VAST Data paves way for vast data storage
VAST Data paves way for vast data storage
About this video
Key Insights:
Vast data specialises in data storage components of a modern data center, aligning with new artificial intelligence capabilities
Vast Data enjoys pronounced growth in its valuation, a trend mirrored by Nvidia, OpenAI, and Coreweave
The global big data opportunity has reportedly expanded by approximately 20 times, largely due to advancements in artificial intelligence
Jeff Denworth, co-founder of Vast Data, shares that their company specialises in data storage aspects of a modern data center, allowing customers to store, protect, and process their data effectively. Jeff claims this approach appropriately aligns with the current wave of artificial intelligence capabilities. He differentiates Vast Data from conventional data centers, presenting the company as a subcomponent of extensive data centers that focuses on data storage.
Vast Data has recently witnessed a significant boost in its valuation, with Jeff mentioning notable increases throughout 2023 - a trend mirrored by organisations such as Nvidia (NASAQ: NVDA), OpenAI, and Coreweave. He attributes this success to the new wave of deep learning and artificial intelligence, reaffirming their proactive partnership with Nvidia to power AI supercomputers that demand large quantities of data.
According to Jeff, Vast Data's platform operates by integrating data organisation and storage with a sophisticated database platform. This methodology allows a convenient translation of raw data into constructive answers. He believes the global big data opportunity has expanded by approximately 20 times, largely due to advancements in artificial intelligence. Moreover, they implement elements of generative AI within their products for specific tasks such as anomaly detection. Jeff also shares insights about sustainable ideas and efficient software to counter the increasing power demand in data centers.
Below is an unedited transcript of the interview:
Data Infrastructure, a company of Vast data, recently completed a US $118 million funding round which valued the AI data platform firm at over US $9 billion. And it is also looking to be a listed soon to tell us more. Vast data's co-founder Jeff Denworth, joins me up here on the desk. Jeff, thank you so much for coming in. Thank you for having me. And welcome to Australia. Oh, thanks. Fresh off the plane, are you? Pretty much yeah. Oh my golly gosh. You must be feeling a bit green. So anyway, this sounds like an amazing story. You started co-founded Vast Data in 2016 as a modern data storage company. Can you explain what is a modern data storage company? And particularly differentiating against a data center, which is an asset class that Australian investors are familiar with? Sure, sure. So. So if you think about your average data center today, typically you
1:00
find three primary infrastructure components. One is you have compute, which is where all of your processing is happening. Two is you have networking which is connecting all the different computers together. And then you have the data storage environment, which is basically where all your data lives. So you think about a modern data center. The data component is really what vast data specializes on. And our job is to basically help customers store, protect and process their data better than they possibly ever could in a manner that we think is very well aligned for this new wave of artificial intelligence that's now hitting us all over the world. So are you actually differentiated from what a data center would do, or you've just taken one aspect of what some data center operators do, like in Equinix or the other big boys around town. Yeah, just again, maybe separated it out a bit. So we think of data centers as almost property. Yeah. Where there's a ton of connectivity and there's a ton of power flowing into these large scale warehouses. And in those warehouses are systems systems
1:59
for processing, systems for networking and then systems for storage. And so we're basically a subcomponent of a large scale data center. Okay. Your subcomponent okay. Cool. So this latest funding round are $9 billion market cap. So you were saying that the valuation in this fundraising has actually gone up for vast data. Is that the case. Yeah. 2023 was a pretty modest year in terms of valuation increases, particularly in the startup community. And so we announced in December that we had grown our valuation by about two and a half x since the previous round. That happened in 2021. So you're talking about roughly almost three x growth in two years time when the market is is down. There's very few organizations around the world that saw that same increase in valuation. Uh, Nvidia is a great example of a processing company. Uh, there are two, um, large language model companies, OpenAI and anthropic that saw that same valuation
2:59
increase. Vast did and then there's one other company, which is a data center company called Core we've, which is basically next generation born in the AI era company that focuses on hosting large amounts of Nvidia GPUs. Uh, interestingly enough, Nvidia as an investor and vast as its core. We've a customer have asked so so a very small number of organizations over a few billion dollars in valuation that saw that that growth. And a lot of it's really driven by this new wave of deep learning. Absolutely. And you were at the Nvidia conference recently as well. That's my understanding. So maybe just slot in because lots of people own Nvidia shares. And if they don't, they probably do through their retirement savings or whatever. Um, so how do you slot in apart from them having a stake in yourselves? Sure. So they're an investor, as you mentioned, but we also are a design partner with Nvidia, where we work with them to help build and power some of the world's hungriest AI supercomputers. Right. So all of these machines are either
3:59
training or inferring next generation language models, and they need to be fed with a ton of data in order to make these models more accurate. And that's really where we come in. So we've built a few new reference architectures using Nvidia's next generation technologies to make it kind of easier and more scalable to build these large scale AI supercomputers. And it's just been a spectacular partnership that we've enjoyed with them. It's really interesting because there are still a lot of investors and that are just very cynical about the whole thing, right? And they just sit, sit, sit there and look at the Nvidia share price and go, it's a bubble. Can you drill down? You're on the ground. You're in America where it's all absolutely happening. Can you explain the enormity almost, of what Jamie Dimon said in his newsletter over a night? Because it just in my opinion, there seems to be a real disconnect about this really is changing
4:59
the. World business on such a large scale, and it's something that we haven't seen for 20 odd years. Jamie equated it to the printing press. If you saw that electricity on the internet. Yeah. So, um, so actually there's an investment firm called Ark Capital that many years ago broke down their investment thesis. And they said organisations will roughly pay for, um, in software, about 20% of the value of, uh, an employee. If you can show that, you can bring an employees level of efficiency. And so if you think about it with respect to artificial intelligence, what's happening is that you're just seeing these radical efficiency gains. And if I go to Jamie's newsletter, he basically said, I'm not quite sure exactly what shape all of this will take when everything is said and done, but we can see some radical improvements in how we actually work, and the types of jobs that we take are going to change and evolve and be accelerated. And, um, new jobs will emerge that people have to really figure out how to embrace AI and,
5:59
and that type of, um, kind of global event where we see just these radical efficiency gains, uh, it will be translated in terms of dollars, right? There will be winners and losers that really, um, get paid on all of these new improvements and, and automations that that will be possible because of AI. And, uh, you know, the interesting thing is, if, you know, I think those same investors would have said that same thing about Nvidia last year at this time. And it continues to run. And so so as you mentioned, we're we're kind of in the center of it in the United States. We also are fortunate to work in some of the world's largest AI projects. And these projects are roughly ten times bigger than they were even two years ago, when that's kind of like, uh, that was a that was a forward indicator for what you would see in financial results last year. Um, some of the, the scale and the scope of the projects that we're seeing now is astronomical. I think probably the most, um, the most public one is you saw, uh, OpenAI and Microsoft talking about building $100 billion, $100
6:58
billion. I've been I've been banging on about that for anyone that'll listen, because a data center is typically $1 billion. Yeah, yeah. So you're talking about 100 fold bigger. It's massive isn't it? It's crazy. It's crazy. That's really interesting. And I don't want to get too far away from from what you guys are doing. So in terms of your expertise you're coming in and it's almost like your expertise is around the data that comes in. The analyzing it, helping the large language models achieve the outcomes. Is that sort of a fair enough summary? That's correct. On a very simplistic level, I think we use the term data platform. Um, and we're a software company that basically, uh, stitches together a form of, of, of data organization, data storage. Uh, we couple that with the next generation database platform. So now I've got my raw data that I might ask questions to if I'm using a large language model. And then I've got the context that gets stored in a
7:58
database that can quickly answer questions from. And then there's a computational layer that we build in software, in containers that allows for you to kind of translate that raw data into something that you can actually answer. And so I think it's really interesting for people that have been invested in big data companies for the last, you know, 5 or 10 years. You think about companies like snowflake as being a very public and prominent IPO. They really focused on business intelligence for the types of data that you could fit inside, like a data warehouse. Yeah. Um, so numerical data that fits very cleanly in tables. But if you think about what Nvidia is doing and their peers, they're now with new processors and new algorithms able to understand data types that never were understandable by systems, video and imagery and audio and free text. And so that data type is roughly 20 times what the kind of structured data warehouse data size was historically. And so if you kind of take two steps back and you say, what's really
8:58
happening is that the global big data opportunity is roughly 20 times larger now thanks to artificial intelligence. Um, how much in your operations as a software company, do you start to employ generative AI into what you're doing? I think to help you. Sure. With the outcome? Sure. I think, uh, in fact, we we embedded into our products probably more than we use it in the design team. Right. Uh, and so there's there's things like anomaly detection that we can apply into the platform so that, for example, you know, one of the the big cornerstones of the vast stories around securing your data. And if we can use machine learning to say, okay, something strange is happening here, and let's notify the users and the administrators of something happening, then it's really valuable for us in terms of accelerating our code development. We're not yet there though. And is that something that you would look to or. Oh, we already are. We're exploring it right now, but we're trying to do it in a relatively measured pace,
9:57
as I think most enterprise software. Development firms are. So there seems like so much money going into this space. One of the things that seems to crop up in my mind is delivering the infrastructure to accommodate all this development, be it particularly on the energy side, power. Yeah. It just seems to me that one's standing out there and it's got to be clean power. Two thoughts around that. Um, well, you're seeing a lot of progress in this space, actually. There's, uh, there's there's data center companies that are building their infrastructure right next to nuclear power plants. Um, we work with a lot of companies that are building sustainable data centers. One actually in Australia, they're called Dug technology, and they're powering this with geothermal energy. And so, so really sustainable ideas are starting to emerge. Um, but there's there's just an absolute limit to the, the amount of power in the world. And these chip companies are now starting to look at things, saying if the demand is what we expect it to be, you know, we have a global power deficit versus
10:57
what's required to even hit our business projections. And so it's it's pretty crazy times. Um, our focus is on using software to make these systems run more efficiently, such that you need less power to get to the same objective. And that's one of the reasons that we're so tightly partnered with Nvidia. Okay, so you're out in Australia. Is this sort of a marketing tool for you? Uh, no, it's a business tour. So we've been fortunate to have many customers across the region, uh, all sorts of different walks of life we work with, for example, like the Walter and Eliza Hall, uh, Research Institute doing Live Science and applying AI for that, uh, all the way to the Australian federal government, where we work with both civilian and also security initiatives that are happening around the country. So, um, and everybody in between. So, so really just here to, to meet with our customers and kind of continue to improve the business posture of the company. We have now a few dozen people in the country so really excited about the activity. And uh, following actually GTC, we saw a lot of our customers
11:57
that are in Australia come back saying, we now understand the kind of the the ambition of the world, and we're looking to apply that in theater. So that's one of the other reasons I'm here. Oh, fantastic. Uh, tell us about the IPO. Is that something you want to do, or is it just market conditions or obviously a lot more favorable? There are more IPOs happening this year. There are, uh, it seems like the dam is just starting to break a little bit. Uh, from a business posture perspective, we're we're we're feeling pretty good. The company is one of the fastest growing infrastructure profitable in history. We're not profitable, but we are cash flow positive. And so if you look at the kind of the dynamics of startups is, they they start to kind of grow into the public markets. Typically it's always a trade off between growth and cash. And so in our case, we've kind of worked out this very unique business model where we maintain cash flow positivity while still growing by 2 to 3 x per annum. Uh, and so that's obviously a story that investors love. And
12:57
as the market seems to be kind of favorable to our story, we're definitely going to pursue that as hard as we can.