Embedded FinOps Takes Center Stage with Pankaj Gupta | Platform Engineering 2.0 Ep 4

Hey everyone. Welcome back. I'm Alan Schimmel of Techstrong TV, and I'm joined by my good friend, Pankaj Gupta of Broadcom.

0. org website. And then we're doing a follow-up here at Techstrong with Pankaj and the Broadcom team too, on what exactly is that platform.

But in the meantime, we're conducting this series of 10-minute videos. This is actually episode four. tv or the Techstrong TV YouTube channel or OTT channel.

But for now, let's welcome Pankaj. My friend, how are you? I'm doing wonderful.

Thank you for inviting me. It's a pleasure to have you here. So Pankaj, before we jump in where we left off, which is we're going to start with pillar three of the five pillars, this one embedded FinOps.

0. Absolutely. Platform engineering adoption has been very universal across all the IT organization, but the maturity varies.

And platform engineering was primarily built for single persona, which is the developer, and also primary focus on increasing the developer's productivity. In last few years, fundamental big shifts are happening mostly driven by AI and agentic AI, which requires a platform to support AI natively, which requires a very strong embedded FinOps, which we are going to talk today. Also requires a very deep security embedded into the platform to complement shift left security, and also the composability for that.

0, which is the evolution, not a reset today. And today we are going to talk about the third pillar, which is the FinOps- Perfect ... which has to be embedded instead of bolt-on.

Yes. Well, you know what? I racked my brain as you were talking, trying to think of where has FinOps been embedded.

Because it was certainly bolted onto cloud, right? Unfortunately, FinOps suffers a similar fate as security does. People don't get serious about it until they've been bit, right?

" And now all of a sudden, well, we better get religion around FinOps. Why is it an afterthought? I think when the originally platform engineering started, it is a primary focus for developer productivity, and that's absolutely right.

In last few years, few drivers are driving much bigger need for FinOps. One is there is a significant cloud waste. Independent of which report you read, you see that 30 to 35% cloud infrastructure waste is already happening.

There is a lot of unpredictability associated with the cloud bills for that. There is a very limited visibility for that. 50% plus developers completely ignore for a FinOps requirement for that, and it is even very hard to implement the FinOps for that.

And in last two, actually last six months, you see the whole requirement evolving from token maxing to token economics for that. And you must have also seen in last few quarters that hardware pricing, especially driven by the memory pricing going up because of AI, is resetting lot of budgets for IT organizations. So this is the critical moment when the embedded FinOps is so critical for any platform engineering or any IT organization.

Absolutely. You know the old saying, Pankaj, watch after your pennies and the dollars take care of themselves. With FinOps, there's an analog to that, right?

The CFO, the CIO, the powers that be can talk FinOps, put in processes and policies, but really FinOps has to take place down at the penny level, not the dollar level. What I mean by that is you got to make every developer, every person using the platform, not just developers, I hate throwing everything on the developer's shoulders. Every platform engineer, every SRE, every DevOps person, everybody in the organization have to become FinOps aware, a FinOps practitioner.

Exactly. We don't give them a course, but yeah. What's your take on that?

Exactly. You make two great points. One is that infrastructure is where the cost originates.

Infrastructure is where the cost has to be governed for that. Fundamental shift for FinOps has to happen. It has to move from visibility to influencing the decisions as provisioning.

When I spin up a new environment, it should be able to tell me that if you do it in public cloud, this is your going to be cost. If you do it in private cloud, you are going to save 35%. If you are going to use the spot pricing, this is the way you are going to save the money.

If you spin up a specific environment, it should be able to tell you that this is the way you can save the money for that, or this is your budget. You can go up to 10%. Above that, you require the exception from your management.

So that's one big part of self-service FinOps embedded into platform by default. It has to give a real-time cost attribution. It has to have pre-deployment cost gates for that.

0 has to do. Next one is all about tokenomics for that. Third area which we see the autonomous janitors and resizing agents for that to further optimize.

And if you do these things with minimum friction to the user, then every developer, every operator becomes the DevOp practitioner. A need for financial accountability or FinOps is very strong right now in this environment, and further fueled with the hardware cost escalation, supply chain issues, budgets being flat. With same budget, CIOs are being asked to do the AI projects for that.

So it's perfect storm to embed the platform now. Agreed. When I co-founded the companies, even here when I'm running Textstrong, I always instruct my people, we want actionable intelligence.

Don't just spit me out information. Give me something we can act on. And now in the age of agentics, I'll let my agent act on it and take that one step further with automation.

I think that this is a very important piece, and you touched on it in what you said about FinOps. It's no longer enough to just report on my cloud spend, my platform cost. How do I actually convert that to actionable intelligence to prevent runaway costs, to stay in budget, to make sure I stay on that line?

Fully agree. Yeah. I think, and as I said, in the age of agentics, we could take it one step further.

We could really have the agents do that. So- And the FinOps also has a huge cultural change which organization have to drive. Yeah.

FinOps, multiple teams have to work together for that. It is not just the developer. That's a bigger change for CIO or the CTO or the platform engineering leaders to drive the economic accountability and FinOps inside the organization.

And that goes to the heart of platform engineering, too, right? One of the reasons I think we saw the rise of platform engineering coming out of the DevOps movement was you can't just keep throwing stuff to the developer and say, "Oh, that's the developer's problem. That's the developer's job.

" No, the developers, they want to develop. They want to code, they want to check code, what have you. That's part of what the platform is for, and that's why it needs to be built in there.

Pankaj, I want to wrap this particular episode up with this, and I touched on it already, but how does AI infrastructure change the FinOps challenge? I think today, if you look at the AI infrastructure, what is the most expensive resource you have? GPUs.

Mm-hmm. What is the most unpredictable behavior you have? Agents.

Who is going to use more platform, just 30 million developers or hundreds of millions of agents? And tokenomics is brand new field for enforcing the financial accountability and FinOps. Absolutely.

Pankaj, thank you for joining us here on our continuing series. This is just, of course, as I said, episode four. We've still got a couple more episodes to go, I think two or three at least, and I'm looking forward to it.

But until next time, thank you for joining me here on Techstrong TV with this. Thank you for watching. org.

com site. And Pankaj, for the Broadcom, VMware side of things, where should people go to stay up on the latest? com website.

We have a very detailed inside blog. 0 is and why it is evolution. I love it.

Until next time, this is Alan Shimel. Thanks very much.