(upbeat music) >> Welcome back, everyone, to theCUBE's live coverage, day one wrap-up. I'm John Furrier, with Dave Vellante. We have Jerry Chen, special guest who's been with us every year on theCUBE since inception. Certainly every AWS re:Invent, nine years straight. Jerry Chen, great to see you for our guest analyst's wrap up VC general partner, Greylock partners, good to see you. >> John, Dave, it's great to see you guys. Thanks for having me again. It wouldn't be re:Invent without the three of us sitting here and we missed last year, right, because of COVID. So we have to make up for lost time. >> John: We did a virtual one- >> Dave: we did virtual stuff= >> John: wasn't the same as in-person. >> Dave: Definitely not the same. >> Jerry: Not the same thing. So, it's good to see you guys again in person, and less than 6 feet apart. >> Cheers, yeah. >> And 7,000 people here, showing that the event's still relevant. >> Jerry: Yeah. >> Some people would kill for those numbers, it's a bad year for Amazon, down from 60,000. >> Jerry: Yeah. >> So, ecosystem's booming. Okay, let's get to it. Day one in the books, new CEO, new sheriff in town, his name's Adam Selipsky. Your take? >> Well, Adam's new, but he's old, right? Something, you know, like something new, something old, something blue, right? It's so, Adam was early Amazon, so he had that founding DNA. Left, you know, CEO of Tableau, acquired by Salesforce, came back few months ago. So I think it was a great move, because one, he's got the history and culture under Jassy, so he's definitely the Bezos Jassy tree of leadership, but yet he's been outside the bubble. Right? So he actually knows what it means to run a company not on the Amazon platform. So, I think Adam's a great choice to lead AWS for what we call it, like maybe act two, right? Act one, the first X years with Jassy, and maybe this is the second act under Adam. >> Yeah. And he's got- and he was very technical, hung around all the techies, James Hamilton, DeSantis, all the engineers, built that core primitives. Now, as they say, this cloud next gen's here, act two, it's about applications. >> Jerry: Yeah. >> Infrastructure as code is in place. Interesting area. Where's the growth come from? So, look, you know, the ecosystem has got to build these super clouds, or as you say, Castles on the Cloud, which you coined, but you brought this up years ago, that the moats and the value has to be in there somewhere. Do you want to revise that prediction now that you see what's coming from Selipsky? >> Okay, well, so let's refresh. Greylock.com/castles has worked out, like we did, but a lot of thought leadership and the two of you, have informed my thinking at Castles in the Cloud, how to compete against Amazon in the cloud. So you'd argue act one, the startup phase, the first, you know, X years at Amazon was from 2008 to, you know, 2021, the first X years, building the platform, digging the moats. Right? So what did you have? You have castle the platform business, economies of scale, which means decreasing marginal costs and natural network effects. So once the moat's in place and you had huge market share, what do you for act two, right? Now the moats are in place, you can start exploring the moats for I think, Adam talked about in your article, horizontal and verticals, right? Horizontal solutions up the stack, like Amazon Connect, CRM solutions, right? Horizontal apps, maybe the app layer, and verticals, industrials, financials, healthcare, et cetera. So, I think Jassy did a foundation of the castle and now we're seeing, you know, what Adam and his generation would do for act two. >> So he's, so there's almost like an act one A, because if you take the four hyperscalers, they're about, maybe do 120 billion this year, out of, I don't know, pick a number, it's many hundreds of billions, at least in infrastructure. >> Jerry: Correct. >> And those four hyperscalers growing at 35% collectively, right? So there's some growth there, but I feel like there's got to be deeper business integration, right? It's not just about IT transformation, it's about deeper- So that's maybe where this Connect like stuff comes, but are there enough of those? You know, I didn't, I haven't, I didn't hear a lot of that this morning. I heard a little bit, ML- >> Jerry: Sure. >> AI into Connect, but where's the next Connect, right? They've got to do dozens of those in order to go deeper. >> Either, Dave, dozens of those Connects or more of those premise, so the ML announcement was today. So you look at what Twilio did by buying Segment, right? Deconstruct a CRM to compete against Adam Selipsky's old acquire of Salesforce.com. They bought Segment, so Twilio now has communicates, like texting, messaging, email, but all the data come from Segment. >> Dave: With consumption-based pricing. >> With consumption-based pricing. So, right? So that's an example of kind of what the second act of cloud looks like. It may not look like full SaaS apps like Salesforce.com, but these primitives, both horizontally vertically, because again, what does Amazon have as an asset that other guys don't? Install based developers. Developers aren't going to necessarily build or consume SaaS apps, but they're going to consume things like these API's and primitives. And so you look around, what's cloud act two look like? It may not be VM's or containers. It may be API's like Stripe and Billing, Twilio messaging, right? Concepts like that. So, we'll see what the next act at cloud looks like. And they announced a bunch of stuff today, serverless for the data analytics, right? So serverless is this move towards not consuming raw compute and storage, but APIs. >> What about competition? Microsoft is nipping at the heels of AWS. >> Dave: John put them out of business earlier today. [John and Dave Laugh] >> No, I said, quote, I'll just- let me rephrase. I said, if Amazon goes unchecked- >> Jerry: Sure. >> They'll annihilate Microsoft's ecosystem. Because if you're an ISV, why wouldn't you want to run on the best platform? >> Jerry: Sure. >> Speeds and feeds matter when you have these shifts of software development. >> Jerry: You want them both. >> So, you know, I mean, you thought about the 80's, if you were at database, you wanted the best processor. So I think this Annapurna vertical integrated stacks are interesting because if my app runs better and I have a platform, prefabricated or purpose-built platform, to be there for me, I'm going to build a great SaaS app. If it runs faster and it cost less, I'm going to flop to Amazon. That's just, that's my prediction. >> So I think better changes, right? And so I think if you're Amazon, you say cheaper, better, faster, and they're investing in chips, proprietary silicon to run better, faster, their machine learning training chips, but if you're Azure or Google, you got to redefine what better is. And as a startup investor, we're always trying to do category definition, right? Like here's a category by spin. So now, if you're Azure or Google, there are things you can say that are better, and Google argued their chips, their TensorFlow, are better. Azure say our regions, our security, our enterprise readiness is better. And so all of a sudden, the criteria "what's better" changes. So from faster and cheaper to maybe better compliance, better visibility, better manageability, different colors, I don't know, right? You have to change the game , because if you play the same game on Amazon's turf, to your point, John, it- it's game over because they have economies of scale. But I think Azure and Google and other clouds, the superclouds, or subclouds are changing the game, what it means to compete. And so I think what's going on, just two more seconds, from decentralized cloud, being Web 3 and crypto, that's a whole 'nother can of worms, to Edge compute, what Cloudflare are doing with R2 and storage, they're trying to change the name of the game. >> Well, that's right. If you go frontal against Amazon, you're got to get decimated. You got to move the goalposts for better. And I think that's a good way to look at it, Dave. What does better mean? So that's the question that's on the table. What does that look like? And I think that's an unknown, that's coming. Okay, back to the start-ups. Category definition. That's an awesome term. That to me is a key thing. How do you look at what a category is on your sub- on your Castles of the Cloud, you brought up how many categories of- >> Jerry: 33 markets and a bunch of submarkets, yeah. >> Yeah. Explain that concept. >> So, we did Castle in the Clouds where my team looked at all the services offered at Azure, Google, and Amazon. We downloaded the services and recategorized them to like, 30 plus markets and a bunch of submarkets. Because, the reason why is apples to apples, you know, Amazon, Google, Azure all have databases, but they might call them different things. And so I think first things first is, let's give developers and customers kind of apples to apples comparisons. So I think those are known markets. The key in investing in the cloud, or investing in general, is you're either investing in budget replacement, replacing a known market, cheaper, better database, to your point, or a net new market, right? Which is always tricky. So I think the biggest threat to a lot of the startups and incumbents, the biggest threat by startups and incumbents, is either one, do something cheaper, better in a current market, or find a net new market that they haven't thought about yet. And if you can win that net new market before the rest, then that's unbelievable. We call it the, you know, the blue ocean strategy, >> Dave: Is that essentially what Snowflake has done, started with cheaper, better, and now they're building the data cloud? >> Jerry: I think there's- it's evolution, correct. So they said cheaper, better. And the Castle in the Cloud, we talked about, they actually built deep IP. So they went a known category, data warehouses, right? You had Teradata, Redshift, Snowflake cheaper, better, faster. And now let's say, okay, once you have the customers, let's change the name of the game and create a data cloud. And it's TBD whether or not Snowflake can win data cloud. Like we talked about Rockset, one of my investments that's actually move the goalpost saying, oh, data cloud is nice, but real time data is where it's at, and Snowflake and those guys can't play in real time. >> Dave: No, they're not in a position to play in real time data. >> Jerry: Right. >> Dave: I mean, that's right. >> So again, so that's an example of a startup moving the goalpost on what previously was a startup that moved the goalpost on an incumbent. >> Dave: And when you think about Edge, it's going to be real-time AI inferencing at the Edge, and you're right, Snowflake's not set up well at all for that. >> John: So competition wise, how do the people compete? Because this is what Databricks did the same exact thing. I have Ali on the record going back years, "Well, we love Amazon. We're only on Amazon." Now he's talking multicloud. >> So, you know, once you get there, you kind of change your tune cause you've got some scale, but then you got new potential entrants coming in, like Rockset. >> Jerry: Correct. >> So. >> Dave: But then, and if you add up the market caps of just those two companies, Databricks and Snowflake, it's much larger than the database market. So this, we're defining new markets now. >> Jerry: I think there's market cap, especially Snowflake that's in the public market, Databricks is still private, is optimism that there's a second or third act in the database space left to be unlocked. And you look at what's going on in that space, these real-time analytics or real-time apps, for sure there's optimism there. But, but to John's point, you're right, like you earn the right to play the next act, but it's tricky because startups disrupt incumbents and become incumbents, and they're also victims their own success, right? So you're- there's technical debt, there's also business model debt. So you're victims of your own business model, victims of your own success. And so what got you here may not get you to the next phase. And so I think for Amazon, that's a question. For Databricks and Snowflake, that's a question, is what got them here? Can they play to the next act? And look, Apple did it, multiple acts. >> John: Well, I mean, I think I- [Crosstalk] >> John: I think it's whether you take shortcuts or not, if you have debt, you make it a little bit of a shortcut bet. >> Jerry: Yeah. >> Okay. That's cool. But ultimately what you're getting at here is beachhead thinking. Get a beachhead- >> Jerry: Correct. >> Get in the market, and then sequence to a different position. Classic competitive strategy, 101. That's hard to do because you want to win the beachhead- >> I know. >> John: And take a little technical debt and business model debt, cheat a little bit, and then, is it not fortified yet? So beachhead to expansion is the question. >> Jerry: That's every board meeting, John and Dave, that we're in, right? It's called you need a narrow enough wedge to land. And it is like, I don't want the tip of the spear, I want the poison on the tip of a spear, right? [Dave and John Laugh] >> You want, especially in this cloud market, a super focused wedge to land. And the problem is, as a founder, as investor, you're always thinking about the global max, right? Like the ultimate platform winner, but you don't get the right to play the early- the late innings if you don't make it out of the early innings. And so narrow beachhead, sharp wedge, but you got to land in a space, a place of real estate with adjacent tan, adjacent markets, right? Like Uber, black cars, taxi's, food, whatever, right? Snowflake, data warehouse, data cloud. And so I think the key with all startups is you'll hit some ceiling of market size. Is there a second ramp? >> Dave: So it's- the art is when to scale and how fast to scale. >> Right. Picking when, how fast, in which- which best place, it was tough. And so, the best companies are always thinking about their second or third act while the first act's still going. >> John: Yeah. And leveraging cloud to refactor, I think that's the key to Snowflake, was they had the wedge with data warehouse, they saw the position, but refactored and in the cloud with services that they knew Teradata wouldn't use. >> Jerry: Correct. >> And they're in. From there, it's just competitive IP, crank, go to market. >> And then you have the other unnatural things. You have channel, you have installed base of customers, right? And then you start selling more stuff to the same channel, to the same customers. That's what Amazon's doing. All the incumbent's do that. Amazon's got, you know, 300 services now, launching more this week, so now they have channel distribution, right? Every credit card for all the developers, and they have installed base of customers. And so they will just launch new things and serve the customers. So the startups had to disrupt them somehow. >> Well, it's always great to chat with Jerry. Every year we discover and we riff and we identify, in real time, new stuff. We were talking about this whole vertical, horizontal scale and kind of castles early on, years ago. And now it's happened. You were right. Congratulations. That's a great thesis. There's real advantages to build on a cloud. You can get the- you can build a business there. >> Jerry: Right. >> John: That's your thesis. And by the way, these markets are changing. So if you're smart, you can actually compete. >> Jerry: I think you beat, and to Dave's earlier point, you have to adapt, right? And so what's the Darwin thing, it's not the strongest, but the most adaptable. So both- Amazon's adapt and the startups who are the most adaptable will win. >> Dave: Where are you, you guys might've talked about this, where do you stand on the cost of goods sold issue? >> Jerry: Oh, I think everything's true, right? I think you can save money at some scale to repatriate your cloud, but again, Wall Street rewards growth versus COGS, right? So I think you've got a choice between a dollar of growth versus a dollar reducing COGS, people choose growth right now. That may not always be the case, but at some point, if you're a company at some scale and the dollars of growth is slowing down, you definitely have to reduce the dollars in cost. And so you start optimizing cloud costs, and that could be going to Amazon, Azure, or Google, reducing COGS. >> Dave: Negotiate, yeah. >> John: Or, you have no visibility on new net new opportunities. So growth is about new opportunities. >> Correct. >> If you repatriating things, there's no growth. >> Jerry: It's not either, or- >> That's my opinion. >> Jerry: COGS or growth, right? But they're both valid, definitely, so you can do both. And so I don't think- it's what's your priorities, you can't do everything at once. So if I'm a founder or CEO or in this case investor, and I said, "Hey, Dave, and John, if you said I can either save you 25 basis points in gross margin, or I can increase another 10% top line this year", I'm going to say increase the top line, we'll deal with the gross margin later. Not that it's not important, but right now the early phase- >> Priorities. >> Jerry: It's growth. >> Yeah. All right, Jerry Chen, great to see you. Great to have you on, great CUBE alumni, great guest analyst. Thanks for breaking it down. CUBE coverage here in Las Vegas for re:Invent, back in person. Of course, it's a virtual event, we've got a hybrid event for Amazon, as well as theCUBE. I'm John Furrier, you're watching the leader in worldwide tech coverage. Thanks for watching. (Gentle music)