Opinion
Your first cloud deal is not a credits deal
How an early-stage startup should think about AWS, Microsoft, and Google over the long term.
Every founder hears the same advice in the first week: apply for cloud credits. Each of the big three will hand a young company tens or hundreds of thousands of dollars of free infrastructure, and a startup should take it. But choosing a cloud by comparing credit amounts is like choosing a co-founder by comparing signing bonuses. Credits are the most visible part of the relationship and, over five years, usually the least important.
The better question is "who can help me build, sell, and fund the business?" At their best, cloud providers are distribution channels, technical partners, and sometimes investors.
Credits are the opening offer
Credits expire, and a bill that was zero last quarter becomes real overnight. Treat the credit period as runway, not as a cost structure, and model what your infrastructure costs without credits from the first architecture decision. Read the fine print: Google's startup credits, for example, cannot be used on third-party marketplace offerings.
Five layers of value
- Commercial terms: credits, discounts, and later, committed-spend pricing.
- Implementation funding and professional services: every cloud has money to help you migrate, build a new workload, or run a proof of concept, and it can pay for the cloud-certified partners and consultants who do the work. Each vendor calls these programs something different. Ask for them by function.
- Go-to-market: marketplace listing and co-selling with the provider's sales teams.
- Capital: equity arms and accelerators.
- Strategic partnership: introductions to large customers and early access to new products.
The marketplace changes the sales conversation
Large buyers often have committed cloud spend they must consume. If buying your product counts against that commitment, procurement gets easier. The rules differ: Azure benefit-eligible offers count at 100%, Google Cloud Marketplace purchases may draw down a commitment with exclusions, and AWS counts only eligible products. Co-selling usually requires a qualifying marketplace listing, so decide early which marketplace your customers buy through.
Startups often hear that a commitment is a risk to minimize, so they sign flat. If real growth is coming, that can be a mistake. Cloud providers reward new consumption, the incremental spend they can see arriving, with richer benefits: bigger discounts, more funding, more attention from sales. Committing ahead of a growth curve you can defend tends to unlock more than committing to today's usage.
The risk is a shortfall: commitments are typically take-or-pay. So negotiate flexibility: adjustable minimums, rollover, and marketplace purchases that count toward the commit. Commit ambitiously, but only to a plan you believe.
Even with a primary provider, build a relationship with a second one early. A second cloud gives you flexibility, a real comparison on price and terms, and better leverage when the commercial conversation gets serious. Keep the stack portable, and you manage the commercial side of the business on your terms rather than theirs.
A short playbook
- Pre-seed: take the credits from all three, but don't architect around them.
- Choose a primary cloud by where your customers are and which models you need, not by credit size.
- List on the marketplace as soon as there is something to sell.
- When spend grows, negotiate one agreement covering price, credits, implementation funding, co-sell, and support.
Questions to ask your cloud partner
• Which credits expire, and when?
• Can credits cover marketplace purchases?
• What funding is available for migrations, new workloads, and partner services?
• What does co-sell require, and who in your sales organization will know about my product?
• What GPU capacity can you guarantee, for which chips and regions, and how fast are quota increases approved?
• If we sign a commitment, what happens if usage falls short?
In my experience, having advised and negotiated these deals from both the startup side and the vendor side, no two deals are alike. A founder who understands the company's own needs and potential, and who knows the full range of what a cloud provider can offer, can build a deal that is a big win for both sides.
The startups that get the most from a cloud provider treat it as a business partner from the first conversation, and negotiate accordingly.
Elad Nardi is the Chief Revenue Officer at Sela, a global provider of cloud and AI services for technology companies.
