Little Green Tags
This week we take a look at the past, present, and future of FinOps strategies for A.I. workloads.
This week’s musical inspiration in title and lyrics:
Getting Informed
Perhaps you’ve been shocked by a public cloud bill. Perhaps you’ve heard from peers that were shocked by a public cloud bill.
Perhaps you are being asked to do more with A.I. in a public cloud. Perhaps you are wondering if your cloud operating model or your cloud adoption framework tagging and metadata strategies of the past will apply today or in the future.
Now it’s time for reading 📖, listening 🎧, watching 📺 suggestions:
-
First, 📖 Managing Cloud Cost Anomalies in which Angel Alves, Keith Knowles, Sounthar Manickavasagam, Victoria Levy, Yogendra Joshi, Usha Ganesh, Mala Vengatesan, Ira Cohen, and Sivaprakash Durairaj make the case for a more financially informed approach to cloud consumption with an eye towards outliers.
-
Second, 🎧 Financial Challenges of M.L./A.I. Ops in which Joe Daly abandons ChatGPT intros before he interviews Thiago Gil on the junction of costs and implications across financial as well as societal concerns for compute selections with respect to the newest generation of workloads.
-
Third, 📺 FinOps Governance & Policy for Onboarding of Workloads in which Beth Marki and Andrew Feig share the wisdom of balancing simplicity in communication with the complexity of what the business expects in cloud fiscal responsibility.
Lookin’ back on the track for a little greenback 🎶
Granted, you might have heard of FinOps or maybe have your FinOps certification. You might even recall all dozen references to FinOps in Fudge Sunday over the past couple years.
Well, this week’s issue will make a baker’s dozen[1] references to FinOps. Specifically, A.I. will require a mix of both conventional wisdom and unconventional wisdom[1:1] when approached from a FinOps perspective.
What does the platform team need to consider for landing zones meant for A.I. workloads?
Lookin’ back on the track, gonna do it my way 🎶
Is there a FinOps for A.I. required? Will there be newer players besides AWS, Azure, GCP, and OCI that come into FinOps calculations and consideration?
Perhaps there was an earlier answer back in late 2021.
CoreWeave, which offers a specialized cloud service for GPU-based workloads, raises $50M led by Magnetar Capital
By Matt Przybyla / VentureBeat. View the full context on Techmeme.
Techmeme
Perhaps that answer was confirmed again last week.
CoreWeave, which offers Nvidia GPUs in the cloud, raised $2.3B in debt, collateralized by Nvidia chips; CoreWeave has raised $421M in equity so far in 2023
/cloudfront-us-east-2.images.arcpublishing.com/reuters/PJ2O4DP37NJL3I737M4ITFX6RI.jpg)
By Krystal Hu / Reuters. View the full context on Techmeme.
Techmeme
In fact, there was a case made last month for avoiding hyperscale public cloud service providers entirely — and embracing the 10th Other “R” of cloud… i.e. Repatriation.
https://www.techmeme.com/220803/p20#a220803p20
So, what will be the next big thing in FinOps strategies for A.I. workloads?
Until then… Place your bets!
Work Plug
As a reminder, after a +25 year walkabout, I’m an IBMer (again). For 2023, in “Work Plug”, I share a new link each week that is educational, accessible, and relevant to platform engineering from fellow IBMers[2] in the wider IBM Community.
Stay tuned!
Disclosure
I am linking to my disclosure.
You should subscribe to Matt Baker’s Unconventional Wisdom newsletter if you haven’t already – good stuff! ↩︎ ↩︎
Shout out to Rosalind Radcliffe ↩︎