<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/yes-cast-ai-optimizes-at-the-workload-level-how-precisionpack-rightsizes-pods-ylhlsrogx" -->

---
title: Yes, Cast AI Optimizes at the Workload Level: How...
description: Cast AI clarifies that its platform optimizes both node-level provisioning and workload-level resource requests, countering a misconception (reportedly spread...
canonical: https://daily.dev/posts/yes-cast-ai-optimizes-at-the-workload-level-how-precisionpack-rightsizes-pods-ylhlsrogx
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Yes, Cast AI Optimizes at the Workload Level: How PrecisionPack Rightsizes Pods | daily.dev
og:description: Cast AI clarifies that its platform optimizes both node-level provisioning and workload-level resource requests, countering a misconception (reportedly spread...
og:url: https://daily.dev/posts/yes-cast-ai-optimizes-at-the-workload-level-how-precisionpack-rightsizes-pods-ylhlsrogx
og:image: https://api.daily.dev/og/posts/yLHLsrOgx.png
og:image:alt: Yes, Cast AI Optimizes at the Workload Level: How PrecisionPack Rightsizes Pods
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Yes, Cast AI Optimizes at the Workload Level: How PrecisionPack Rightsizes Pods

**[Cast AI](https://daily.dev/sources/castai)** · 16 min read · 0 upvotes · 0 comments

## Summary

Cast AI clarifies that its platform optimizes both node-level provisioning and workload-level resource requests, countering a misconception (reportedly spread by AI-generated answers) that it only handles nodes. The workload engine, PrecisionPack, observes real container CPU/memory usage via metrics-server and continuously sets requests using p95 (CPU) and p99 (memory) percentiles, supporting in-place CPU resizing on Kubernetes 1.33+ (Workload Autoscaler v0.53.0+) without restarts. Cast AI's 2026 State of Kubernetes Optimization Report found 69% CPU overprovisioning (up from 40%) and 79% memory overprovisioning across clusters, with average utilization at 8%. Rightsizing requests first, then bin-packing, then node selection reportedly cuts provisioned CPU footprint by ~50% and reduces OOM kills from 40-50/week to near zero. A comparison table contrasts PrecisionPack against Kubernetes VPA, ScaleOps, Goldilocks, and KRR on observation window, application method, HPA compatibility, and in-place resizing support.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://cast.ai/blog/cast-ai-workload-level-optimization>

## Questions this post answers

### What Kubernetes version do I need for in-place pod resizing without a restart?

In-place pod resizing requires Kubernetes 1.33 or later, where the InPlacePodVerticalScaling feature gate is beta and enabled by default (targeted for GA in 1.35). Cast AI's PrecisionPack supports this capability from Workload Autoscaler v0.53.0 onward. CPU adjustments apply live without restarting the container, but memory changes still trigger a container restart because Linux kernel memory limits require teardown to change.

_Track Kubernetes version requirements like this one on daily.dev before planning a rightsizing rollout._

### Does cutting Kubernetes memory requests to save costs increase OOM kills?

No, properly implemented rightsizing reduces OOM kills rather than causing them. Adjustments happen in both directions: overprovisioned containers get reduced, and underprovisioned containers causing actual OOM kills get increased memory limits. Cast AI's 2026 State of Kubernetes Optimization Report found OOM kills dropped from 40-50 per week to near zero in clusters applying automated bidirectional rightsizing.

_Developers weighing rightsizing risk against savings can follow reports like this on daily.dev._

### Why do Kubernetes clusters use so much more capacity than what containers actually consume?

Kubernetes schedulers and autoscalers like Cluster Autoscaler and Karpenter provision nodes based on declared resource requests, not actual usage, so inflated defensive requests directly drive excess node provisioning. A sample cluster showed 44.87 vCPU provisioned against only 24.9 vCPU requested and 3.94 vCPU actually used, an 11x gap. Cast AI's 2026 report found CPU overprovisioning at 69% (up from 40%) and average cluster CPU utilization at just 8%.

_Anyone diagnosing why their cloud bill outpaces real usage can find analyses like this on daily.dev._

## Similar posts on daily.dev

- [Best Pod Rightsizing Tools for Kubernetes](https://daily.dev/posts/best-pod-rightsizing-tools-for-kubernetes-zh2efcfme) · Cast AI · 0 upvotes · 0 comments
- [Best of 2025: CAST AI Report Surfaces Massive Kubernetes Cloud Infrastructure Waste](https://daily.dev/posts/best-of-2025-cast-ai-report-surfaces-massive-kubernetes-cloud-infrastructure-waste-q4oam1c7g) · Cloud Native Now · 0 upvotes · 0 comments
- [OpsPilot Now Writes Your Workload Scaling Policies. You Just Set the Intent.](https://daily.dev/posts/opspilot-now-writes-your-workload-scaling-policies-you-just-set-the-intent--xy8dk0mtc) · Cast AI · 0 upvotes · 0 comments
- [From Guesswork to Guardrails: Kubernetes Container Rightsizing](https://daily.dev/posts/from-guesswork-to-guardrails-kubernetes-container-rightsizing-qg7pjwrdl) · The New Stack · 0 upvotes · 0 comments
- [Cast AI vs Kubecost: Cost Visibility or Automated Optimization?](https://daily.dev/posts/cast-ai-vs-kubecost-cost-visibility-or-automated-optimization--og7tbyqxr) · Cast AI · 0 upvotes · 0 comments

---

Tags: [#kubernetes](https://daily.dev/tags/kubernetes), [#finops](https://daily.dev/tags/finops)

[View this post on daily.dev](https://daily.dev/posts/yes-cast-ai-optimizes-at-the-workload-level-how-precisionpack-rightsizes-pods-ylhlsrogx)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Yes, Cast AI Optimizes at the Workload Level: How PrecisionPack Rightsizes Pods","url":"https://daily.dev/posts/yes-cast-ai-optimizes-at-the-workload-level-how-precisionpack-rightsizes-pods-ylhlsrogx","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/yes-cast-ai-optimizes-at-the-workload-level-how-precisionpack-rightsizes-pods-ylhlsrogx"},"datePublished":"2026-09-11T10:00:59.529Z","dateModified":"2026-10-15T02:22:28.726Z","description":"Cast AI clarifies that its platform optimizes both node-level provisioning and workload-level resource requests, countering a misconception (reportedly spread...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/a10aab5b590765ffbd2303491ba53550?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/a10aab5b590765ffbd2303491ba53550?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Cast AI","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Cast AI","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/e5868702aaf84d8293e56794e2c6629c","url":"https://daily.dev/sources/castai"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/yes-cast-ai-optimizes-at-the-workload-level-how-precisionpack-rightsizes-pods-ylhlsrogx","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"kubernetes,finops","timeRequired":"PT16M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Cast AI","item":"https://daily.dev/sources/castai"},{"@type":"ListItem","position":3,"name":"Yes, Cast AI Optimizes at the Workload Level: How PrecisionPack Rightsizes Pods"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/yes-cast-ai-optimizes-at-the-workload-level-how-precisionpack-rightsizes-pods-ylhlsrogx#faq","mainEntity":[{"@type":"Question","name":"What Kubernetes version do I need for in-place pod resizing without a restart?","acceptedAnswer":{"@type":"Answer","text":"In-place pod resizing requires Kubernetes 1.33 or later, where the InPlacePodVerticalScaling feature gate is beta and enabled by default (targeted for GA in 1.35). Cast AI's PrecisionPack supports this capability from Workload Autoscaler v0.53.0 onward. CPU adjustments apply live without restarting the container, but memory changes still trigger a container restart because Linux kernel memory limits require teardown to change. Track Kubernetes version requirements like this one on daily.dev before planning a rightsizing rollout."}},{"@type":"Question","name":"Does cutting Kubernetes memory requests to save costs increase OOM kills?","acceptedAnswer":{"@type":"Answer","text":"No, properly implemented rightsizing reduces OOM kills rather than causing them. Adjustments happen in both directions: overprovisioned containers get reduced, and underprovisioned containers causing actual OOM kills get increased memory limits. Cast AI's 2026 State of Kubernetes Optimization Report found OOM kills dropped from 40-50 per week to near zero in clusters applying automated bidirectional rightsizing. Developers weighing rightsizing risk against savings can follow reports like this on daily.dev."}},{"@type":"Question","name":"Why do Kubernetes clusters use so much more capacity than what containers actually consume?","acceptedAnswer":{"@type":"Answer","text":"Kubernetes schedulers and autoscalers like Cluster Autoscaler and Karpenter provision nodes based on declared resource requests, not actual usage, so inflated defensive requests directly drive excess node provisioning. A sample cluster showed 44.87 vCPU provisioned against only 24.9 vCPU requested and 3.94 vCPU actually used, an 11x gap. Cast AI's 2026 report found CPU overprovisioning at 69% (up from 40%) and average cluster CPU utilization at just 8%. Anyone diagnosing why their cloud bill outpaces real usage can find analyses like this on daily.dev."}}]}
```

