Enterprise Cloud Pricing Is a Tax on Being Small – This Removes It

01 /The Concept

Cloud infrastructure pricing is built for enterprises. Startups and small companies spending $1K–$100K per month on AWS are paying full retail while Amazon’s volume discounts quietly sit out of reach. Pump changes the math through group purchasing: it aggregates companies with relatively modest cloud usage, buys resources in bulk at the discounts that scale warrants, and distributes those resources back to members at wholesale pricing. The result: each company gets its small slice of cloud infrastructure at up to 60% off standard rates. The service costs users nothing directly – Pump doesn't charge a fee. Instead it retains a portion of the discount it earns from Amazon, so customers pay slightly above Pump's cost but still below what they'd pay dealing with Amazon alone.

On top of the group purchasing model, Pump runs proprietary optimization algorithms that continuously monitor each company's actual resource usage and automatically cut idle or underused capacity – maintaining or improving performance while driving additional savings.

Before connecting, a company can grant the Pump service read-only access to its cloud account. The system analyzes the account and produces a savings estimate. If the company likes what it sees, it grants modification access and waits up to 24 hours for onboarding approval.

Once approved, the company routes its cloud payments through Pump rather than directly to Amazon. Pump's systems handle billing – and in the background, the optimization layer keeps working.

Pump has also launched a new product: PumpGPT. As the name suggests, it's a purpose-built AI assistant for answering questions and solving problems related to Amazon cloud services.

The goal is to give Pump's users the kind of technical support they'd otherwise have to seek from Amazon directly. That might sound like a feature, but the economics make it meaningful: Amazon's priority support for startups runs around 3% of the monthly cloud bill. For standard business accounts, it's closer to 10%.

PumpGPT gives users a path to eliminating that cost. And by all accounts, it's genuinely capable – it passes Amazon certification exams, answering 91% of questions correctly in each attempt with fresh test sets. The startup claims this outperforms standard GPT-4 on the same benchmarks by roughly 90%

One design decision worth noting: PumpGPT is specifically trained to recommend the most cost-efficient solutions to user problems. Saving money is the whole point of Pump, and the AI is built to reflect that.

Pump went through Y Combinator in spring 2022. It now has around 200 customers, collectively saving roughly $12M per year.

02 /Why It Matters

As [covered previously](/review/otlichnaja-mehanika-dlja-novyh-marketplejsov), Ply – a marketplace for small businesses to purchase equipment and materials for interior work – appears to operate on a similar group purchasing model, promising savings of up to 30%. Ply raised $5.7M.

Group purchasing as a concept isn't new – Groupon practically invented its internet-era incarnation and famously described its own model as "doing for pizza what we're now doing for cloud infrastructure" But historically, group purchasing worked best for physical goods, where vendors gained real logistics and inventory benefits from large orders – justifying the discount.

Cloud providers don't have those problems. No warehouses, no shipping, no inventory to turn. At first glance, it's not obvious why Amazon would want to deal with an intermediary at all.

But Pump's founder says Amazon is actively supportive. In part because Amazon is already comfortable with intermediaries – large consulting firms have long recommended and resold Amazon cloud services to enterprise clients. With Pump, Amazon found itself happy to hand off the cost and hassle of onboarding, billing, and supporting hundreds or thousands of small accounts to a partner capable of aggregating them. That's cost savings for Amazon too.

The trend this points to: a new layer of intermediaries is emerging in cloud services distribution – a channel that was previously sold almost exclusively direct. Other startups in the same wave:

- SaaShop (covered previously) – a marketplace for selling cloud services to small businesses. Raised $3.5M.

- Vendr (covered previously) – a service helping larger companies negotiate better terms on cloud service contracts and renewals. Raised $216M (much of it raised after the original review).

03 /Opportunities

Reselling cloud services is both a compelling business model and a genuinely large market.

Global cloud spending reached nearly $500B in 2022. It grew to close to $600B in 2023. By 2024, it hit $725B.

Individual company spend keeps rising, meaning the percentage savings from optimization are translating into amounts companies actually care about. And the number of companies using cloud infrastructure keeps expanding – Gartner estimates that by 2026, 75% of all organizations will have undergone digital transformation, moving most of their platforms to the cloud.

The kicker: 99.9% of companies in any market are small companies. Cloud providers are spending increasing resources servicing small accounts – and as the Pump-Amazon relationship illustrates, they're willing to hand that cost and complexity to intermediaries who can aggregate even a few hundred of those accounts.

For anyone building or selling cloud services, intermediaries are a chronically underused lever for reaching and servicing small customers – broader coverage, lower operational cost. And for anyone not already in cloud infrastructure, becoming that intermediary – offering customers savings and additional value without owning the underlying infrastructure – is a legitimate business model in its own right.

The most defensible reseller position combines the group discount with a complementary service layer – cost optimization tooling, technical support, or billing automation. The discount gets the customer in; the service layer keeps them.