$150K is the realistic upper bound of standard AWS credit programs without entering competitive cohorts (Generative AI Accelerator) or specialized pathways (MAP for migrations). It's built from three stacked pools: Activate Portfolio ($100K) + Build for Startups ($25K) + Bedrock POC ($25K). This page walks through how each layer composes, why the stacking math works (and where it stops working), and the exact application pattern partners use to land this stack in 18–21 days.
There are theoretically larger credit tiers — the Generative AI Accelerator funds up to $1M, MAP can fund $200K+, EDP gives committed-spend discounts. But those are different mechanics with different eligibility windows. $150K is the largest pool achievable through "standard" Activate-family applications, which is what most Series-A and seed startups actually navigate.
The structure of the $150K stack reflects AWS's credit-program design philosophy: separate pools for separate purposes, each capped to prevent any single startup from absorbing too much credit budget. The pools don't conflict with each other because they're scoped to non-overlapping use cases. AWS reviewers explicitly check for use-case overlap before approving stacked applications.
The $100K Portfolio pool funds your general AWS infrastructure — the workloads that would exist whether or not you had a specific new project. The $25K Build for Startups pool funds a discrete project, usually a new product line, feature, or vertical. The $25K Bedrock POC pool funds AI inference for a defined use case with an evaluation plan. Each pool answers a different "what is this for?" question.
Pushing beyond $150K requires moving to a different program family: Generative AI Accelerator (competitive cohorts, $200K–$1M), MAP (substantial migrations, $200K+), Solution Provider Program (committed-spend discounts), or Enterprise Discount Program (multi-year contracts). These programs have qualitatively different application processes — competitive selection, partner certifications, or sales-team negotiation. They're available, but they require different time horizons (60–180 days vs 18–21 for the standard $150K stack).
For a Series-A or late-seed startup whose primary objective is "get the most credits I can in the next 30 days without entering a competitive cohort," $150K is the answer.
AWS reviewers don't approve "$150K of credits" as a single decision. They approve three separate applications, each evaluated against its own criteria. Understanding what each application contains is what makes the stack achievable.
What it funds: general AWS infrastructure. Compute (ECS, EKS, Lambda, EC2), database (RDS, Aurora, DynamoDB), storage (S3, EFS), networking (VPC, CloudFront, Route 53), observability (CloudWatch, X-Ray). The default-purpose pool.
Application form (partner-attested): the ACE record describes your company's overall AWS use case — what your product is, what your stack looks like, what you'll consume across AWS services. Projected annual AWS spend at $4K–$15K/month is the typical band; AWS calibrates the $100K award to fund 18–24 months of that consumption.
What gets it rejected/downgraded: projecting implausibly large AWS spend (e.g., $50K/month at Series-A); use case overlapping with AWS's own products; missing institutional vouch (the partner attestation is the substitute when the VC isn't in Portfolio Sub-Program).
What it funds: a discrete, scoped project that's distinct from your "general AWS infrastructure" use case. Examples that approve cleanly:
• "We're adding a video transcoding pipeline using MediaConvert + S3 + CloudFront for a new product feature."
• "We're building a B2B compliance portal using Cognito + API Gateway + Step Functions — a distinct vertical from our consumer SaaS."
• "We're launching a data-pipeline product using Glue + Athena + S3 for a new customer segment."
What it does NOT fund: the same workload as Portfolio (that's double-counting and gets silently rejected). "Generic infrastructure" framing (Portfolio covers that). Marketplace SaaS purchases (separate billing channel).
What gets it approved: a clear paragraph describing the discrete project, the AWS services it uses, the projected spend ($1K–$2K/month is typical), and a 6–12 month implementation window. The application takes 10 additional minutes on top of the Portfolio application.
What it funds: Amazon Bedrock inference + the supporting compute/storage for a defined AI POC. Earmarked — these credits can only be spent on Bedrock and tightly-coupled supporting services (OpenSearch for vector search, Lambda for orchestration, S3 for prompt logs).
The POC plan: a structured proposal AWS reviewers need to see explicitly. Components: the use case (one paragraph, concrete), the chosen model with reasoning ("Claude Sonnet 4 because it balances cost and quality at our scope"), the eval methodology ("we'll measure accuracy on N=500 examples per week against a held-out test set"), the projected inference budget ($1K–$2K/month at POC scale), and the POC window (60 days).
What lands $50K instead of $25K: larger POC scope ($5K+/month projected Bedrock spend), demonstrated commercial outcome trajectory (e.g., a Letter of Intent from a customer who'll pay for the AI feature), explicit commitment to scale Bedrock usage post-POC. Most Series-A applicants land at $25K; ambitious AI-native startups land at $50K.
Stacking is approvable when applications describe distinct use cases. When they don't, AWS reviewers detect the overlap and silently downgrade — typically by approving Portfolio at full ceiling and zeroing out Build for Startups.
The detection mechanic isn't magic. ACE records include itemized "AWS services to be used" fields and a use-case-description paragraph. When two ACE records list the same services and similar descriptions, the reviewer flags overlap. The default action is to approve the larger (Portfolio) and reject the smaller (Build for Startups).
Founders sometimes interpret a Build for Startups rejection as "AWS denied my credits." That's usually wrong — Portfolio likely approved. But the founder didn't notice because the rejection email for Build for Startups arrives separately from the approval email for Portfolio. CloudRoute partners explicitly track both submissions and confirm both outcomes before declaring the engagement complete.
The fix for overlap: write the Build for Startups application around a use case that genuinely is distinct from Portfolio. The vetted partner walks through this with you on the discovery call — "is there a real second workload here, or is this one workload framed two ways?" If one workload, file Portfolio only. Don't pad applications.
A startup that hit $150K via the stack and wants more has four further programs to consider. None are "Activate sub-tiers" — they're separate program families with different mechanics.
A competitive cohort program (~50 startups per cohort globally) for AI-first startups. Median award: $200K–$400K. Top awards: $1M for outstanding AI-native companies.
Application: cohort window (usually quarterly), competitive selection, 60–90 days from application to credit availability.
Eligibility: AI-first product (not just "uses AI features"), pre-Series-B, commitment to Bedrock as inference backbone.
When this fits: AI-first product, willing to wait 60+ days, comfortable with mentorship + go-to-market commitments. When this doesn't fit: need credits in 21 days; AI is feature-level not product-level; bootstrapped without traction signal.
For companies migrating to AWS from another cloud or on-prem. Credits scale across "Assess → Mobilize → Migrate" phases. 25% of migration cost at Mobilize; 50% at Migrate.
For a typical Series-A migration ($500K–$1M total migration cost), MAP funds $125K–$500K. For Series-B+ migrations, MAP routinely covers $500K+.
When this fits: substantial migration with $5K+/month projected post-migration AWS spend. When this doesn't fit: greenfield AWS build (no migration); migration too small for partner-led MAP work.
A partner-resold AWS billing arrangement. The partner becomes your billing intermediary; AWS gives them a wholesale discount; they pass most of that discount to you.
Typical net discount: 8–12% on AWS bills. Compounds over multi-year contracts. For a $200K/year AWS spend, that's $16K–$24K/year saved.
When this fits: profitable bootstrapped at $5M+ ARR; predictable AWS growth; willing to consolidate billing through a partner. When this doesn't fit: pre-revenue; want flexibility to switch providers; AWS bill too small ($24K/year saved isn't meaningful if AWS spend is $20K/year).
Direct from AWS (no partner intermediary). You commit to a minimum annual AWS spend; AWS gives tiered discounts beyond a threshold.
Typical discount: 10–25% for mid-market customers ($100K–$1M/year AWS spend); 25–40% for enterprise ($1M+/year).
When this fits: mid-market or enterprise scale; multi-year revenue predictability; AWS account team relationship already established. When this doesn't fit: startup-scale AWS spend under $100K/year — EDP minimum thresholds don't apply.
| Path | Ceiling | Time-to-credits | Founder hours | Approval profile |
|---|---|---|---|---|
| $150K stack (Portfolio + Build + Bedrock POC) | $150K | 18–21 days | ~70 min | Standard partner-filed; ~80% approval at full ceiling |
| Generative AI Accelerator | up to $1M | 60–90 days | cohort application + interview | Competitive; ~5% acceptance rate for typical applicants |
| Migration Acceleration Program | $200K–$500K+ | 14–28 days (Assess) + ongoing across phases | ~2 hours initially + ongoing partner engagement | Tied to substantial migration; partner-led |
| Solution Provider Program | 8–12% bill discount (compounds) | 30–60 days to onboard | contract negotiation | Open to profitable bootstrapped at scale |
| Enterprise Discount Program | 10–25% bill discount | 60–120 days to negotiate | sales-cycle engagement | Open at $100K+/yr AWS spend |
Day 0 — Submit a CloudRoute inquiry. Indicate that you want the full stack ($150K). 3 minutes.
Day 1 — Routed within 24 hours to an Advanced or Premier-tier partner with both Portfolio + Bedrock POC track records. Calendly link arrives.
Day 2 — 45-minute discovery call. Two halves: (1) confirm Portfolio + Build for Startups use cases are distinct (~20 min); (2) confirm Bedrock POC use case + eval plan + budget (~25 min).
Day 3–4 — You provide: company info, AWS account ID, deck, two distinct use case paragraphs (one for Portfolio/Build, one for Bedrock POC), and a one-paragraph eval methodology. Time: ~30 minutes.
Day 5 — Partner files three ACE records: Portfolio (Layer 1), Build for Startups (Layer 2), Bedrock POC (Layer 3). All within the same business day.
Day 8–14 — AWS reviewer assigns. Portfolio typically approves first (Day 10–12); Build for Startups second (Day 12–14); Bedrock POC last (Day 14–18) because the POC review queue is slightly slower.
Day 16–21 — All three credit balances land in your AWS account. Portfolio and Build for Startups show as general "promotional credits." Bedrock POC shows tagged for Bedrock use with a 60-day POC checkpoint.
Day 21+ onward — Credits auto-apply against your monthly invoice in order of expiration. At typical Series-A burn rates, $150K lasts approximately 24 months. Unused Bedrock POC credits expire at month 12 if the POC didn't complete; Portfolio and Build for Startups expire at month 24.
The Bedrock POC review queue runs 4–7 days slower than the general partner-filed queue because the Bedrock team specifically reviews POC plans. Founders pursuing only Portfolio + Build for Startups can hit 14 days; the Bedrock POC layer adds the extra week. CloudRoute partners file all three on the same day to minimize total wall-clock.
Credit pools are easier to understand when they're translated into months of actual AWS consumption. Here's the distribution from CloudRoute's routed-engagement data, averaging across Series-A SaaS + Series-A AI startups.
| AWS service category | Monthly avg | 24-month total | % of stack |
|---|---|---|---|
| Compute (ECS Fargate / Lambda / Aurora compute) | $2,500 | $60,000 | 40% |
| Database (Aurora / RDS / DynamoDB) | $1,250 | $30,000 | 20% |
| Bedrock inference (POC pool, earmarked) | $1,050 | $25,000 (Bedrock POC only) | 17% (earmarked) |
| Storage + CDN (S3 + CloudFront) | $420 | $10,000 | 7% |
| Networking (VPC + NAT Gateway + DT) | $330 | $8,000 | 5% |
| Observability (CloudWatch + X-Ray) | $250 | $6,000 | 4% |
| Build for Startups discrete project | $460 | $11,000 ($25K Build pool) | 7% |
| Total | $6,260 | $150,000 | 100% |
Founders sometimes ask "is it worth the extra paperwork for $50K more?" Here's the honest comparison.
| Variable | $100K (Portfolio only) | $150K (full stack) |
|---|---|---|
| Credit pools | 1 (Portfolio) | 3 (Portfolio + Build + Bedrock POC) |
| Founder time | ~30 minutes | ~70 minutes |
| Time-to-balance | 11–18 days | 18–21 days |
| Approval risk | ~8% | ~20% (each layer has its own approval risk) |
| Use case requirement | 1 use case (general infra) | 3 distinct use cases |
| Best for | Series-A with single coherent product | Series-A with AI workload + distinct project |
| What the marginal $50K covers | N/A | ~8 months additional AWS runway OR a Bedrock POC at production scale |
| Stack-friendliness with future credits | High (room to grow) | Already near the ceiling |
Situation: Migrating off GCP (Vertex AI + GKE) to AWS Frankfurt for GDPR-compliant data residency. Existing $4K/month GCP spend, projected to scale to $8K/month on AWS within 6 months. AI-first product running RAG pipelines on Vertex AI. a16z portfolio company (Portfolio-eligible). Wanted to maximize the credit stack to fund both the migration AND the Bedrock POC for moving inference off Vertex AI.
What CloudRoute did: Routed within 19 hours to an EU-Central Advanced-tier partner with Bedrock + GCP-migration competencies. Partner filed three ACE records on day 4: Activate Portfolio ($100K for general infrastructure post-migration), Build for Startups ($25K for the migration project — distinct workload), Bedrock POC ($25K with a Claude Opus + Llama 3 multi-model RAG eval plan).
Outcome: All three approved within 19 days. Total credits applied: $150K. GCP→AWS Frankfurt migration completed week 8. Vertex AI→Bedrock migration completed week 11. eu-central-1 data residency confirmed for GDPR. Total cost to customer: $0; CloudRoute commission paid by partner from AWS engagement funding.
engagement window: 13 weeks · founder time: ~11 hours · credits secured: $150K · cost to customer: $0
CloudRoute routes you to a partner who files all three ACE records — Portfolio + Build for Startups + Bedrock POC — within the same week.