Sacra estimates that Pump hit $25M in annualized revenue in 2025, up from roughly $7M at the end of 2024 and $500K as of March 2024, reflecting rapid expansion in spend under management as the company scaled its pooled billing model.

Pump’s revenue is driven by its position as a cloud reseller and intermediary, earning a percentage of aggregated volume discounts negotiated with AWS and GCP rather than charging customers directly. As more companies onboard and route billing through Pump, total cloud spend under management increases, unlocking deeper pricing tiers and expanding the margin pool available to Pump.

Growth is tightly coupled to customer acquisition and average cloud spend per account. The company reports hundreds to over 1,000 customers and a pooled purchasing base that enables 20–40% typical savings, suggesting strong alignment between customer value and revenue expansion as usage scales.

Looking forward, revenue growth depends on expanding beyond AWS into GCP and Azure, increasing penetration into higher-spend mid-market customers, and layering additional monetization through visibility, security, and AI-driven optimization products.

Pump has raised a total of $1.72M in disclosed funding. Its most recent round was a $215K seed closed in June 2024, with Leonis Investment and Y Combinator as key investors.

Pump is a Y Combinator S22 company.

Pump is a cloud cost optimization platform that sits between a company and its AWS, GCP, or Azure bill. Its core function is to change how cloud workloads are purchased, not how they are built or run, by automating purchases of discount instruments such as Reserved Instances, Savings Plans, and Committed Use Discounts on behalf of customers.

Onboarding is lightweight: a company connects its cloud account by granting read-only billing access, and Pump pulls up to a year of usage history to generate a savings estimate. No infrastructure changes are required. If the customer grants autopilot permissions, Pump begins executing commitment purchases automatically, buying and rotating discount instruments based on actual usage patterns instead of requiring the customer to forecast demand.

The product's main claim is risk transfer. Startups often underuse cloud discount instruments because a one- or three-year commitment can create overcommitment risk if usage changes. Pump pools that risk across its customer base. Because many companies have partially offsetting usage patterns and buy through the same purchasing layer, Pump can make more aggressive commitment purchases for an individual customer than that customer could typically make on its own.

The platform has three live modules. Pump Save is the savings engine. Pump View is a multi-cloud spend visibility layer that normalizes cost data across AWS, GCP, and Azure into a single dashboard with anomaly detection, forecasting, custom reports, and Slack or email alerts. Pump Secure adds a security posture layer that scans hundreds of AWS configuration issues and provides step-by-step remediation guidance based on Well-Architected practices.

A fourth module, PumpGPT, is described as an AI assistant trained on AWS documentation and the customer's own environment. It can answer infrastructure questions, recommend cost-optimized configurations, and help file support tickets, but remains more of a product direction than a fully commercialized pillar.

Pump is a B2B cloud reseller and optimization engine. The product is free to end customers: Pump earns revenue from cloud providers through reseller economics and the margin created when aggregated customer spend qualifies for pricing tiers that no individual startup could reach on its own. After a customer joins, its cloud usage is billed through Pump, and Pump invoices the customer directly each month.

The model scales with aggregate spend. More customers increase spend under management, which can improve volume discount economics, raise savings per customer, and make the free product easier to sell and refer. Pump's site references a $150M aggregated purchasing pool, and the company cites average customer savings of 30%, with typical outcomes ranging from 20% to 40%.

The business also warehouses risk. Pump buys commitments on behalf of customers and offers a money-back guarantee for overcommitment attributable to its own decisions, so it is underwriting forecast risk across a diversified portfolio of companies. That diversification reduces the risk of commitment purchases for each customer and can improve the accuracy of Pump's optimization engine over time.

Gross margins are likely lower than those of pure SaaS FinOps vendors like CloudZero or Vantage, because a meaningful share of economics flows through resale and payment operations. The tradeoff is that removing the software fee can collapse a procurement objection and accelerate adoption, especially in the startup and SMB segment where budget approval for a new SaaS tool can be a meaningful friction point.

The FinOps market is fragmenting into three layers: visibility, commitment automation, and workload optimization. Pump competes across all three with a free, reseller-funded model that most rivals do not match.

ProsperOps is the closest direct threat in automation, offering autonomous discount management across AWS, Azure, and GCP and monetizing as a percentage of realized savings. Its main advantage versus Pump is that it automates commitments without requiring the billing-intermediation model Pump uses, which can be easier to clear in enterprise procurement reviews where finance teams are cautious about changing the commercial relationship with a hyperscaler.

Zesty and nOps compete in similar territory, with Zesty concentrated in AWS Savings Plans automation and nOps leaning into AWS-native depth and Kubernetes tie-ins. Flexera One Cloud Commitment Management targets the same problem for large enterprises that already use Flexera for broader IT asset management, where commitment optimization is bundled into a wider technology-spend program rather than bought as a standalone tool.

CloudZero, Vantage, and Finout compete primarily on spend intelligence. CloudZero, which Sacra estimates at $42M ARR in March 2026, is oriented around unit economics, mapping cloud cost back to specific products, features, or customers so engineering and finance teams can make architectural decisions, not just purchasing decisions. Vantage competes on breadth and clean UX, supporting AWS, Azure, GCP, and many adjacent providers with fixed-price plans and lightweight autopilot for Savings Plans. Finout differentiates on allocation depth, with a MegaBill approach that handles shared infrastructure costs across Kubernetes, Snowflake, Datadog, Databricks, and OpenAI alongside the major clouds.

DoiT has the closest structural overlap with Pump's model: it combines FinOps software with a cloud commerce and reseller motion, and its acquisition of CloudWize in late 2025 moved it into security posture and compliance, mirroring the same Save-View-Secure bundle Pump is building. DoiT is more enterprise-oriented and service-heavy, but it competes for the same underlying buying motivation.

CAST AI competes for the same cost-reduction budget through a different mechanism, runtime optimization of Kubernetes and AI workloads via autoscaling, rightsizing, and spot orchestration rather than billing-layer commitment arbitrage. For customers with large containerized workloads, CAST AI can produce larger savings than commitment optimization alone, making it a substitution risk rather than just a feature gap.

AWS Cost Explorer, Azure Advisor, and GCP FinOps Hub continue to improve their native recommendation engines and are free to existing cloud customers. They cannot aggregate spend across companies, but they raise the baseline Pump must beat on single-cloud accounts and reduce the perceived urgency of adopting a third-party tool for customers that are not yet multi-cloud.

Pump's core wedge is cloud commitment optimization for startups and SMBs, but the same billing relationship and spend data that power the savings engine create expansion paths into adjacent products, larger customers, and new categories of cloud spend.

The FinOps Foundation's 2026 data shows that nearly all practitioners now manage AI spend, up sharply from 63% the prior year. GPU training and inference workloads are spikier, harder to forecast, and more expensive per unit than traditional cloud infrastructure, the conditions where Pump's commitment automation and risk-transfer model appear most applicable.

Expanding into AI-specific optimization, including GPU commitment planning, cross-region placement recommendations, and anomaly detection for bursty inference spend, would let Pump target the fastest-growing segment of the cloud bill without requiring a fundamentally different product architecture.

Pump's current messaging skews toward founders and engineering teams at startups, but Pump View's executive reporting, finance-friendly summaries, and anomaly alerts already fit a broader buyer. Moving upmarket into mid-market companies with distributed engineering teams and formal finance oversight is a logical extension, especially as those organizations look for a lighter-weight alternative to enterprise FinOps suites like IBM Apptio Cloudability or Flexera.

The channel motion also broadens distribution. Pump's existing partnerships with YC and Techstars reach companies when cloud credits roll off and paid usage begins. Extending that channel to startup banks, cloud-credit ecosystems, MSPs, and cloud consultancies would pull Pump into accounts earlier and at lower customer acquisition cost than direct outbound.

Pump already operates an AWS Reserved Instance resale capability and, in some onboarding paths, takes on management-account billing ownership. That level of integration points toward a broader cloud purchasing network where Pump intermediates commitment purchases, secondary-market rebalancing, cross-cloud procurement workflows, and potentially brokered capacity for AI infrastructure.

As more spend flows through Pump's billing layer, it accumulates more data on usage patterns across its customer portfolio. That should improve the accuracy of its commitment models and expand the set of optimization actions available through the platform relative to what any individual customer could do alone.

Hyperscaler channel risk: Pump's economic model depends on maintaining authorized reseller status and favorable discount-sharing arrangements with AWS, GCP, and Azure, so a unilateral change to reseller economics, partner program terms, or API access by any of the three cloud providers could compress Pump's margin pool faster than the company could pivot to a software-fee model.

Billing-layer trust barrier: Because Pump becomes the billing intermediary and in some cases takes on management-account ownership, it faces a higher security review burden and more procurement friction than read-only FinOps dashboards like Vantage or CloudZero, which can limit how quickly it moves upmarket into enterprises with formal cloud governance and legal review processes.

Customer graduation risk: Pump's value proposition is strongest for companies too small to negotiate bespoke hyperscaler economics directly, but as customers scale their cloud spend into the millions per month, they may secure equivalent or better direct pricing from AWS, GCP, or Azure and build internal FinOps functions, creating a ceiling on retention unless Pump's software and security layers are sticky enough to justify staying on the platform beyond the savings alone.