Why Better Technology Does Not Win Hyperscaler Partnerships with Alan Chhabra

0:10Welcome to Revenue Builders, hosted by John Kaplan and John McMahon. Alan Chhabra is the Executive Vice President of Worldwide Partnerships at Cerebris Systems. Before joining Cerebris, he spent nearly a decade building MongoDB's global partner ecosystem, where partners touched more than 80 % of new business. In today's AI market, hyperscaler partnerships can determine whether an infrastructure company reaches customers at scale. In this clip, Alan explains why winning those partnerships takes more than strong technology. Leaders must understand the partner's internal priorities, competitive pressures, power structure, and existing investments.

0:45He breaks down how Cerebrus aligned its architecture with AWS Tranium and why the strongest partner strategies create value for both sides without ignoring the politics inside the account. Let's dive in. So at MongoDB, we had a challenge with the hyperscalers. At the time, for folks who are listening who are in the data space in the 2015 to 20 area, they'll know what I'm talking about. For folks that weren't, it's okay. But the hyperscalers were offering their own open source data as a service. And they were bringing in all the open source players like MongoDB and Elastic and Redis and things like Databricks, like Apache Spark.

1:32So we had to convince the hyperscalers, do not use our open source alone, partner with us and sell our enterprise versions. and it was like a lot of creating FOMO amongst each other. Like, hey, you could have our entire sales teams backing you or we could compete with you. So which hyperscaler wants to get with us first? And at the time we got Google first at MongoDB and then quickly when we were competing with Google with the other guys, AWS was like, yeah, we want in. And then we did a massive partnership with them and then Microsoft as well. So same thing here, right? So if I unwind your question, cracking the hyperscalers is extremely political.

2:18First, they have deep relationships with NVIDIA. Deep. And they don't want to upset the supply chain they have with NVIDIA. Second, they're not stupid. They're creating their own chips. Okay? Like I mentioned before, they have their own AI chips. They're becoming, you know, they want to be more relevant in the stack. Not only are they just cloud providers, they're writing their own language models, even though they're partnering with Anthropik and OpenAI. They're creating their own models and they're developing their own chips. So AWS has Tranium, Microsoft has Maya, and Google has their TPUs.

2:57And like, so number one, they have deep relationships with NVIDIA and they don't want to upset NVIDIAs because if they somehow their supply gets disrupted, that's a big problem. second when you say that because there's a shortage and if nvidia says okay well we're going to put you on the slow boat yeah it's a lot about supply and demand right now right it is not perfect i mean video has the same like they can't get make enough of what they want like no one can right so so you have to kind of understand the politics of there so they want to keep nvidia as a good partner they're not stupid they're building their own chips and then you have this competitive advantage cerebris that comes into the fold.

3:37So what we decided to do with AWS is come up, and this is a lot of like kind of what going back to MongoDB, where we came up with a joint offering where they still were selling Mongo, but we were working on it as an OEM platform. Our sellers were compensated to work with them. Their sellers were compensated work with us. In this model, we use technology a little bit differently, where when you go to AWS in the future and want Cerebris, the architecture is a combination of Tranium plus Cerebris. Tranium for input of tokens, Cerebris for output of tokens. Tranium for pre-fill, asking it, you know, pre-fill and Cerebris for the decode, the answers.

4:28And that architecture still is just as fast. It leverages AWS's own IP, which is Tranium. And it's probably even better for price performance. So when I work with these phenomenal leaders at AWS, like Peter DeSantis and Dave Brown and Rubel Borno and Posada runs Data Center, they love that architecture, right? Because it is using their own IP as well as ours. It's competitive to GPUs and video, but they're not just like bringing in another third party. were almost working in tandem on the architecture. So it worked. So AWS signed a big deal with us and we're rolling that out. That's all public this year.

5:20So we had to understand the politics around it. You can't just like walk in and be like, hey, I have a faster widget. Why don't you use it? That's hard to do at AWS. You really have to kind of come up with the angle that they're going to buy into. That's a win for them long-term. And what does that do to you as being an expert on how these channels work. First of all, the masterclass that you just gave on the stack is amazing. I can't imagine that anybody could be in a role like yours and not being able to articulate that. That's a must because you just put the stack in there, but now you're putting the relationships in there and you're putting the money in there.

6:00And some things that don't make sense, like the models, the frontier models that made all the money in the beginning and they have all this influence and all this money and they're buying their way into this. I don't want to be disparaging, but they're influencing. A lot of them have their own VC funds that are sending money up the channel, right? Yes. So what does it really take today to now kind of explain your role? Like the definition of your role is so much is like, it's like a superhuman. Seriously. way john mcmahon he'll tell you i'm not superhuman but but i do but he always has been a master like even going back to mongo db for the audience you know aws for instance had a competitive product and they wanted to come right after mongo db and he went and did a big deal with aws where they were selling LongoDB.

6:59So he, Alan does go into these sales situations very strategically. He doesn't just show up and throw up. He's, he's knowing exactly the angle he's going to take when he goes in the door. It's the hashtag of coopetition. Like you have to be a student of under, you started this conversation with landscapes. And I think that's so appropriate. I think if you don't understand number one, the landscape that you're in and you don't understand your customers and your partner or the landscape that each, the lens that each one of these layers has, there's no way that you can be successful. So it's not a typical - Think about how the enemy inside the account, right?

7:40So you're talking about you going and selling your product into, let's say, the hyperscale. And the hyperscaler has a product also. So someone inside that hyperscaler is essentially your enemy. And they'll politically try to do anything to kill you off. to not allow you into their company, right? So you're going up against that. Talk about having to get strong champions and how to neutralize enemies. There's a masterclass right there. Well, your champion in one situation could be you're an enemy in another, and that's just reality, right? And an enemy doesn't mean wailing and gnashing of teeth.

8:16It's just not in their best interest to do X with you, but is in their best interest to do Y. Correct, Alan? I agree. I mean, so I think if you're a CEO of a AI company that is coming out and you're trying to, every CEO is going to be like, yeah, I need a partner leader. Okay. It's like a checkbox. You need a sales leader. You need a partner leader. Obviously, you need all the engineering folks. Like that's the core. But then you need these two other people. Okay. And you need a CMO and a marketing leader. If I'm a CEO of an AI company and you come across a candidate, the candidate you should first ask, tell me about the landscape of my business.

9:04And if they can't answer what we went through, they haven't done their homework. Or if you're like a salesperson, you haven't done your research in the account. So I think it's so important before you just like jump in and do a bunch of activity and do a bunch of pipeline generation and set up a bunch of meetings for your CEO at a conference. Any candidate and CEO of an AI company that's going to hire for partners needs to first show them the landscape. Ask them the interview question. What's the landscape of the of all the potential partners we could have and why relevant to us? the second test I would give anyone give me the org chart not like all 20 ,000 people but at the top level show me the org chart of the number one partner you're going to work with at our company and if that's AWS show it to me so I do that all the time I mean Andrew did that with me right so when I joined Cerebus hey I said I don't know if he asked me but I just showed it to him Let me show you the power base of these brilliant people at AWS.

10:13And what you realize is it's like Game of Thrones. There's five kingdoms, six kingdoms. Oh, good. Some kingdoms work for the parent, for Amazon, all the way up to Andy Jassy. And others work for the subsidiary, AWS, up to Matt Garman. And within AWS, there's different components as well, right? There's great leaders like Dave Brown, like Prasad, like Swami, like Ruba, like Greg Pearson. So it's not about being superhuman. But I do respect that, what John's saying, like we had to learn a lot at Mongo the hard way. But it's about doing your research. and all these you also have to have political capacity meaning you can't look at one deal or one relationship you're not yeah if you look at one deal by itself you never get it so what john didn't tell you at mongo one of the ways we cracked in the hyperscalers was not just partnering with their competitor we partnered with alibaba in china so both you guys know we had a great leader who worked for me, Steve Su in China.

11:33So I was on a plane to Beijing, Guangzhou, Shenzhen. I think I went like seven times while I was there. And we cracked into Alibaba. And the last thing AWS or Azure wanted was MongoDB to have a great partnership with the Chinese before we had a good one with them too. So same thing in this new world, right? When we crocked in AWS, we embraced Tranium. Tranium plus Cerebrus to compete with NVIDIA. A lot of neutralization of enemies if you can do something like that.

12:13Alan Chhabra: Thanks for listening to today's episode. If you enjoy the content, please subscribe. Rate and review the show to help us reach more people. This show is brought to you by Force Management, where we help companies improve sales performance, executing the growth strategy at the point of sale. Check out forcemanagement.com for more information.