How Much Are Free Startup Credits in Brazil Worth?
A company incorporated in Brazil can realistically assemble somewhere between $10,000 and $200,000 in credits across cloud, models and tooling, because nearly every major program is global rather than gated to a US entity.
Brazilian founders tend to get this wrong in two opposite directions. One group assumes the large programs are American and never applies at all. The other sets up a Delaware or Cayman holding company early, partly believing it is what unlocks the credit stack. It is not. An offshore structure earns its legal and accounting cost through fundraising from US funds and contracting with US customers, not through whether Google Cloud will approve a Ltda in Belo Horizonte.
AI Perks tracks $7.7M in credits across 194 companies, and the large majority accept a Brazilian entity with a CNPJ.
Which Providers Give Brazilian Startups Credits
The stack has three commercial tiers - cloud platforms worth tens to hundreds of thousands of dollars, model providers from the low thousands up to around $25,000, and tooling in the hundreds to low thousands per product - plus a Brazilian public funding layer that pays in reais rather than vouchers.
| Provider | Layer | Typical headline value | Open to a Brazilian entity |
|---|---|---|---|
| Google for Startups Cloud Program | Cloud | Up to $200,000 for AI-first teams | Yes |
| Microsoft for Startups Founders Hub | Cloud | Up to $150,000 in Azure, tiered | Yes |
| AWS Activate | Cloud | $1,000 to $100,000 depending on route | Yes |
| Anthropic | Models | Up to $25,000 via partner routes | Yes |
| OpenAI | Models | Low thousands, usually via accelerator or investor routes | Yes |
| NVIDIA Inception | GPU and compute | No fixed figure, discounts plus partner cloud credits | Yes |
| Vercel, Supabase, Linear and similar | Tooling | Hundreds to low thousands each | Yes |
| FINEP and BNDES innovation finance | Public funding | Reais, as grant or subsidised credit | Brazil only |
| FAPESP PIPE | State research funding | Allocated research funding, not a dollar balance | São Paulo state |
Treat every figure there as a headline maximum, not an expectation. The top of each range is rarely the default outcome, and the conditions attached to it move often enough to be worth checking against current terms on getaiperks.com.
Why São Paulo Region Pricing Changes the Math
Brazilian cloud regions list above their North American equivalents, often by a wide margin, so an identical architecture burns through a credit balance faster in São Paulo than it would in Virginia.
AWS sa-east-1, Google Cloud southamerica-east1 and Azure Brazil South all carry that premium, driven by import duties, power and the cost of operating capacity locally. Check current list prices rather than trusting any single percentage, because the gap moves, though its direction has been consistent for years.
Two consequences follow. A $100,000 grant buys a Brazilian team meaningfully less real capacity than the same grant buys a US team, so headline figures overstate your runway. And the obvious workaround, running everything in a cheap US region, costs you well over 100 milliseconds of round trip latency to Brazilian users, which is fine for a batch pipeline and bad for an interactive product.
There is a compliance edge too. LGPD does not itself mandate local storage, but public sector procurement and regulated buyers frequently do, contractually. Credits make it affordable to sit in a Brazilian region during the period when you can least afford it.
What Credits Are Actually Paying For
Credits do not make infrastructure free. They move the moment you start paying, and the shape of the bill is different on each side of that moment.
| Line item | How it behaves for a Brazilian startup |
|---|---|
| Cloud fixed cost | Databases, load balancers, logging and staging run whether or not you have users |
| Model inference | Almost perfectly linear in usage, with no economies of scale until committed spend |
| Data egress | Charged per gigabyte leaving the region, usually covered by credits and invisible until they end |
| FX exposure | Every bill is in dollars, so your real cost in reais moves with the exchange rate |
| Transaction taxes | Brazil applies IOF to foreign currency operations, and the rate has been revised repeatedly |
That IOF line is the one to verify rather than assume, since the rate has been revised several times. Confirm the current figure with your accountant.
The FX point is underrated. A dollar-denominated balance is a hedge while it lasts, so a devaluation quietly raises both the value of your credits and the pain of the day they run out.
The Brazil-Only Layer Most Founders Miss
Beyond the global programs sits a domestic layer of public innovation finance, R&D tax incentives and corporate accelerators whose real value is often the partner tier they unlock.
Public finance pays cash, not vouchers. FINEP and BNDES fund innovation in reais through grants and subsidised credit lines. That is more flexible than a cloud voucher and far heavier on paperwork. FAPESP PIPE does something similar at São Paulo state level for research-driven companies.
Lei do Bem is the incentive most startups cannot use. Brazil's R&D tax incentive applies to companies taxed under the lucro real regime. Most early startups sit on Simples Nacional or lucro presumido, which puts the benefit out of reach until later.
Marco Legal das Startups changed the wrapper, not the funding. The 2021 startup legal framework created lighter structures such as Inova Simples and clarified investment instruments. It makes operating easier. It does not hand you compute.
Corporate accelerators are routers. Cubo Itaú, Wayra and the Google for Startups presence in São Paulo give modest direct value. Their leverage is that membership often moves you into a higher tier at the cloud and model providers, which is where six-figure numbers live. Current conditions per program are listed on AI Perks.
Where the Value Actually Concentrates
Cloud is the heaviest layer, model credits are the most perishable, tooling is small per product and meaningful in aggregate, and the domestic public layer moves on a calendar of its own.
Cloud carries the weight. It is the largest single line in the stack and the decision everything else sits on. Committing to model credits locked to one vendor before the cloud question is settled is how teams end up paying two bills.
Model credits are perishable. A balance you cannot consume while it is live is worth a fraction of face value, so it is worth most to a team with something already in production or close to it.
Tooling compounds quietly. Analytics, error tracking, CRM and design tools are small individually and meaningful together. The easiest value in the stack, and the most often ignored.
Domestic funding runs in parallel. Public grant timelines follow their own calendar and do not conflict with anything on the provider side.
Conditions move constantly, which is exactly why a tracked list beats a bookmark folder. Filter by category at getaiperks.com.
What Brazilian Founders Get Wrong About Credits
The expensive mistake is not missing a program. It is treating credits as revenue and building an architecture that only works while someone else pays for it.
Architecting for the credit, not the bill. Teams provision generously in an expensive region while credits absorb it, then meet their real run rate the day the balance hits zero. Check your uncredited monthly cost in reais at least quarterly, even while you are not paying it.
Applying to one program. Approval is inconsistent and criteria are opaque. Three approvals out of nine applications beats one out of one, and the applications mostly ask for the same information.
Flipping offshore for credits. If reincorporation is being justified by program access, check whether the programs actually require it. Most do not. AI Perks lists which have a genuine entity requirement.
Ignoring Pix in the cost model. Payment processing is one of the few line items where Brazil is structurally cheaper than the US, so payment credits matter less here than cloud and model credits do. Spend your application effort where the bill actually is.
Frequently Asked Questions
Can Brazilian startups get the same credits as US startups?
In almost all cases, yes. Cloud, model and tooling programs run globally and accept companies with a CNPJ. A small number of US accelerator routes are closed to foreign entities, and a separate domestic layer of grants is open only to Brazilian companies. Current status per program is tracked at getaiperks.com.
How much are free startup credits in Brazil worth in total?
A realistic assembled stack runs from roughly $10,000 for a pre-product team to well over $150,000 for a funded, AI-heavy company with investor introductions. AI Perks tracks $7.7M in credits across 194 companies, and no single startup qualifies for all of it. Regional pricing means those dollars buy less capacity locally.
Do I need a Delaware or Cayman company to get cloud credits?
No. The major cloud and model programs accept Brazilian entities directly. Flipping offshore is a fundraising and contracting decision with real legal and accounting cost attached, and doing it purely to unlock credits is close to always the wrong trade.
Should I run in the São Paulo region or a cheaper US region?
It depends on who your users are. Brazilian regions list higher than US ones, so credits stretch further in Virginia, but you take well over 100 milliseconds of additional round trip latency. Interactive consumer products should stay local. Batch pipelines and training jobs usually should not.
Do Brazilian public grants replace provider credits?
No, they cover different things. FINEP, BNDES and FAPESP award cash in reais with significant reporting obligations. Provider credits are larger in headline terms and locked to one vendor. Most well funded Brazilian startups end up holding both instruments at once.
Which credits are worth the most to a Brazilian founder?
Cloud, by a wide margin, because it is the largest line in the stack and everything else sits on top of it. Model credits come next in headline value but are the most perishable. Tooling is the smallest per product and the easiest to underrate, since the value is cumulative rather than headline. Category filters are on getaiperks.com.
Build from São Paulo. Let someone in Seattle cover the compute.