S12 Bonus: Sharad Kumar & Harshit Omar, FluidCloud

Episode Summary

Executive Summary: Sharad Kumar and Harshit Omar, co-founders of Fluid Cloud, describe building a platform that lets enterprises migrate infrastructure across cloud providers with a click, driven by their own painful experience with long, expensive post-acquisition migrations. They explain why AI-assisted Terraform generation failed, how they built a deterministic mapping engine instead, and why product discipline, scalability, and distribution matter more than hype.

Main Topics: Origin of Fluid Cloud from real migration pain (Priority: 5/5): The founders’ prior acquisition forced an 8-9 month AWS-to-AWS migration, exposing how broken and costly cloud portability is. That experience became the direct motivation for Fluid Cloud. Why AI was not the right first solution (Priority: 5/5): They initially tried AI-generated Terraform and model-based automation for four months, but it could not produce 100% accurate migrations without humans. They pivoted to building their own mapping engine in-house. Building for multi-cloud portability at scale (Priority: 5/5): Fluid Cloud is designed to unify cloud APIs and enable click-button migration across AWS, Azure, and GCP, targeting a Fortune 2000-scale infrastructure problem rather than a small MVP use case. Product philosophy and MVP discipline (Priority: 4/5): The team rejected shortcuts, built production-like MVP foundations early, and emphasized testing, automated deployments, and performance benchmarking before broad release. Team culture and hiring for startup mentality (Priority: 4/5): They prioritize problem solvers, fearless builders, and people who believe in the vision, with a strong emphasis on ambition, resilience, and culture fit over pure years of experience. Market timing, VMware disruption, and cloud volatility (Priority: 5/5): They see Broadcom’s VMware changes, Terraform’s licensing shift, cloud outages, and broader infrastructure instability as validating the need for cloud portability and multi-cloud escape paths. Founders’ advice on entrepreneurship and distribution (Priority: 4/5): They stress reality checks, perseverance, avoiding trend-chasing, and building distribution alongside product—because even strong products need a clear go-to-market strategy.

Key Arguments: The founders experienced the problem firsthand when a company acquisition led to an 8-9 month cloud migration, proving current cloud transfer processes are too slow and consultancy-heavy. AI was not capable of producing a fully accurate Terraform-based migration without human intervention, so it could not power a true one-click migration MVP. Cloud infrastructure portability requires a custom mapping engine because cloud APIs differ and the configuration space is enormous, making generic automation insufficient. The product had to be built for scale from day one because the target problem affects large enterprises and multi-cloud environments, not just early adopters. A strong startup should prioritize problem solvers and people aligned with the mission, not just resumes or experience level. The VMware/Broadcom and Terraform/HashiCorp disruptions created urgency and market demand for alternatives, strengthening Fluid Cloud’s timing. Great companies require both a great product and a distribution strategy; content or product alone is not enough. Founders should not chase hype cycles like AI-only branding if it does not solve a real customer problem.

Data Points: Migration time after acquisition: 8 to 9 months - Their previous company’s move from one AWS account to another exposed the pain that inspired Fluid Cloud. Initial AI/MVP experiment time: 4 months - They spent four months trying AI-based Terraform generation before pivoting away from it. Approximate AWS configuration combinations: 3 to 4 billion combinations - Used to illustrate why building a universal mapping engine is extremely hard. Estimated total cross-cloud combinations: 9 to 10 billion combinations - AWS multiplied by similar complexity in Azure and GCP, underscoring the scale challenge. Time to achieve click-button migrate breakthrough: First week of December - They say the in-house mapping breakthrough enabled zero-human-intervention migration at that point. Unit test coverage at start: 10% - They mention starting with low test coverage and improving the system as they built. Performance benchmark size: 35 million records - Their MVP was stress-tested with no downtime at this scale. Team size: 35 people - Current company size mentioned during discussion of scaling. Engineering team size: 26 people - Subset of the team focused on engineering. Sprint cadence: 2 weeks - They ship features on a two-week release cycle. Revenue donation to foundation: 2% - They say 2% of revenues go to the Fluid Cloud Foundation. VMware price increase: ~1000% - Approximate figure cited as a catalyst for market disruption and customer churn. Expected VMware customer exits: 70% to 75% - They cite Gartner-related expectations that many VMware customers will leave over 2-3 years. VMware customer reconsideration rate: 74% - A poll cited to show many users are rethinking their relationship with VMware.

Pivotal Quotes: "There was always a human intervention needed." — Harshit Omar: Explaining why AI could not fully automate Terraform generation for migration. "The product has to be built on day one for scaling." — Sharad Kumar: Describing why Fluid Cloud was architected for enterprise-scale multi-cloud use from the start. "You can build a great company not being an asshole." — Sharad Kumar: Summarizing the company’s culture and leadership philosophy.

Implications: Fluid Cloud is betting that real enterprise pain, not AI hype, will drive the next infrastructure shift. The episode suggests multi-cloud portability, vendor escape, and disciplined execution will matter more as cloud pricing and outages increase.