Akamas

Akamas is an autonomous optimization platform for Kubernetes and cloud-native applications, headquartered in Milan and founded in 2019 as a spinoff from Moviri, a longtime performance-engineering firm.

Reviewed by 7wData

On this page

Publisher review

Akamas is an autonomous optimization platform for Kubernetes and cloud-native applications, headquartered in Milan and founded in 2019 as a spinoff from Moviri, a longtime performance-engineering firm. The platform uses patented reinforcement learning AI to continuously optimize full-stack application and infrastructure configurations—spanning serving engines (vLLM, batching), GPUs (NVIDIA, DCGM), runtimes (CUDA, PyTorch, JVM, Node.js), and orchestration (Kubernetes)—to balance cost, performance, reliability, and service quality against user-defined constraints. Rather than surfacing static recommendations, Akamas identifies configuration inefficiencies and either guides teams toward safe changes or applies them autonomously in production with zero downtime, backed by safety policies that test changes incrementally and respect SLOs.

It operates in three modes: Insights (full-stack Kubernetes optimization with observability), Offline (automated application performance testing in pre-production), and Platform (production-grade autonomous optimization). Recent releases include AI Infrastructure Optimization capabilities targeting GPU workload efficiency (published results show +57.5% decode throughput, +54.1% prefill throughput, -15.8% GPU memory reduction). The company raised €9.6 million (approximately $10M USD) in February 2025 led by United Ventures, marking its first institutional funding round; it plans to expand into North America with a Boston office.

Customer testimonials from Dynatrace, Navan (VP of Engineering: "rapidly matured on performance tuning" with "significant cost savings"), and TeamSystem highlight the platform's ability to improve microservice efficiency and reduce operational toil at scale. Akamas competes with observability and cost-optimization platforms but differentiates through full-stack awareness and reinforcement learning rather than rules engines.

Get the AI & data signal, daily.

48k+ subscribers read this every morning. One email, both newsletters. Unsubscribe anytime.

How it works

  1. Full-Stack Application-Aware Optimization

    Continuously optimizes serving engines, GPUs, runtimes, and infrastructure together while respecting application constraints, SLOs, and service quality requirements.

  2. Patented Reinforcement Learning AI

    Uses proprietary ML algorithms to explore large configuration spaces efficiently and identify optimal settings without manual tuning or trial-and-error.

  3. Production-Grade Autonomous Optimization

    Applies configuration changes in live Kubernetes environments with zero downtime, backed by safety policies that validate incremental changes and prevent harmful modifications.

  4. Insights Module for Kubernetes & Microservices

    Unifies developers, SREs, and platform engineers with observability-driven recommendations for pod sizing, resource allocation, and configuration across distributed systems.

  5. Offline Performance Testing & Tuning

    Automates application performance testing in test environments, systematically identifying optimal configurations before production deployment.

  6. GPU & AI Infrastructure Optimization

    Specialized capabilities for optimizing AI workloads on GPU-based infrastructure, addressing token throughput, memory efficiency, and cost per inference.

  7. Multi-Technology Integration

    Supports Kubernetes, JVM, Node.js, .NET, Golang, Spark big-data jobs, and public cloud instances (AWS, Azure, GCP) with environment-specific optimization profiles.

Strengths and trade-offs

Strengths

  • Autonomous optimization reduces operational toil; handles full-stack tuning (app + runtime + infrastructure) in a single system rather than siloed tools.
  • Reinforcement learning explores high-dimensional configuration spaces faster than rules engines; customers report 60% cloud cost reduction and 30% performance gains.
  • Safety-first approach with incremental rollout and SLO-aware guardrails prevents configuration errors from degrading availability or service quality.

Trade-offs

  • Pricing not publicly available; enterprise-only sales model requires direct outreach, making cost comparison and ROI estimation difficult for prospective buyers.
  • Requires observability instrumentation (Prometheus, Datadog, Dynatrace) to function; doesn't operate independently, adding operational dependency and upfront integration cost.
  • Nascent market (founded 2019, first institutional funding Feb 2025); limited long-term track record at scale; customer base concentrated in Europe and early US adoption.

Pricing context

Akamas operates on a subscription model with three product tiers—Insights, Offline, and Platform—tailored to different deployment maturity levels. Specific pricing is not publicly disclosed; the company uses custom enterprise quotations. Recent $10M Series Seed funding (Feb 2025) indicates venture-backed growth trajectory with plans to expand beyond Europe.

Usage is typically based on infrastructure scale (nodes, pods, workload count) and optimization frequency (real-time vs. periodic). No freemium or free tier is offered.

Alternatives

User reviews

No user reviews yet. Be the first to write one.

Sources

Reporting on this tool draws on these publicly available sources.

  1. akamas.io — Product overview, core features, Kubernetes and AI optimization capabilities
  2. akamas.io — Company founding (evolution from Moviri), Milan headquarters, mission statement, recognition as GigaOm leader in cloud resource optimization
  3. akamas.io — Full-stack optimization, reinforcement learning AI, supported technologies (Kubernetes, JVM, Node.js), Insights/Offline/Platform modules
  4. www.featuredcustomers.com — Customer testimonials and case studies from Dynatrace, Navan, and TeamSystem; 4.8/5.0 rating; documented cost savings and performance improvements
  5. akamas.io — Funding round (€9.6M / $10M USD Feb 2025), US expansion plans, investor backing
  6. unitedventures.com — Investment rationale, market positioning, autonomous optimization approach