Updated June 2026 · Originally published February 27, 2026
AWS credits for AI projects run from $1,000 (Activate Founders) to $100,000 (Activate Portfolio), $500,000 for Y Combinator companies, and up to $1,000,000 for the Generative AI Accelerator. The win is not the headline number; it is stacking the programs you actually qualify for and getting an Activate Org ID from your VC or your fintech bank first.
- ✓ Start with Activate Founders ($1,000, self-serve), then upgrade to Portfolio (up to $100,000) through a VC, accelerator, or fintech Org ID.
- ✓ Brex and Mercury both issue Activate Org IDs in 1 to 2 business days, so no VC is required for the first $5,000.
- ✓ Since April 2024, credits work on third-party models on Amazon Bedrock (Anthropic Claude, Llama, Mistral).
- ✓ Credits expire (12 to 24 months) and AWS enforced this harder in 2026, so model your steady-state spend before they land.
Most AI founders paying full-price AWS bills are leaving real money on the table. Training on SageMaker, running inference on Amazon Bedrock (Amazon managed model service), storing datasets on S3, and deploying agents at scale runs $4,000 to $18,000 a month before you earn a dollar. AWS credits are not a secret, but the strategy to stack them is. The problem is rarely access; founders knock on the wrong door first and walk away with $1,000 when they qualified for $100,000.
Our team has set up AWS infrastructure for AI builds across the US, UK, UAE, and Singapore, and we tested every one of these paths with real client accounts. At Braincuber Technologies we help AI-first companies get their AWS infrastructure right from day one, and that starts with not paying full price for it. Every figure below is checked against AWS's own pages as of June 2026.
Which AWS credit programs exist in 2026?
Not all programs pay the same or approve at the same speed. Here is what exists, what it pays, and who qualifies, ranked by realistic dollar value.
| Program | Credit value | Who qualifies | Approval speed |
|---|---|---|---|
| AWS Activate Founders | $1,000 | Bootstrapped, pre-Series B, founded within 10 years | 48–72 hours |
| AWS Activate Portfolio | $5,000–$100,000 (up to $200,000 with select providers) | VC / accelerator / fintech-backed with an Org ID | 1–2 weeks |
| Y Combinator + AWS | $500,000 | YC companies (2024 cohorts onward) | On enrollment |
| Generative AI Accelerator | Up to $1,000,000 | GenAI / agentic startups (under 2% accepted) | Annual cohort |
| Cloud Credit for Research | $5,000–$250,000+ | Academic / institutional researchers | 90–120 days |
| New-account free tier | Up to $200 | Any account opened after July 15, 2025 | Instant |
The most-used tier is Activate Portfolio: AWS says more than 280,000 startups have gone through it, and it now carries up to $10,000 in business-support credits on top of the compute. The single biggest mistake we see is applying for Founders ($1,000) when you already qualify for Portfolio, simply because nobody asked the accelerator or VC for the Org ID. We have watched that $99,000 gap happen more than a dozen times in the past year.
How do you stack credits across programs?
AWS does not limit you to one program, so the real play is to run several in parallel. Here is the order we recommend:
- Week 1: Open a Brex or Mercury account, request their AWS Activate Org ID, and apply for Portfolio ($5,000 minimum, often $25,000+). Apply for Activate Founders ($1,000) in parallel if you have not used it.
- Week 2: If you are at an accredited institution, submit a Cloud Credit for Research proposal. Rolling deadline, 90 to 120-day review.
- Month 3: If your investor is an AWS Activate Provider, request their Org ID for the full $100,000 Portfolio credit.
- Mid-year: Apply for the Generative AI Accelerator. Start the application about 3 months before the deadline.
- At signup: Claim the new-account free tier: $100 on sign-up plus up to $100 more for completing five service activities (EC2, RDS, Lambda, Bedrock, Budgets), for up to $200 total.
Proof it works
We have seen teams layer $106,000+ from the Founders, Portfolio, and fintech paths alone, before ever applying to the Accelerator.
Want this mapped to your funding stage and AWS spend? We will tell you exactly which programs to apply for, in order.
Book a free call →What can AWS credits actually be spent on?
Credits are not locked to one service. A credit-funded AI stack can spend them on:
- Amazon SageMaker — model training, tuning, and MLOps pipelines.
- Amazon Bedrock — Anthropic Claude Sonnet 4, Meta Llama, Mistral, Cohere, and AI21 models through one API; third-party models became credit-eligible in April 2024.
- Amazon EC2 GPU instances (P4, P5, G5 families) — raw compute for custom training.
- AWS Trainium and Inferentia — purpose-built AI chips that cost less per inference hour than comparable GPU instances.
- Amazon S3 — dataset storage at about $23 per terabyte per month on S3 Standard.
- AWS Lambda and Step Functions — serverless inference and agentic workflow orchestration.
Why do most founders get rejected or underfunded?
The two application killers
Mistake 1: Applying for Founders ($1,000) when you qualify for Portfolio ($25,000–$100,000) because you never asked your accelerator for the Org ID.
Mistake 2: a vague use case
Do not write "cloud infrastructure." Write "training a fine-tuned Llama 3 model on SageMaker ml.p4d.24xlarge instances, serving inference through Amazon Bedrock at 10,000 daily API calls."
AWS has said plainly that AI and ML startups get priority and enhanced support through 2025 and 2026. Name specific services in your application: SageMaker, Bedrock, Trainium, Inferentia, EC2 GPU instances. Be surgical. Generic applications get generic rejections.
Not sure which AWS credit tier you qualify for?
Book a free call and we will map your stage and spend to the right programs.
How does AWS compare to Azure and Google Cloud?
Compared to the other two clouds, AWS wins on ceiling and on stacking. Microsoft for Startups offers up to $150,000 in Azure credits, but since July 2025 that full tier is tied to its Investor Network; the bootstrapped path caps at $5,000. Google Cloud is the most aggressive for AI specifically, with up to $350,000 over two years ($250,000 in year one, $100,000 in year two). Unlike either, AWS reaches $500,000 for Y Combinator companies and up to $1,000,000 for Generative AI Accelerator picks, and it lets you layer $106,000+ from several programs at once. The 2025 Accelerator cohort took 40 startups worldwide from thousands of applicants, kicked off in Seattle in October, and wrapped at re:Invent in December.
Frequently asked questions
How much can a startup get in AWS credits in 2026?
It ranges from $1,000 (Activate Founders, self-serve) to $100,000 (Activate Portfolio through a VC or accelerator, with select providers reaching $200,000), $500,000 for Y Combinator companies, and up to $1,000,000 for the Generative AI Accelerator cohort. New AWS accounts also get up to $200 in free-tier credits.
Do AWS credits work on Amazon Bedrock for third-party models?
Yes. Since April 2024, AWS Activate credits are redeemable on third-party foundation models on Amazon Bedrock, including Anthropic Claude, Meta Llama, Mistral AI, Cohere, AI21 Labs, and Stability AI, all through a single API. Activate credits cannot be spent on most AWS Marketplace purchases, but Bedrock model usage is the explicit exception.
What is the fastest way to get AWS credits this week?
Open a Brex or Mercury business account. Both are AWS Activate providers and can issue an Org ID within 1 to 2 business days. Apply for Activate Portfolio with that Org ID and $5,000 to $25,000 in credits is typically approved within 1 to 2 weeks, faster than any other route.
Can I apply for multiple AWS credit programs at once?
Yes. AWS does not restrict you to one program. Founders, Portfolio (via a fintech Org ID), and Cloud Credit for Research can run in parallel. Our team has seen startups stack $106,000 or more before ever touching the Generative AI Accelerator.
Do AWS credits expire?
Yes, and AWS enforced expirations more strictly in 2026. Activate Founders credits last about 12 months; Activate Portfolio credits run 12 to 24 months; Research credits expire in roughly 1 year. They cannot be rolled over or exchanged for cash, so plan your deployment timeline before they hit your account.
Do I have to be a US company to qualify?
No. AWS Activate is global. DPIIT-recognized Indian startups can get $5,000 to $10,000 through the Startup India partnership, and more through VC-backed Portfolio applications. The 2025 Generative AI Accelerator drew 40 startups worldwide, including 10 from Asia Pacific and Japan.
Stop paying full-price AWS bills on day one.
Book a 15-minute review. We will map your current AWS spend against available credits and tell you exactly which programs to apply for, in order. No fluff, just numbers.
Sources: AWS Activate credits, AWS Generative AI Accelerator, AWS Free Tier update, Activate credits on Bedrock, The Register on the $500K YC deal, Microsoft for Startups, and Google Cloud for Startups. Figures are current as of June 2026 and change often. Client results reflect Braincuber's own deployments.
Written by Dhwani Tarwani — Co-founder & AI Practice Lead, Braincuber Technologies. Sets up AWS infrastructure (SageMaker, Bedrock, Trainium) and ships production AI agents for US, UK, UAE, and Singapore clients.Founder and CEO of Braincuber. Has scoped and shipped 500+ Odoo, AI, and cloud projects for US mid-market and global brands. Takes every founder call personally — no SDR layer between buyers and the people building the system.