The AWS credit landscape went through three distinct phases between 2024 and 2026. The Generative AI Accelerator launched. Bedrock POC funding standardized then expanded. AWS Activate consolidated former Founders, Builders, Builders Plus, and Portfolio variations into a cleaner tier structure. Older promotional code programs reached end-of-life. This page is the single reference for what is materially different in 2026 — what evolved, what stayed identical, and what to expect heading into 2027.
The credit landscape did not change in a single moment. It evolved across three distinct phases — each driven by a different competitive pressure on AWS. Knowing the phase a program is in explains why a track exists at all, what funding pool it draws from, and where it is likely going next.
Phase one ran from early 2024 through mid-2024. The Generative AI Accelerator launched as a 20-startup pilot cohort in the first quarter of 2024, driven by competitive pressure from OpenAI's startup program and the broader perception that AWS was behind on AI distribution. Bedrock POC funding existed in 2023 but only in ad hoc form — individual partner-development managers granting Bedrock-earmarked credits case by case. In early 2024, AWS standardized Bedrock POC funding into a real program with documented eligibility criteria, a published credit range, and a defined ACE submission path.
Phase two ran from mid-2024 through 2025. The Generative AI Accelerator scaled from 20 startups in its pilot cohort to roughly 50 per quarterly cohort by the end of 2025, and the geographic acceptance broadened from primarily US-based startups in the 2024 launch to a more balanced mix including APAC, MENA, and LATAM by late 2025. Bedrock POC expanded its model coverage — the original 2024 program was Anthropic-focused (Claude inference workloads); the 2025 expansion added Llama, Mistral, and Nova workloads to the eligible scope. Build for Startups, which existed in narrower form in 2023, had its scope materially expanded during 2024 to cover discrete projects across a wider range of AWS services.
Phase three ran through late 2025 into 2026. AWS Activate consolidated its tier structure — former variations such as AWS Activate Founders, Builders, Builders Plus, and the various sub-tracks within Portfolio were collapsed into a clearer four-tier presentation that is closer to what the self-serve application form shows today. Older promotional code programs (early AWS Builders codes, certain re:Invent codes from pre-2023 events) reached end-of-life through 2025 and are no longer redeemable in 2026. The Activate self-serve form was revised in early 2026 to streamline eligibility verification — fewer free-text fields, more dropdown-driven inputs, and tighter coupling to the AWS account creation flow.
Each phase reflects a different competitive pressure on AWS. Phase one was about AI distribution. Phase two was about geographic and model diversification. Phase three was about operational tidiness — making the Activate program legible to founders who had been confused by the proliferation of tier names accumulated over five years.
A reference-grade view of what happened to each named program in the AWS credit landscape, year by year. Each row reflects what a founder filing today should know about that program's history — both for context on why the program exists in its current form and for predicting what is likely to change next.
2024 state: the public Activate page showed four overlapping tiers — Activate Founders ($1K), Activate Builders ($1K with different eligibility), Activate Builders Plus (intermediate), and various Portfolio sub-tracks. The proliferation was a legacy of programs accumulated since AWS Activate's 2013 launch. Founders routinely confused which tier they were applying to.
2025 evolution: internal partner communications signaled an upcoming consolidation. The dollar amounts at each tier stayed largely the same, but the naming was rationalized — what AWS internally called "Activate" began to refer to a clearer four-tier ladder (self-serve Builders $1K, self-serve Founders $5K, partner-filed Founders $5K–$25K, partner-filed Portfolio $50K–$100K).
2026 state: the consolidation is now reflected on the public Activate page and in the revised early-2026 self-serve form. The four tiers are presented as the canonical structure. Older tier names appear in legacy documentation but no longer route to distinct application paths.
What stayed identical: the dollar ceilings. $1K self-serve Builders, $5K self-serve Founders, and $100K partner-filed Portfolio remain the same numbers founders saw in 2024. The change is presentation and routing, not the credit amounts.
2024 state: the Portfolio Sub-Program — the mechanism that lets specific VCs submit credit applications on behalf of their portfolio companies — included roughly thirty US-based tier-1 firms. The list was opaque and rarely published; founders typically learned whether their VC had access by asking directly.
2025 evolution: the Sub-Program list expanded gradually through 2025. Notable additions included several regional VCs — specific MENA funds gaining access, India-focused early-stage funds joining the program, and a handful of LATAM funds being added. The expansion reflected AWS's broader push into emerging markets where the institutional venture base had matured enough to justify the partnership investment.
2026 state: the Sub-Program now includes a larger cross-section of regional VCs, though the US-tier-1 firms still dominate the list numerically. The opacity remains — AWS does not publish the full list — so the practical advice for a founder unsure whether their VC has access is to ask the VC directly or to route through a partner via ACE in parallel.
The implication for routing: the expansion narrowed the gap between US and non-US founders in how easily they could access the $100K Portfolio tier through their investor. A MENA-based Series-A with a regional VC that joined the Sub-Program in 2025 now has a structurally similar Portfolio access path to a US-based Series-A with a Silicon Valley investor. Partner-filed ACE routing still tends to be faster, but the VC-filed path is more broadly available than it was in 2024.
2024 launch: the standardized Bedrock POC funding program was rolled out in early 2024 with Anthropic models as the primary focus. Claude inference workloads were the dominant use case, the eligibility template assumed an Anthropic-flavored evaluation harness, and the credit range was published as $10K–$50K per approved POC.
2025 expansion: the eligible model scope broadened materially. Llama (Meta) workloads on Bedrock became fully eligible during 2025, as did Mistral and Nova (Amazon's own foundation model family). The credit range remained $10K–$50K but the range of POCs being approved widened — eval harnesses for Llama-based agents and Mistral-based content generation became routine.
2026 state: the program now explicitly includes embeddings workloads (where the POC is built around Bedrock's embedding models for retrieval, semantic search, or vector-DB-augmented generation) and Knowledge Bases workloads (where the POC is built around the Bedrock Knowledge Bases feature for managed retrieval-augmented generation). The 2026 application template asks specifically about whether the POC involves embeddings, fine-tuning, or Knowledge Bases — fields that did not exist in the 2024 template.
The credit range: still $10K–$50K typical, with occasional approvals into the higher band for ambitious POCs. The dollar amount has not inflated; the program scope has broadened.
The competitive context: the AI credit gold rush of 2024–2026 was driven by AWS competing directly with OpenAI's and Anthropic's direct-distribution channels for AI startups. Bedrock POC funding expanded model and use-case coverage to keep parity with what startups could obtain by going directly to model vendors. The Anthropic-AWS partnership deepening through 2025 and into 2026 also pulled more Claude-based workloads into the Bedrock-funded pool specifically.
2024 state: Build for Startups existed in a narrower form before 2024 — it was historically scoped to specific compliance workloads (HIPAA, FedRAMP-adjacent) and a small set of AWS services. During 2024 the program scope was materially expanded to cover discrete projects across a much wider range of AWS services and use cases.
2025 evolution: the broader scope settled into the pattern still in effect — Build for Startups funds a discrete project (a new feature, a new vertical, a new compliance push) up to $25K. The eligibility no longer requires the project to be compliance-flavored; any discrete, well-scoped AWS workload qualifies.
2026 state: stable. The $25K ceiling has not changed since the 2024 expansion. The application mechanic remains partner-filed via ACE. Reviewers typically approve Build for Startups records that are filed alongside a Portfolio application and that describe a workload distinct from the general infrastructure covered by Portfolio.
2024 launch: the Generative AI Accelerator was announced in the first quarter of 2024 with an initial cohort of roughly 20 startups, primarily US-based. The program offered up to $1M in Bedrock-earmarked credits per accepted startup, with median funding around $300K. The application window for the inaugural cohort closed quickly relative to demand.
2025 scaling: AWS expanded the program to roughly 50 startups per quarterly cohort by mid-2025. Geographic acceptance broadened — APAC, MENA, and LATAM startups began appearing in the accepted cohorts at meaningful percentages, compared to the 2024 cohort that was predominantly US-based. The median funding remained around $300K, with outliers approaching the $1M ceiling.
2026 state: the program runs roughly four cohorts per year at approximately 50 startups per cohort globally. The application window is typically 30 days. Selection runs 30–45 days after the window closes. First-tranche credit issuance follows roughly 15 days after selection. Subsequent tranches issue across 6–12 months tied to program milestones — use of Bedrock at scale, completion of mentorship modules, joint go-to-market work with the Bedrock team.
What the expansion reflects: the Accelerator scaled because demand far exceeded the 20-startup pilot, and because AWS's competitive posture against OpenAI's startup program and Anthropic's direct distribution required a larger, more visible Bedrock-funded cohort. The Anthropic-AWS partnership deepening through 2025–2026 has pulled the Accelerator further toward Claude-focused workloads in practice, though officially the program supports any Bedrock-eligible foundation model.
2024 state — MAP 2.0: the Migration Acceleration Program had been on the 2.0 framework since 2022. Three phases (Assess, Mobilize, Migrate), each with credit funding scaled to migration scope. Credit award percentages were standardized: roughly 25% of projected migration cost at the Mobilize phase, roughly 50% at the Migrate phase, with the Assess phase funded as a flat band ($5K–$25K).
2025 evolution: internal partner signals began pointing to a MAP 3.0 transition during 2025. The 3.0 framework increases the Mobilize and Migrate phase coverage percentages for specific verticals — financial services and healthcare in particular were called out for higher coverage because the migration complexity in those verticals justifies it. The transition has been gradual; individual MAP engagements during 2025 were filed under either 2.0 or 3.0 depending on partner and reviewer familiarity.
2026 state: the transition is still in progress as of mid-2026. New MAP engagements filed by partners with current MAP credentials are increasingly being routed through 3.0; legacy engagements continue under 2.0. The vertical-specific coverage increase is the headline change — a healthcare migration filed under MAP 3.0 may see Mobilize coverage at 30–35% of projected cost rather than the 25% baseline, and a financial services migration may see similar uplifts. Non-vertical migrations continue at the baseline 25%/50% pattern.
The practical impact: for most CloudRoute-routed migration engagements, the 3.0 transition has not materially changed the credit math. For migrations specifically in healthcare or financial services, the difference is meaningful and worth confirming with the partner before filing.
The end-of-life landscape: AWS has historically run a variety of promotional code programs — early AWS Builders codes distributed through partner conferences, re:Invent attendee codes, codes from specific accelerator partnerships, and various time-bounded promotional grants. Most of these were small ($100–$1K) and supplementary to the core Activate program.
What deprecated through 2025: early AWS Builders promotional codes from pre-2023 events reached end-of-life through 2025 and are no longer redeemable in 2026. Certain re:Invent attendee codes from pre-2023 events are similarly no longer redeemable. The cutoff was not announced as a single event; codes simply stopped being honored as their internal expiration logic ran out.
What is still live: Activate self-serve Founders ($5K) and Builders ($1K) remain the current self-serve channels. They are not promotional codes — they are first-class Activate tiers that flow through the standard self-serve application form. The deprecation affected promotional codes specifically, not the core Activate program.
For founders holding old codes: if a code is from a pre-2023 event and has not been redeemed, it is likely no longer valid in 2026. The practical workaround is to apply through the current self-serve Activate Founders form at $5K, which clears in 3–7 days.
What changed: AWS adjusted the requirements for the Premier partner tier during 2025. The adjustments affected which partners maintain access to certain ACE submission categories — particularly the higher-ceiling Portfolio tiers and Bedrock POC submissions. The changes were operational rather than structural — the tier ladder (Select, Advanced, Premier) remained intact; the qualification criteria at each rung shifted slightly.
Why it matters for founders: partners who held Premier tier in 2024 may or may not still hold it in 2026 depending on whether their consulting revenue, certified-headcount, and case-study output met the adjusted thresholds during 2025. The implication for founders routing through ACE is that partner standing now varies more across the partner ecosystem than it did in 2024.
The CloudRoute response: partner standing is one of the routing signals CloudRoute uses to match founders. Partners with current Premier tier and recent ACE submission velocity are routed for higher-ceiling engagements; partners at Advanced tier are routed for engagements where Premier-tier credentials are not required. The 2025 rebalancing did not break the routing model; it required CloudRoute to refresh the partner standing data more frequently than once a year.
The most visible 2026 change to the typical founder is the revised Activate self-serve form. The form was rebuilt in early 2026 to streamline eligibility verification and to align the self-serve tier presentation with the consolidated tier structure described above.
The 2024 self-serve form was a free-text-heavy questionnaire. Founders filled in company description, AWS use case, projected spend, and a number of explanatory fields with paragraph-length answers. Eligibility verification ran partly on automated checks (email domain matching, AWS account existence) and partly on a human reviewer reading the free-text fields. The wall-clock from submission to credits-in-account ran 3–7 days for the $5K Founders tier.
The 2026 revised form replaces several of the free-text fields with dropdown-driven inputs. The company stage field is now a dropdown rather than a paragraph. The use case is selected from a curated taxonomy rather than written in prose. The projected spend is structured into specific service categories. The result is a form that takes a similar amount of founder time (perhaps slightly less) but produces a more structured submission that the reviewer can evaluate faster.
The wall-clock did not change meaningfully. The $5K Founders tier still clears in 3–7 days. The $1K Builders self-serve tier still clears in 24 hours. The form is faster to fill in and faster to review, but the queue cadence that drives the wall-clock — reviewer batch assignments, weekend pauses, weekly volume — did not change.
Coupling to the AWS account creation flow also tightened. In 2024, founders could submit a Founders application without an AWS account and be invited to create one as part of the approval process. In 2026, the form expects an AWS account ID at submission. If one does not exist, the form links into the AWS account creation flow before allowing the Founders application to be filed. The practical implication: founders should create the AWS account first, then file the Founders application — the reverse order no longer works.
The combined effect of the Generative AI Accelerator launch, the Bedrock POC funding expansion, and the deepening Anthropic-AWS partnership produced a meaningful surge in AI-targeted credit funding from 2024 through 2026. Understanding the competitive context explains why the AI tracks scaled the way they did and why they are likely to continue scaling into 2027.
The competitive backdrop is the central driver. Through 2023 and into early 2024, OpenAI established a clear lead in API-distributed inference for startups, with Anthropic following closely. AWS's response was twofold — deepen the Bedrock distribution channel (which already offered Claude, Llama, Mistral, and Nova through a unified API surface) and dramatically expand credit funding for startups choosing to build on Bedrock. The Generative AI Accelerator launch in 2024 and the Bedrock POC funding standardization were the visible outputs of that response.
The Anthropic-AWS partnership specifically deepened materially through 2025 and into 2026. AWS's investment posture toward Anthropic and the corresponding commitment to making Claude the flagship Bedrock model produced downstream effects on credit funding — more Claude-focused POCs being approved, more Anthropic-flavored eligibility templates in the application materials, and a Generative AI Accelerator cohort whose accepted workloads skewed toward Claude-based agent and content-generation use cases.
The credit math reflects the competitive economics. AWS's funding cost for a $300K-median Generative AI Accelerator credit grant is recovered when the accepted startup graduates off credits and becomes a paying Bedrock customer at scale — typically 12–24 months post-acceptance. The break-even arithmetic only works when the startup's eventual Bedrock spend is materially larger than the credit grant. For accepted startups operating at $50K+/month of Bedrock inference, the break-even runs comfortably; for accepted startups that fizzle, the credit cost is absorbed.
The volume increase has been notable. Bedrock POC approvals in 2024 ran at a fraction of the 2026 cadence. Generative AI Accelerator applications in 2024 ran into a 20-startup acceptance window; the equivalent applications in 2026 run into a 50-startup window. The absolute dollar volume of AI-targeted credit funding from AWS to startups roughly doubled between 2024 and 2026, even before counting the broader Bedrock POC scope expansion.
The implication for founders applying in 2026: the AI tracks are larger, more diverse in eligible workloads, and more geographically inclusive than they were in 2024. The competitive pressure that produced the scaling has not abated heading into 2027 — if anything, it has intensified as the Anthropic-AWS relationship deepens further and as OpenAI's direct distribution continues to anchor a competing pole.
A reference page on what changed is incomplete without an equally clear statement of what stayed identical. The core mechanics of the AWS credit landscape remain stable from 2024 through 2026. The changes summarized above are scope, scale, presentation, and forms — not the underlying structure.
| Program | 2024 state | 2026 state | What changed materially |
|---|---|---|---|
| Activate Founders (self-serve) | $5K, free-text form, 3–7 days | $5K, dropdown-driven form, 3–7 days | Form revised early 2026; dollar amount unchanged; AWS account ID now required at submission |
| Activate Builders (self-serve) | $1K, 24h wall-clock | $1K, 24h wall-clock | No material change |
| Activate Founders (partner-filed) | $5K–$25K via ACE, 10–14 days | $5K–$25K via ACE, 10–14 days | No material change — stable since 2024 |
| Activate Portfolio (partner-filed) | $100K via ACE, 11–18 days | $100K via ACE, 11–18 days | VC Sub-Program list expanded with regional additions; ceiling unchanged |
| Build for Startups | $25K, scope-expanded mid-2024 | $25K, stable | Scope expansion happened in 2024; stable since |
| Bedrock POC funding | $10K–$50K, Anthropic-focused | $10K–$50K, Llama + Mistral + Nova + embeddings + Knowledge Bases | Model and use-case scope materially broadened 2025–2026 |
| Generative AI Accelerator | $300K median, ~20 startups, US-skewed | $300K median, ~50 per cohort, APAC/MENA/LATAM accepted | Cohort size more than doubled; geographic mix broadened |
| MAP — Assess phase | $5K–$25K, 14–21 days | $5K–$25K, 14–21 days | No material change at the Assess phase |
| MAP — Mobilize phase | ~25% of projected cost | ~25% baseline, 30–35% for FS/healthcare under 3.0 | Vertical-specific uplift under MAP 3.0 transition |
| MAP — Migrate phase | ~50% of projected cost | ~50% baseline, vertical uplift under 3.0 | Vertical-specific uplift under MAP 3.0 transition |
| Older promotional codes | Pre-2023 codes still partially redeemable | Pre-2023 codes deprecated and no longer honored | End-of-life through 2025 for legacy promotional codes |
Founders planning runway calculations for 2027 fundraising cycles ask reasonably whether the credit landscape is about to shift again. The honest answer is that the core mechanics are stable but several specific programs are likely to evolve further. The speculation below is grounded in partner-program signals and the visible competitive context, not in announced AWS roadmaps.
The Generative AI Accelerator cohort size is the most likely candidate for further expansion. The scaling from 20 startups in the 2024 pilot to roughly 50 per quarterly cohort by 2026 was driven by competitive pressure that has not abated. Partner-program signals through late 2025 and into 2026 suggest AWS is comfortable with the operational mechanics of a larger cohort and that the funding pool has room to support additional accepted startups. A cohort size in the 60–75 range per quarter by late 2027 is plausible based on the current trajectory.
The Anthropic-AWS relationship is the second factor likely to drive change. The partnership deepened materially through 2025 and into 2026 — investment posture, Bedrock distribution priority, and joint go-to-market commitments. Downstream implications for credit funding are that Claude-focused workloads may see further preferential treatment within the Bedrock POC program, and the Generative AI Accelerator may continue to skew toward Claude-based use cases in practice even if officially the program supports any Bedrock-eligible foundation model. Founders building on Claude specifically are positioned to benefit from this trajectory.
The Portfolio Sub-Program VC list is likely to continue expanding. The 2025 additions of regional VCs (MENA, India, LATAM) suggest AWS's broader emerging-markets push will continue producing further VC partnership additions during 2027. The implication for founders is that institutional access to the $100K Portfolio tier through their VC will become more broadly available globally — though partner-filed ACE routing will likely remain the faster path in most cases.
MAP 3.0 transition is likely to complete fully during 2027. The vertical-specific coverage uplifts (financial services, healthcare) are likely to be fully consolidated, and additional verticals (potentially manufacturing, telecommunications, retail) may be added to the uplift list based on internal partner-program signals from 2025–2026. The phase architecture (Assess → Mobilize → Migrate) is structurally stable and unlikely to change.
The Activate self-serve form is unlikely to be revised again in 2027. The early-2026 revision was a multi-quarter project and the operational benefits (faster reviewer evaluation, cleaner submission data) are likely to be sufficient for at least 18–24 months. A further revision is more likely in 2028.
What is unlikely to change in 2027: the $100K Portfolio ceiling, the $5K self-serve Founders tier, the partner-filed ACE mechanism, the institutional vouch requirement, and the $0 cost-to-founder structure. These are foundational mechanics rather than tunable parameters, and AWS has shown no indication of materially restructuring them.
For founders planning runway calculations through 2027, anchor on the stable core: $100K Portfolio + $25K Build for Startups + $10K–$50K Bedrock POC remains the realistic standard stack regardless of how cohort sizes or VC lists evolve. The Generative AI Accelerator path remains an outlier — competitive selection rather than a standard application — and should be treated as upside rather than baseline. The partner-filed ACE mechanism is the operationally stable channel; route through there as the primary path with the self-serve $5K Founders as a bridge during the partner-filed wait.
A reference document on the year-over-year evolution is useful only insofar as it translates into operational decisions. The five operational implications below are the actionable distillation of what changed and what it means for a founder filing today.
One — AI tracks are now first-class, not afterthoughts. In 2024, Bedrock POC funding was a relatively new addition that founders often overlooked. In 2026, the Bedrock POC track is a routine layer in a standard stacked engagement. If a founder is filing a Portfolio + Build for Startups stack and has any AI workload — even a small one — the Bedrock POC layer is worth filing as the third record in the same engagement. The marginal effort is small; the marginal credit gain is meaningful.
Two — geographic eligibility has broadened. Series-A startups based in MENA, India, LATAM, or APAC have materially more credit access in 2026 than they did in 2024. The Generative AI Accelerator accepts non-US cohorts at meaningful percentages. The Portfolio Sub-Program VC list includes regional VCs. Non-US founders should not assume that AWS credit programs are US-only — they were predominantly US-flavored in 2024 but are not in 2026.
Three — the revised self-serve form requires an AWS account first. The early-2026 form revision tightened coupling to the AWS account creation flow. The practical sequencing is: create the AWS account, then file the Founders application. The reverse order (apply first, create account on approval) no longer works on the current form. For partner-filed engagements, the partner can guide the AWS account setup as part of the discovery call, but the founder should still expect to create the account before the ACE record is submitted.
Four — partner standing matters more in 2026 than in 2024. The 2025 Premier tier rebalancing produced more variability in partner standing across the partner ecosystem. Routing through a partner with current Premier tier and recent ACE submission velocity is more clearly preferable than it was in 2024 when more partners held stable Premier tier. CloudRoute's routing model accounts for this; founders going directly to partners should verify current partner standing rather than assuming continuity from 2024 perceptions.
Five — the Generative AI Accelerator is on a different track than the standard stack. The Accelerator scaled in cohort size and broadened geographically, but it remains a competitive selection process on a 60–90 day wall-clock — not a credit application on a 14–21 day wall-clock. Founders who need credits in the near term should file the standard partner-filed stack ($100K + $25K + $25K) and consider the Accelerator as a separate parallel application. The two tracks are not mutually exclusive; accepted Accelerator startups can supersede or supplement the standard stack depending on overlap with already-issued Bedrock POC credits.
For a typical Series-A founder filing the standard partner-filed stack, the dollar math has not changed between 2024 and 2026. The structural composition is similar. What is different is the scope of eligible workloads at the Bedrock POC layer and the broader geographic eligibility for the Generative AI Accelerator layer.
| Variable | Standard 2024 stack | Standard 2026 stack | What changed |
|---|---|---|---|
| Portfolio ceiling | $100K | $100K | No change |
| Build for Startups | $25K | $25K | Scope expanded 2024; stable since |
| Bedrock POC | $10K–$50K (Claude-focused) | $10K–$50K (Claude + Llama + Mistral + Nova + embeddings + Knowledge Bases) | Eligible model and workload scope broadened |
| Total standard stack | $135K–$175K | $135K–$175K | Dollar range unchanged |
| Wall-clock | 11–21 days | 11–21 days | No change to standard wall-clock |
| Generative AI Accelerator upside | ~20 startups/cohort, US-skewed | ~50 startups/cohort, geographically broader | Cohort more than doubled; non-US acceptance materially higher |
| MAP Mobilize coverage (FS/healthcare) | ~25% | 30–35% under MAP 3.0 | Vertical-specific uplift for financial services and healthcare |
| Application form (self-serve) | Free-text questionnaire | Dropdown-driven, AWS account ID required | Revised early 2026 |
| Cost to founder | $0 | $0 | Structural economics unchanged |
Situation: Series-A AI-ML startup building a retrieval-augmented agent on Bedrock with Claude as the primary reasoning model and Bedrock embeddings for the retrieval layer. In 2024, the embeddings workload would not have been eligible for Bedrock POC funding under the original Anthropic-focused program scope. By 2026, embeddings and Knowledge Bases workloads are explicitly within scope. The founder wanted to file the full standard stack — Portfolio for general infra, Build for Startups for a vertical-specific compliance push, Bedrock POC for the agent + embeddings work — and confirm the embeddings layer would clear under the revised 2026 eligibility template.
What CloudRoute did: Routed within 20 hours to an EU partner with Premier tier (verified 2026 standing post-rebalancing) and demonstrated Bedrock POC track record including 2025–2026 embeddings approvals. Discovery call confirmed eligibility under the current 2026 Bedrock POC template — the application explicitly flagged the embeddings + Knowledge Bases components as eligible workloads. Partner filed all three ACE records on day 5: Portfolio for general infra, Build for Startups for a GDPR telemetry project, Bedrock POC for the Claude agent + embeddings POC plan.
Outcome: All three records cleared within the standard wall-clock. Portfolio approved day 13. Build for Startups approved day 15. Bedrock POC approved day 17 — the embeddings workload was approved without write-back, confirming the 2026 scope expansion holds in practice. Total credits applied: $145K ($100K + $25K + $20K). Founder reported the timeline accurately matched the Day 18 planning anchor. CloudRoute commission paid by partner from AWS engagement funding; customer paid $0.
engagement window: 17 days · founder time: ~6 hours · credits secured: $145K · scope: included 2026-eligible embeddings workload · cost to customer: $0
CloudRoute routes you to a partner with current Premier tier and recent ACE submission velocity. Partner files the ACE record under the 2026 application template; credits land in 11–18 days for the standard stack. Customer pays $0.