AWS is Too Expensive: Here is the Open Source Alternative

Umur Cubukcu, co-founder of Ubicloud, explains the principles of an "open cloud," centered on an open-source control plane, portability, and freedom from data lock-in. He details Ubicloud's strategy to compete with hyperscalers by offering superior price-performance on core services like PostgreSQL and compute, particularly for startups and enterprises seeking control and data sovereignty.

4 min read
AWS is Too Expensive: Here is the Open Source Alternative
source: https://www.youtube.com/watch?v=fRbnTywzvmA

The current cloud landscape, dominated by hyperscalers like AWS, Azure, and GCP, has evolved into a set of closed ecosystems. While they run a vast amount of open-source software, the cloud platforms themselves—specifically their control planes—are proprietary. An “open cloud” challenges this paradigm by ensuring that the core management and orchestration software is open source, providing essential transparency and portability.

The Philosophy of an Open Cloud

The minimum requirement for an “open cloud” is an open-source control plane. This is the software that takes bare metal machines as input and produces cloud services (VMs, databases, networking) as output. The key principles that stem from this foundation are:

  • Transparency: With an open-source control plane, there are no hidden mechanisms. Users can see exactly how their data and resources are being managed, which is crucial for security and trust.
  • Portability: An open-source stack can be deployed anywhere—on-premise data centers, rented bare metal, or even on top of a hyperscaler. This gives users a consistent experience and operational model across different environments, breaking the lock-in that defines the current cloud era. The control plane governs the resources, while the “data plane” (where the data resides) can be on a user’s own hardware or a provider of their choice.

Tackling Data Gravity and Storage Lock-In

A primary driver of cloud lock-in is “data gravity,” centered around proprietary storage services like Amazon S3. The combination of massive data volume, service-specific APIs, and punitive egress fees makes migration difficult and expensive. An open cloud manifesto for storage would include:

  • Portable Formats and APIs: The storage stack itself should be portable, allowing users to run it in their own data center if desired.
  • No Punitive Egress Fees: While moving data incurs real network costs, these should be passed on minimally. Hyperscalers often charge 90-100x the actual cost, creating an economic barrier to leaving.
  • Decoupled Compute and Storage: A recommended modern practice is to use storage services with free or low-cost egress, such as Cloudflare’s R2. This allows data to be stored affordably and moved freely to any compute provider—including Ubicloud—for processing, after which the results can be written back. This pattern is not yet mainstream due to the path dependency of existing data on services like S3.

Ubicloud’s Value Proposition

Ubicloud was founded on the thesis that portability and superior economics could be combined to offer a compelling alternative to the status quo. The value proposition differs by customer segment:

  • For Startups: The primary appeal is price-performance. By building on bare metal and using a minimal, modern stack, Ubicloud can offer services at roughly one-third the cost of hyperscalers. Performance is often better because they use newer hardware, such as the latest AMD EPYC CPUs and local NVMe disks, whereas hyperscalers often rely on older infrastructure.
  • For Enterprises: The key driver is control and sovereignty. Many large organizations face a false choice between legacy on-premise solutions (e.g., VMware) and handing over control to a hyperscaler. Ubicloud provides a middle ground, allowing them to run a modern cloud stack on their own hardware or in specific geographic regions to meet data residency and privacy requirements.

Core Services and AI Workloads

Ubicloud focuses on a curated set of core services rather than attempting to replicate the entire AWS catalog, positioning itself as the “Trader Joe’s of the cloud.”

  • Current Offerings: The main services include high-performance VMs, managed Kubernetes, managed PostgreSQL, and managed GitHub Actions runners. The GitHub runners are a popular entry point, offering a 10x cost reduction and faster builds with a single-line configuration change.
  • PostgreSQL Expertise: The founding team has deep, long-standing expertise in PostgreSQL. They offer a highly optimized managed PostgreSQL service that can run on Ubicloud’s bare metal or directly on AWS. When run on AWS, it leverages local NVMe disks on EC2 instances to deliver significantly higher performance than native managed services like RDS.
  • AI and ML Workloads: Ubicloud is not a GPU-focused cloud but supports AI workloads by focusing on the broader stack.
    • It’s well-suited for CPU-heavy data preparation, agentic workloads, and inference on open-weight models.
    • It fully supports the emerging PARK stack (PyTorch, open-weight AI models, Ray, and Kubernetes), providing the necessary compute and orchestration layers.
    • For GPU-intensive tasks, users can leverage Ubicloud’s integration with bare-metal providers or connect to specialized GPU clouds, using Ubicloud for the surrounding components.

Ecosystem and Vision

The long-term vision aligns with the “Sky Computing” concept, where cloud infrastructure is an abstracted commodity and providers compete on value, not lock-in. By aggregating bare metal from various providers, Ubicloud can offer a unified, high-performance, low-cost public cloud experience.

The open-source nature of the control plane also enables a powerful ecosystem strategy. For instance, Ubicloud has partnered with ClickHouse, where the ClickHouse team runs an instance of the Ubicloud control plane to offer a best-in-class managed PostgreSQL service to their own customers. This co-development model, where partners contribute improvements back to the upstream project, benefits the entire ecosystem.

Share post

Stay In The Loop! Subscribe to Our Newsletter.

Get updates straight to your inbox. No spam, just useful content.

Recommended for You

Pipecat Cloud: Enterprise Voice Agents Built On Open Source - Kwindla Hultman Kramer, Daily

Pipecat Cloud: Enterprise Voice Agents Built On Open Source - Kwindla Hultman Kramer, Daily

A deep dive into the challenges of building production-grade, low-latency voice AI agents, and how the open-source, vendor-neutral framework Pipecat provides a comprehensive solution for development, deployment, and scaling. Learn about voice AI architecture, the trade-offs between speech-to-speech and text-based models, and practical deployment strategies.

Inference at Scale:Breaking the Memory Wall

Inference at Scale:Breaking the Memory Wall

Sid Sheth, CEO of d-matrix, details their memory-centric approach to AI inference hardware, focusing on their Digital In-Memory Compute (DIMC) architecture. He explains how DIMC, an augmented SRAM technology, minimizes data movement to solve the memory bottleneck, delivering significant gains in latency and energy efficiency, particularly for the 'decode' phase of large language models.