2012. The first conference that Andy Jassy did. I was at SAP at the time, and as president there, I was one of his first guest speakers on stage. And there was only a thousand people in the audience. Only a thousand people there.

Daniel Newman: Did you hear the number that Adam Selipsky threw out?

Sanjay Poonen: It’s like 60,000 now, right-

Daniel Newman: That live.

Sanjay Poonen: Yeah, that’s pretty amazing.

Daniel Newman: 300,000 people watching this came out.

Sanjay Poonen: Watching it live, yeah. It’s a phenomenon. I think it’s whatever… It’s the B2B show, it’s the Davos, it’s the Comdex, whatever your favorite analog from the past is, it is now the show to be at. And they’ve certainly created a cult life phenomenon here.

Patrick Moorhead: Well, listen, you’ve been paying attention to Six Five and we’ve been paying attention to your moves. And it’s been fun tracking you at the multiple companies that you’ve gone to. And I felt like you really brought this… You brought a bunch of partners in to crease out the fold. You brought a bunch of advisors, customer advisors to help you, and brought some big names as well. From the IBMs and the AWSs, it’s definitely a new look for the company. And I know you had a great technical grounding that you came in, the company to run, but you really look like you’re an overdrive now.

Sanjay Poonen: Yeah. Listen, we try to stay humble and hungry, guys. And one thing that’s true about the software business, from my experience, watching at least the last 18-20 years at VMware and SAP is success does not breed success the next day, you’ve got to always, like AWS, Jeff Bezos and Andy Jassy, call it day one mentality. So, I think 15-16 months ago when I joined the company, I saw a really good… This company had the best tech in its category evidenced by the success that Mohit had had as the founder of Nutanix. And when he founded Cohesity, we invented hyper-converged for secondary storage. But I saw a tremendous opportunity at junction of not just the cloud world, which was where AWS fits in the story, and they’d made an investment in the company three years ago, but also security and AI. So, we first started off with some incredible tech innovations that advanced our security and now AI portfolio we’ve talked to, and you guys have written a lot about both of those that have been profound. But we also needed to, beyond the tech, get a ecosystem of partners and a set of personalities involved in the company. So, one of the first calls I made was I asked myself, who is probably the biggest security name in the world that fights cyber threats every day for governments? And that’s Kevin Mandu. So, Kevin is probably… I affectionately call him the Sherlock Holmes of the cybersecurity industry. If some big brand in this country gets hit, he’s been ex-Air Force, been helping… He’s not allowed to go to Russia and China because he’s been fighting cyber, and he knows where every one of these gangs are working. I made a call to him and he graciously accepted, looked at our tech, felt we were the best tech.

So, that was the first thing we did as I joined the company. And then we began advancing our capabilities with Amazon and a few other clouds, and then prominently, a couple of very key security players who are the big names in the space, Palo Alto, CrowdStrike, Microsoft security and so on. And then I would not have even seen this new advances that could have been possible with generative AI, because 12-15 months ago, I wasn’t thinking GPT-4. But 9-10 months, and we’ve chronicled how that happened. Went up first to Microsoft and then to Google and now to Amazon and studied what they were doing with generative AI, Open AI, Vertex and Bedrock. And what we are about to ship there, as you’ve written about it, is groundbreaking. We’ve patented some technology there. So, this cloud world, security and AI, are the three vectors that we feel there’s a tremendous opportunity to create a very special company here.

Daniel Newman: Well, congratulations on the progress. Let’s talk a little bit about what’s going on here. You made a pretty important announcement, a touring integration touring with Bedrock. Tell us a little bit about what this is, what this means, and why this is such a big moment for Cohesity.

Sanjay Poonen: Yeah. I think if you back up, first off, the partnership with Amazon is profound for us because three-four years before I joined, Mohit had set up both an investment of Amazon, they were the first cloud to invest in us, and we built a sort of a cloud solution that was first and preferred for Amazon. So, we’ve had a headstart of a large number of our cloud services, data protection service, security, now of course AI and capabilities that we call Cohesity Cloud services running on Amazon. So, that’s been very good for the partnership. About nine months ago, when generative AI started to become a big topic of discussion, I was playing around with the GPT, Chat GPT. And I asked it, you guys are profound thought leaders and writers and speakers. 15 years ago, I was in the analytics space. And when I came here to speak at the first conference, I was talking at SAP and my role as president about the analytics. So, I asked GPT, Chat GPT, summarize what Sanjay Poonen talked about analytics 15 years ago, and give me a half page summary. I want to see how accurate it was. And it was actually pretty close to it. Second question I asked, Patrick, you know my father. I said, my father has been very involved in planting churches that summarize what Zach Poonen has talked about the last 30 years. And it was not so bad. And I sent it to my dad, and he was like, “Wow, where did this come from?” So, I asked then myself a very natural question that sort of followed that could we use this summarization capability to summarize our own data? The world’s data is protected by Cohesity, so we are the repository of all that secondary… Think of us like a Snowflake or a Databricks for secondary data. And what if we could have a GPT type window that searched the vast amount of data that customers had. They used it, and it summarized a half page summary of everything that was in there. That could be enormously powerful.

So, at that time, it was only Open AI and Microsoft really taking the forefront. So, I went up to Microsoft, spent some time with Satya and his team, and they said, “Yeah, there’s this technique called retrieval augmented generation.” I knew nothing about it. Why don’t you and your founders download every computer science paper on RAG, study it and see if it can be applied to secondary data because it is being applied to primary data. The Cassandra type folks like DataStax and others were doing that. So, that’s what we did. And Mohit and our genius engineers downloaded, they’ve got a stack this big of all the computer science papers. And we started applying retrieval augmented generation, and we filed some patents because we fully expected as we did it, many other people would try to… We did that first with Open AI. But we, from the get go, decided we would not make it optimized only for one cloud. Very shortly there afterwards, Vertex AI, which was Google’s generative AI. We announced the Microsoft partnership in April. We announced what we’re doing with Google in August, and then we were closely working with Amazon to see when Bedrock would come available. And the moment it did. So, we’re now the only data protection vendor that has a generative AI capability optimized for all three clouds. And here, obviously, this conference is all about Bedrock. So, that use case of being able to retrieve a augmented generation to allow our… And most of our customers when they’re using the cloud are to pick Amazon as their first cloud, to allow them to be able to search their large amounts of tens or hundreds of petabytes. To summarize, maybe a query like this, if I was the Cisco of a bank, and 10 years ago, there was a breach that happened, I might want to ask my data sitting on the Cohesity platform, please summarize how we responded to that breach 10 years ago, in a half page summary. I’m going to go get a cup of coffee and come back in 10 minutes. Or better still, summarize something even deeper. I’m going to come back tomorrow morning, come back with a full summary. That was hard to impossible in the past before these generative AI techniques existed.

Patrick Moorhead: So, I think you illustrated pretty well how this can add value looking back. What type of value does it provide going forward as it relates to cybersecurity?

Sanjay Poonen: Yes, that’s tremendous. So, everything I’ve described is a e-discovery search, analytical on historical, because secondary data is usually historical. Forward looking techniques for AI in general and machine learning doesn’t have to be just generative AI. We’ve been doing for 10 years. So, our entire AI framework is called Turing. We’ve been using a number of ransomware detection, classification techniques of AI that once you’ve seen, it’s a lot like disease prevention. Detecting some kind of pattern from an existing set of signatures allows you to predict what may happen in the future. So, the more we are doing it on things that we see, but also integrating deeply with the AI that we get from and feeds we get from Microsoft or Palo Alto or CrowdStrike or Cisco who are the big, they have trillions of signals they’re seeing, allow us to then scan our secondary data at scale and then prepare it for a clean room cyber vault. So, many of those things… There’s a lot of parallels, Pat, between the way of which we fight cyber warfare and the way in which the doctors fight disease. And a lot of the warnings that you can do looking at strands of viruses that predict something about in the past COVID or whatever have you, are the exact same science that we apply in cybersecurity.

Patrick Moorhead: I got that.

Daniel Newman: It seems like being on this generative AI train is going to be the demarcation right now between the companies that are going to quickly accelerate like in your space. You’re doubling, tripling down, meaning… Pat, and I, I think we could say, we’ve talked to many of your peers, we’ve had some of them on the show and all are’s using the words AI. I don’t think right now, there’s a architecture that exists that doesn’t use ai. But I think what you’re being able to say in terms of the sophistication of generative at this level and how we’re outstripping and outpacing. Because right now, I think it’s a bit of a cat and mouse game. Cybersecurity is a cat and mouse game. And so, every tool that you’re using, Sanjay, and your team is building on, there’s some black cat that’s trying to use similar RAG or whatever to build more sophisticated tools to outstrip what you’re doing. So, it’s really interesting to watch how much it’s forcing companies like you and it’s forcing these partnerships to become more sophisticated as no company has a monolithic ecosystem.

Sanjay Poonen: Yeah. I think Dan, the way to think about it is for what we were just describing a few seconds ago with Pat, for the security use cases, everyone in our space is saying similar things. And to be honest, maybe there are text approaches that are similar. So, we’re all basically using some kind of feed to detect ransomware inside a bunch of secondary data using a variety of algorithms. We could argue whose algorithms better. But let’s assume that the techniques are similar. It’s not like the science there’s unique. There’s no one who holds a patent on my algorithm to scan is better than yours. Scanning algorithms… And then large language models allow you to use some kind of natural language to feed that query for searching. Nobody in our space is doing retrieval augmented generation. We’ve studied that, we patented that and no far have seen nobody doing that. That’s a different use case from cybersecurity. That’s search and discovery and analytics. So, our view of it is, if you think about secondary data, like the bottom of an iceberg, primary data is what sits in your files and snowflake and what have you. It’s the top of the iceberg. Second data, what’s happened, grows into the bottom of the iceberg. You want to do three things to that bottom of the iceberg. You want to manage it well. That’s what classically backup tools are done. You want to secure it really well. This conversation we just had about cybersecurity. But you also want to get insights onto that. That third piece of it, which is analytics and insights into it, not… Cybersecurity is like the brakes of a car. It helps you stop the bad guys from coming in. But just protecting that bottom of the iceberg gives you no value add if you can’t get insights into it. The insights part of it, we are uniquely doing. I have not seen anybody in the data protection space who’s using generative AI, retrieval augmented generation analytical technologies to give you insights into that data. That’s unique. And I’ve been watching this really closely. Of course, you’re seeing something else. But I’ve not seen any where else. And that we think is a huge bonanza for this company.

Daniel Newman: And I want to make sure we pivot and talk about a couple other announcements, but I just want to double click, what are those insights… When you say secondary and getting insights from it using RAG to do this. Okay? What kind of insights are your customers… Because I’d love for you to talk to the audience about that, because this is a differentiator for you.

Sanjay Poonen: It’s going to vary by vertical industry, because this becomes very quickly, a vertical industry. So, I gave you the example of a bank. What did I see 10 years ago, can you summarize that for me? That’s a search based insight. I was talking to a media company at the show who collects a lot of movies and video shows and whatever you can imagine, it’s a Fortune 100 brand. If they’ve indexed enough of that movie content, at some point in time, it’s age because it shows from 10 years ago. Summarize something from that that’s in my backup that I’d want to see. And today, being able to search and summarize video content if it’s not indexed is hard, but assume that this happens, or over time, there will even be ways to search video content just as easily as there was to search a text file or a Microsoft Word or PowerPoint document. Any question that you would want to ask on your data, you should be able to ask on that iceberg transparently knowing that there’s no boundary between primary and secondary. Why do you care that after a year, it got backed up? I want to ask something about sales, if I’m in a retailer, for all the history of time. And maybe the company can now decide, I’m actually going to back up after five days because my speed of being able to ask a question on historical data is just as fast. In the past, the reason you didn’t do that is, oh, trying to ask a question on backup data was really slow to impossible, because you had to get the data out of the backup rehydrated. Imagine now that an entire iceberg of primary and secondary data was infinitely queryable for both unstructured and structured data. It’s enormously powerful. So, by industry, people are going to have a set of unstructured and structured data questions they want to answer that we think that this is going to unlock a whole lot of value, not for us, but for the companies. And it’s their data. The other thing we all have to do, Dan and Pat, is to ensure that the way in which you ask those questions are done in a responsible way. We coined the term responsible AI. You wrote about it. I was actually excited to see Adam Selipsky talking about it. So, those will also be important in the framework of how we solve this.

Patrick Moorhead: So, I’d like to shift gears here. And by the way, thanks for the explanation. I think you are going to have to work really hard to get that message out, because it is a differentiator, and I think this is part of it, getting that message out. I want to shift gears. You’re amping up support for VMware workloads on AWS and how you’re protecting those. Can you talk us through that? You’ve already have these relationships in place, but it appears that you’re supersizing this.

Sanjay Poonen: Yeah. Listen, VMware, I have a soft corner in my heart for that, having been COO there and-

Patrick Moorhead: You know a little bit about this, don’t you?

Sanjay Poonen: A a lot of my colleagues’ still there. It turns out Broadcom is one of our top customers. When I had breakfast with Hawk 10 a few months ago, I told them, “Let’s wait for the transaction to close. But I want to double down on our partnership with VMware inside Broadcom, both as a customer but also as a partner.” The largest customers who backup… Sorry, who have VMware workloads use our platform today, among the Fortune 500 companies, the large ones. So, what we’re seeking to do is make sure that we protect those workloads wherever they are. Today, they’ve been mostly on-prem, but a growing number of them are also now in AWS or the other clouds. These were the VMware AWS. And when I was there, Raghu and I worked on all of these deals, the VMware on AWS. And Pat announced that I think in 2014 and 2015 on stage here at AWS. So, there’s a substantial number of customers who are taking advantage of VMware cloud on AWS and their analogs in the other clouds, Azure, whoever have you. We want to make sure that our data protection technologies work seamlessly across those. And all the power, what you like about Cohesity, maybe the beautiful user experience or the power to protect it and cyber recover at speed and scale, all come very tangibly with you wherever that workload is.

Daniel Newman: Excellent. All right, Sanjay. So, here’s the trick question. No, I’m playing. Do appreciate you doing the breakdown. I’ve actually learned quite a bit here-

Patrick Moorhead: Sometimes you get done with these conversations and you’re like, “I need to go find out more.”

Daniel Newman: I knew all that, but this one, I didn’t know all that. Let’s give a little glimpse into the future. I know our audience is going to be interested. Clearly, you’re taking next level techniques to differentiate your product. But where does Cohesity go from here, Sanjay?

Sanjay Poonen: Listen, for now, we have… I tell our teams we have enough strategic differentiation coming out of our ears. We have to execute.

Daniel Newman: Execute.

Sanjay Poonen: At the junction of cloud, security and AI is a very special company. And yes, there are other same things, but we have got unique differentiated tech. It’s very nice when some of our competitors are now emulating our thing. 10 years ago, we were the first to do scale out architectures, hyper-converge. Every one of the other competitors followed us. So, we invented that. We were going to continue to be ones that invent new areas and then we expect the rest of the competitors in that space to follow us. Come back to that iceberg diagram, we estimate there’s about a hundred exabytes of data in primary systems, and there’s a thousand exabytes, a zettabyte, at least in secondary systems. We want the bulk of those to be on our platform. So today, the majority of the Fortune 500, Fortune 100 of any modern platform runs on our platform, way ahead of everybody. So, we are significantly ahead of other modern players in the Fortune 100, 500. So, the largest customers rely on our platform. We need to get more of them. And the ones who are already are customers, standardized on our platform. So, I believe it’s an execution game. We are advancing our workload proliferation to multi-cloud use cases. But standing through all of this is this North Star, which is Cohesity’s platform. When you do a bake off, like a Ferrari, you boys both love F1s, we need to be the fastest car on the track for speed of cyber recovery. We’ve talked a little bit about that. I think the game of the game of the game is going to be when you test this engine relative to anybody else for speed of cyber recovery, you’re going to see Cohesity shines. You do a 10,000 VM test of how quickly can you recover. Because see, I think what’s happening in this world today is CISOs and CIOs are saying, “Listen, we have to assume we’re going to get breached. If we get breached, I want to have the best platform that can recover the fastest.” That’s where we shine. And if we do that, all these other capabilities of cloud and AI just make it work better.

Daniel Newman: We like testing stuff.

Patrick Moorhead: I like testing stuff and I love F1.

Daniel Newman: So, fast Cars and labs.

Patrick Moorhead: There we go. All right.

Daniel Newman: I like the combo.

Patrick Moorhead: Let’s do it. Yep.

Sanjay Poonen: Thank you so much, Dan and Pat.

Daniel Newman: Sanjay Poonen, thanks so much for joining us here in the Six Five.

Patrick Moorhead: Thanks.

Daniel Newman: All right, everybody, hit that subscribe button. Join us for all of the episodes here at AWS Reinvent 2023 in Las Vegas. We appreciate you stopping by The Six Five lounge, Pat.

Patrick Moorhead: Yes.

Daniel Newman: And we appreciate Sanjay Poonen joining us once again. So, hit that subscribe button. We’ll see you all really soon.

Patrick Moorhead
+ posts

Patrick founded the firm based on his real-world world technology experiences with the understanding of what he wasn’t getting from analysts and consultants. Ten years later, Patrick is ranked #1 among technology industry analysts in terms of “power” (ARInsights)  in “press citations” (Apollo Research). Moorhead is a contributor at Forbes and frequently appears on CNBC. He is a broad-based analyst covering a wide variety of topics including the cloud, enterprise SaaS, collaboration, client computing, and semiconductors. He has 30 years of experience including 15 years of executive experience at high tech companies (NCR, AT&T, Compaq, now HP, and AMD) leading strategy, product management, product marketing, and corporate marketing, including three industry board appointments.