AWS credits for Entrepreneur First (EF) portfolio companies — the post-cohort partner-filed Portfolio path most EF founders never run.

IEntrepreneur First — the London-founded talent investor and its global cohort network

EF is structurally different from accelerators that accept companies. The model shapes both how founders move through cohort and how AWS Activate interacts with the resulting portfolio companies. The first step is understanding the model.

Entrepreneur First was founded in London in 2011 by Matt Clifford and Alice Bentinck. Both founders had a thesis that the bottleneck for deep-tech and ambitious technology company formation in Europe was not capital or ideas but the co-founder matching problem — highly capable individuals were not finding each other early enough to start companies together. EF’s answer was a structured program that selected for individual talent, paid the selected talent a stipend to participate in the program full-time, and ran a six-month cohort across co-founder matching and idea formation.

The early EF cohorts ran in London and were small — single-digit to low-double-digit company outputs per cohort. The model proved itself with the 2015 sale of Magic Pony Technology to Twitter (~$150 million), which was an EF cohort 4 company; subsequent EF portfolio successes included Cleo (the AI-driven personal finance app), Tractable (computer-vision-based vehicle damage assessment), Aztec Network (privacy infrastructure for Ethereum), and PolyAI (conversational AI for enterprise customer service). The portfolio mix skews toward deep tech, AI, biotech, and hardware-software companies — a reflection of the founder pool EF attracts and the explicit EF preference for ambitious technical bets.

EF expanded geographically from London into Singapore (2017), Bangalore (2018), Paris (2019), Berlin (2019), Toronto (2020), and Tokyo (2021). Each location runs cohorts on a roughly six-month cycle, with one or two cohorts per year per city. At any given moment EF is running active cohorts across multiple continents — the London cohort, the Singapore cohort, the Bangalore cohort, the Paris and Berlin cohorts (sometimes run jointly as the European cohort), the Toronto cohort, and the Tokyo cohort all overlap in time at different stages.

The cohort mechanics are the unusual part. EF accepts individual founders, typically through a multi-stage application and interview process that screens for technical depth, startup orientation, and what EF calls "edge" — some combination of domain expertise, technical capability, or insight that suggests the founder could build a defensible company. Accepted founders receive a stipend (the exact figure varies by location and cohort year; the principle is to make full-time participation financially possible for high-quality candidates regardless of their savings position). The cohort itself runs in two distinct phases: Form (co-founder matching and idea exploration) and Launch (focused company-building toward a pre-seed investment decision).

In the Form phase — typically the first two to three months of cohort — founders work in shared space, run structured exercises with each other, prototype ideas in micro-teams, pitch to the EF investment team and to each other, and converge toward two- or three-person founding teams paired with a scoped company idea. In the Launch phase — the back half of cohort — the resulting teams move into focused company-building: customer development, early product work, fundraising preparation. At the end of Launch, EF’s investment committee makes pre-seed funding decisions, and the companies that pass investment committee continue as funded EF portfolio companies. Companies that do not pass investment committee remain EF alumni and continue to receive EF community access and follow-on support, but without the EF check.

The check size at the EF pre-seed stage varies by location. EF London pre-seed checks have historically been in the £80K–£150K range (roughly $100K–$190K at recent FX). EF Singapore pre-seed checks have been in a comparable range denominated in SGD (~SGD 100K–150K). EF Bangalore pre-seed checks denominated in INR have been at the EF-typical range adjusted for local currency. EF Paris and Berlin pre-seed checks in EUR have been similar in absolute scale (€90K–€140K). EF Toronto and Tokyo follow the same pattern with the local currency adjustment. The structural point: EF is a pre-seed talent investor, not a growth-stage investor, and the check size reflects that positioning.

IIEF is inside the AWS Activate Portfolio Sub-Program for post-cohort portfolio companies

AWS Activate maintains a Portfolio Sub-Program with a list of institutional partners whose portfolio companies receive standing access to the Portfolio credit tier. Y Combinator, Techstars, and 500 Global are inside this program with multi-year standing arrangements. EF is inside it as well — specifically for post-cohort EF portfolio companies that have completed Launch and received the EF pre-seed check.

The Portfolio Sub-Program operates as a standing arrangement between AWS Activate and the participating institutional partner. The partner commits to a defined volume of portfolio company introductions to Activate; AWS commits to standing approval thresholds for those portfolio companies; the two organizations integrate at the operational level so that eligibility verification is consistent across applications.

For EF specifically, the standing arrangement covers post-cohort portfolio companies — the subset of cohort participants who completed Launch, passed EF’s investment committee, and received the EF pre-seed check. EF maintains a list of these post-cohort companies, refreshed at the end of each cohort cycle. AWS Activate reviewers can verify EF portfolio company status against this list when reviewing Portfolio applications, which is the operational mechanism behind the standing approval. Partner-filed Portfolio submissions through ACE that reference EF as the institutional sponsor are routed against this verification track.

The implication for credit ceilings: EF portfolio companies see Portfolio approvals at $50K–$100K, with the upper end ($100K) reached when the use case is well-scoped, the projected AWS consumption is meaningful, and the application narrative is tight. This is materially higher than the Antler partner-intermediated ceiling ($50K–$75K) and roughly comparable to the YC and Techstars standing-arrangement ceiling ($75K–$100K). The EF institutional signal is strong enough with AWS Activate reviewers that the standing-arrangement verification carries the application through to the upper Portfolio tier reliably for well-scoped submissions.

The practical caveat: EF cohort participants who do not complete Launch or do not receive the EF investment-committee check are not covered by the standing arrangement. These EF alumni — cohort participants who either did not pass investment committee or who chose to step out of cohort before Launch — still have an EF affiliation, and that affiliation is real institutional signal, but the partner-filed Portfolio path for them runs more like the Antler partner-intermediated path than the YC standing-arrangement path. Approvals for EF alumni without investment-committee backing tend to land in the $50K–$75K range rather than reaching the $100K upper end.

IIIThe pre-team-formation cohort timing and what it means for credit applications

EF’s pre-team-formation model creates a credit-application timing pattern that does not exist for accelerators that accept teams. The pattern is similar to Antler’s in structure but plays out on EF’s specific cohort calendar. Founders who do not understand the timing miss two to three months of the application window.

At EF cohort kickoff — day one of Form — the founder population is unsorted. A typical EF cohort starts with several dozen individual founders, none of whom have an incorporated company at the cohort. Co-founder matching, idea exploration, and team formation happen across Form, which runs roughly two to three months depending on the cohort location and year. By the end of Form, most founders have converged toward a co-founder pair or trio and a scoped idea. The corresponding legal entities — typically incorporated as a UK Ltd for London cohort, a Singapore Pte Ltd for Singapore cohort, a French SAS or a German GmbH for Paris and Berlin cohorts, an Indian Private Limited for Bangalore cohort, an Ontario Inc for Toronto cohort, or a Tokyo-jurisdiction entity for Tokyo cohort — typically get registered during the transition from Form into Launch, somewhere around cohort weeks ten to fourteen.

AWS Activate applications require a registered business entity. The application form asks for company name, registration number, registered address, and tax ID. Until the EF team incorporates, none of those fields can be completed accurately. Partner-filed Portfolio applications through ACE require an even more substantive customer record: registered entity, AWS account ID owned by the entity, named technical contact, scoped use case description, projected AWS consumption.

The implication: the first two to three months of EF cohort cannot access the partner-filed Portfolio path because there is no entity yet. The standing residency code or any self-serve credits can be applied against a personal AWS account during this window, which covers prototyping costs (typically modest — the Form-phase prototypes tend to run on Vercel, Supabase, Render, or a personal AWS account at low monthly burn). The higher-ceiling partner-filed path is gated on incorporation. Effectively the credit-application timeline starts when the team incorporates, not when the founder joins cohort.

The right operational pattern: monitor your incorporation timeline closely. The week your entity is registered (entity name confirmed, registration number issued, tax ID issued, a business bank account opened, an AWS account migrated to or created under the entity), submit a CloudRoute inquiry. The partner-filed Portfolio path can move within five to seven days of incorporation. Approvals land 18 to 25 days later. If incorporation happens in cohort weeks twelve to fourteen, credits are in your account by cohort weeks sixteen to eighteen — late-Launch or just-post-Launch — which is exactly the moment your need for production-class AWS infrastructure starts ramping (customer development moves from interviews to pilots, early product work moves from prototype to production, fundraising preparation references the cloud stack in data-room materials).

A common mistake worth flagging early: deferring the credit application until after EF investment committee, or until the post-cohort seed round closes, or until "we finalise our incorporation structure properly." Each of these defers the application by months. Incorporation in any reasonable jurisdiction is sufficient to start the partner-filed Portfolio path; if you later re-domicile (UK Ltd flip to a Delaware C-Corp for US expansion, for example), the credit balance stays with the original entity but the architectural and operational learning carries over. Waiting for the "perfect" final structure costs you the credit windowing during the moment you most need cloud-cost coverage.

IVThe realistic EF post-cohort credit stack in 2026

The full EF post-cohort stack reaches $105K–$155K: $5K self-serve Activate Founders + $50K–$100K partner-filed Portfolio + $25K Build for Startups + $10K–$50K Bedrock POC. The ceiling matches the YC and Techstars range because EF’s standing arrangement in the Portfolio Sub-Program operates similarly. The components compose cleanly when filed by a partner who understands the EF cohort timing.

VLocation dynamics — London, Singapore, Bangalore, Paris, Berlin, Toronto, Tokyo

EF operates cohorts in seven cities. Each cohort is structurally identical from the AWS Activate perspective — standing-arrangement Portfolio is the path — but the regional AWS partner ecosystem, the regional follow-on VC landscape, and the regional AWS region selection differ. The location of your cohort shapes which partner CloudRoute routes you to.

EF London is the founding cohort and the largest by graduate count. London-based EF portfolio companies typically route to eu-west-2 (London) for production workloads, occasionally with eu-west-1 (Ireland) as the cross-EU region or for documented exit-path purposes. London EF portfolio companies often raise follow-on rounds from the broader UK VC ecosystem — Index Ventures, Atomico, LocalGlobe, Octopus Ventures, Hoxton Ventures, Balderton Capital, and Notion Capital all show up regularly. The London AWS partner ecosystem is dense and includes multiple partners with prior EF cohort track records. CloudRoute routes London EF alumni to UK-fluent partners who have pre-loaded the UK GDPR + Data Protection Act 2018 + ICO compliance language into prior EF application narratives.

EF Singapore anchors the APAC footprint. Singapore-based EF portfolio companies route to ap-southeast-1 (Singapore) for production workloads, occasionally with multi-region setup into ap-southeast-3 (Jakarta) or ap-northeast-1 (Tokyo) depending on customer base. Follow-on VCs include Peak XV (Sequoia’s Southeast Asia successor), East Ventures, Vertex Ventures Southeast Asia, Wavemaker Partners, and Jungle Ventures — the Singapore seed-to-Series-A scene is competitive and well-capitalized. CloudRoute routes Singapore EF alumni to APAC AWS partners with PDPA compliance experience and prior EF Singapore submissions on record. EF Singapore pre-seed checks are typically denominated in SGD (SGD 100K–150K equivalent).

EF Bangalore is the Indian subcontinent cohort. Bangalore-based EF portfolio companies route to ap-south-1 (Mumbai) and increasingly ap-south-2 (Hyderabad) for data localization purposes under the Digital Personal Data Protection Act 2023 (DPDP). Follow-on VCs include Peak XV India, Lightspeed India, Nexus Venture Partners, Blume Ventures, and the broader NASSCOM 10,000 Startups community. CloudRoute routes India EF alumni to partners with DPDP compliance experience and ap-south-1 deployment track records. EF Bangalore pre-seed checks are denominated in INR at the EF-typical range adjusted for local currency.

EF Paris anchors the French cohort. Paris-based EF portfolio companies route to eu-west-3 (Paris) for French data residency or eu-west-1 (Ireland) for cross-EU customer bases. Follow-on VCs include Partech, Elaia Partners, Daphni, Kima Ventures, Serena, Idinvest, and the broader French VC ecosystem (which has expanded substantially through La French Tech and Bpifrance co-investment programs). CloudRoute routes Paris EF alumni to partners with GDPR + French CNIL guidance experience and prior EF Paris track records. Pre-seed checks are typically denominated in EUR (€90K–€140K).

EF Berlin anchors the German cohort. Berlin-based EF portfolio companies route to eu-central-1 (Frankfurt) for German and DACH data residency, occasionally with eu-west-1 (Ireland) for broader EU service availability. Follow-on VCs include Speedinvest, HV Capital, Project A, Cherry Ventures, Earlybird, and the broader DACH VC ecosystem. CloudRoute routes Berlin EF alumni to partners with GDPR + BfDI (the German federal data protection authority) experience and prior EF Berlin or DACH track records. Pre-seed checks denominated in EUR at the EF-typical range. Some cohort years see EF Paris and EF Berlin run jointly as the European cohort — the credit-application mechanics are unchanged but the partner matching draws from both French and German partner pools depending on the company’s primary jurisdiction.

EF Toronto anchors the Canadian cohort. Toronto-based EF portfolio companies route to ca-central-1 (Canada) for Canadian data residency under PIPEDA (Personal Information Protection and Electronic Documents Act) and the Quebec-specific Law 25 framework where relevant. Follow-on VCs include Real Ventures, BDC Capital, Inovia Capital, Golden Ventures, and Garage Capital. CloudRoute routes Toronto EF alumni to partners with PIPEDA experience and ca-central-1 deployment track records. EF Toronto pre-seed checks denominated in CAD at the EF-typical range adjusted for local currency.

EF Tokyo is the most recent cohort addition. Tokyo-based EF portfolio companies route to ap-northeast-1 (Tokyo) for Japanese data residency under the Act on the Protection of Personal Information (APPI). Follow-on collaboration with regional VCs (Globis Capital Partners, JAFCO Group, Coral Capital, ANRI, World Innovation Lab) and Japanese corporate VCs is common. CloudRoute routes Tokyo EF alumni to APAC partners with APPI compliance experience and prior EF Tokyo submissions where available; the partner pool with EF Tokyo track records is smaller than the London or Singapore pools but growing.

VIEF vs Antler vs Seedcamp — the three London-relevant pre-seed paths compared

EF, Antler, and Seedcamp are the three most-cited London-relevant pre-seed investor comparisons. Each has distinct cohort or fund mechanics that shape the AWS credit interaction.

EF and Antler are both pre-team-formation programs. EF accepts individual founders in London (2011-founding), Singapore, Bangalore, Paris, Berlin, Toronto, and Tokyo — seven cities total. Antler accepts individual founders in 30+ cities globally including Singapore (its headquarters), Berlin, Stockholm, Sydney, Nairobi, Bangalore, São Paulo, Riyadh, and others. The shared mechanic: both programs accept founders before teams form, run a structured matching-and-formation phase, then make pre-seed investment decisions in companies that emerge from the cohort. The shared implication for AWS credits: both programs have the incorporation-timing pattern where the entity does not exist at cohort kickoff and the partner-filed Portfolio path opens once the team incorporates mid-cohort.

The structural difference between EF and Antler is concentration and selectivity. EF is more concentrated — fewer cities, fewer cohorts per year, smaller cohort sizes, more selective acceptance into Form, and a higher-stakes investment-committee decision at the end of Launch. Antler is broader — more cities, more cohorts per year, larger cohort sizes, broader acceptance into residency, and a higher absolute volume of post-residency portfolio companies. The credit-mechanic implication: EF post-cohort companies have the standing-arrangement Portfolio Sub-Program access ($50K–$100K ceiling), Antler post-residency companies have the partner-intermediated Portfolio path ($50K–$75K ceiling). The EF concentration plus institutional standing arrangement supports the higher credit ceiling; the Antler breadth plus partner-intermediated path supports the somewhat lower ceiling.

EF and Seedcamp are both London-headquartered seed-stage investors with European reach, but the models are different. EF is pre-team-formation — individual talent enters cohort before the company exists. Seedcamp is post-team-formation — Seedcamp invests in companies that already have a team, an idea, and typically an early product or traction signal. Seedcamp’s mechanic is closer to a traditional seed fund than to a cohort program; portfolio companies are funded individually rather than batched into a cohort. The credit-mechanic implication: Seedcamp portfolio companies do not have the incorporation-timing pattern — they are already incorporated when Seedcamp invests, so the partner-filed Portfolio application can start immediately upon institutional vouch. EF post-cohort companies have to wait through Form and into Launch for incorporation before the partner-filed path opens.

On credit ceilings: EF portfolio companies and Seedcamp portfolio companies both have access to the AWS Activate Portfolio Sub-Program with $50K–$100K ceilings, though through different operational tracks. Seedcamp is recognized in the broader UK VC ecosystem and partner-filed Portfolio applications referencing Seedcamp as the institutional sponsor land cleanly. EF is in the standing arrangement specifically for post-cohort portfolio companies, which provides the verification mechanism. The ceiling outcomes are roughly equivalent ($50K–$100K); the cohort timing dynamic only applies to EF and Antler, not Seedcamp.

Realistic full stacks: EF post-cohort companies land $105K–$155K (this page). Antler post-residency companies land $95K–$130K (the partner-intermediated Portfolio ceiling is the gap). Seedcamp post-investment companies land $105K–$155K (comparable to EF, with the timing advantage of immediate incorporation). The differences are real but modest — a $20K–$30K spread across roughly comparable stacks. The bigger operational variable is the cohort or fund-investment timing relative to when you need the credits to land.

VIIThe EF partner-filed application narrative — what the partner writes into ACE

Every Partner-Filed Activate application is an ACE record. The record has structured fields (company info, use case, AWS services, projected spend) and a free-text narrative section. The narrative is where the EF-specific institutional vouch and the cohort-and-location context go. The structural pattern a CloudRoute-routed EF partner uses follows a consistent shape.

Company-info block. The narrative opens with cohort identification: EF cohort city, cohort year and number, Form-to-Launch transition date, EF investment-committee outcome and date, EF pre-seed check size and date. AWS Activate reviewers verifying against the EF standing-arrangement list look for exactly this information — cohort city and cohort number map to the EF portfolio list; the investment-committee date confirms post-cohort status; the check size grounds the projected AWS consumption.

Use-case paragraph for Portfolio. Production workload described with the AWS service surface. EF deep-tech and AI companies typically describe a service mix of EKS or ECS for the application control plane, RDS Aurora PostgreSQL or DynamoDB for application data, S3 with KMS for object storage, CloudFront for asset delivery, Cognito or a self-managed auth layer for customer authentication, Lambda for asynchronous work, and SQS or EventBridge for inter-service messaging. AI-native EF companies add SageMaker for training, EC2 GPU instances for batch inference or training, and OpenSearch for vector search. The region selection is explicit and tied to the cohort location (eu-west-2 for London, ap-southeast-1 for Singapore, ap-south-1 for Bangalore, eu-west-3 for Paris, eu-central-1 for Berlin, ca-central-1 for Toronto, ap-northeast-1 for Tokyo). Projected 12-month AWS consumption is calibrated to the EF pre-seed check size — typically $3K–$8K/month at month-twelve burn.

Use-case paragraph for Build for Startups. A discrete workload distinct from the Portfolio-funded SaaS infrastructure. Common EF Build scopes: the multi-account AWS Organization restructure for compliance posture (GDPR-aligned for London/Paris/Berlin, PDPA-aligned for Singapore, DPDP-aligned for Bangalore, PIPEDA-aligned for Toronto, APPI-aligned for Tokyo); the customer-managed KMS hierarchy with workload-specific key policies; CloudTrail with multi-year retention in a dedicated log-archive account; GuardDuty + Security Hub + Inspector for continuous monitoring; AWS Backup with documented RPO/RTO posture; documented incident response runbook. Implementation target is typically Q3 or Q4 of the cohort year. Projected consumption for the Build workload runs $1K–$2K/month at steady state.

Use-case paragraph for Bedrock POC (AI-native EF companies). The specific AI feature being built on Bedrock: model selection (Claude Sonnet 4 for the typical reasoning workload, Claude Opus 4 for the high-quality drafting workload, Amazon Titan Embeddings for the RAG layer, sometimes Llama 3.x or Mistral Large 2 for cost-sensitive volume workloads), evaluation methodology (N=200–500 evaluation set with human-reviewed gold-standard outputs, bi-weekly evaluation cadence over the 60- to 90-day POC window), projected inference budget ($2K–$8K/month at POC scale; $5K–$20K/month at production scale post-POC). Region-resident model variant specified where the customer data residency commitments require it.

Compliance addendum (cohort-location-specific). The compliance language varies by cohort city. London EF: UK GDPR + Data Protection Act 2018 + ICO; AWS DPA executed via AWS Artifact. Paris EF: EU GDPR + French CNIL guidance; AWS DPA executed; EU-resident model variant for Bedrock. Berlin EF: EU GDPR + German BfDI guidance; AWS DPA executed; EU-resident or DACH-resident model variant for Bedrock. Singapore EF: PDPA; AWS DPA executed; Singapore-resident model variant for Bedrock where supported. Bangalore EF: DPDP 2023; AWS DPA executed; ap-south-1 deployment for data localization. Toronto EF: PIPEDA (and Quebec Law 25 where relevant); AWS DPA executed. Tokyo EF: APPI; AWS DPA executed. The partner pre-loads the relevant language for the specific cohort city.

VIIIThe cohort-aligned 25-day EF stack timeline

Cohort weeks 1–9 (Form phase). Co-founder matching and idea exploration. No incorporated entity exists yet. Self-serve Activate Founders ($5K) can be filed against your personal AWS account or a placeholder account if useful for prototyping; partner-filed Portfolio is not yet available because there is no entity. Existing prototypes typically run on Vercel, Supabase, Render, or low-burn personal AWS at modest monthly cost.

Cohort weeks 10–14 (Form-to-Launch transition). Team formation completes; the EF team incorporates. Typical jurisdictions by cohort city: UK Ltd (London), Singapore Pte Ltd (Singapore), Indian Private Limited (Bangalore), French SAS (Paris), German GmbH (Berlin), Ontario Inc or Canadian Federal incorporation (Toronto), Japanese KK or GK (Tokyo). A business bank account is opened; an AWS account is created or migrated under the entity name; the tax ID is registered. This is the trigger week for partner-filed Portfolio.

Day 0 (incorporation week). Submit a CloudRoute inquiry indicating EF portfolio company status, cohort city, cohort number, and Form-to-Launch transition date.

Day 1. Routed within 24 hours to a partner with prior EF cohort application experience, matched on cohort region (UK, EU, APAC, India, Canada, Japan) and on vertical (deep tech, AI, biotech, B2B SaaS, hardware-software).

Day 2. Discovery call (30 minutes). Partner confirms EF cohort membership, scopes Portfolio + Build for Startups + Bedrock POC applicability, verifies incorporation status and AWS account setup.

Day 3–5. You provide company info (entity registration, AWS account ID, named technical contact), use case paragraphs, deck, and EF cohort confirmation. Around 45 minutes total founder time.

Day 6–7. Partner files three ACE records (Portfolio + Build for Startups + Bedrock POC if AI-native) within the same business week. Each names EF as the institutional sponsor and your cohort as eligibility evidence; the standing-arrangement verification track applies.

Day 14–20. Portfolio approval lands first ($50K–$100K depending on use case quality and projected consumption). Build for Startups by Day 18–22 ($25K). Bedrock POC by Day 20–25 ($10K–$25K non-AI, $25K–$50K AI-native).

Day 18–25. Full stack visible in AWS Billing dashboard under "promotional credits." $105K–$155K total auto-applies against monthly invoice.

Cohort weeks 18–26 (Launch phase, customer development and fundraising). Production AWS infrastructure operates on the credit balance through Launch and into the post-cohort window. The credit-funded cloud cost coverage supports customer pilots, early product work, and the post-cohort fundraising data room.

Optional post-cohort month 6+ (AI-native EF companies). Apply to the AWS Generative AI Accelerator. Selection in 60–90 days; additional $200K–$500K for accepted AI-native applications. EF-alumni acceptance rates run comparable to YC and Techstars AI cohorts; the EF deep-tech and AI orientation produces strong Accelerator narratives.

IXEF Paris cohort 12, AI-native infrastructure company — anonymized engagement walk-through

A representative anonymized engagement, drawn from a recent CloudRoute EF post-cohort match. Cohort city, cohort number, vertical, and check size are presented faithfully; the company name and specific product detail are obscured.

The team. Three-person founding team — two technical co-founders (ML and distributed systems backgrounds) and one technical-commercial co-founder (prior B2B sales-engineering experience at an enterprise infrastructure company). Met in EF Paris cohort 12 during Form weeks 4–7; agreed to a co-founder pairing in Form week 8 with the third co-founder joining in Launch week 2. Incorporated as a French SAS in cohort week 12; AWS account created under the entity the following week.

The thesis. An infrastructure layer for managing large-scale LLM inference across multiple model providers, with cost optimization and routing logic optimized for European customer data residency requirements. The category positioning is "LLM inference infrastructure for European enterprises" — narrow enough to scope, broad enough to support meaningful AWS consumption projection. EF investment committee approved the pre-seed check (€130K) at the end of Launch; cohort ended cohort week 26.

The credit application. CloudRoute inquiry submitted cohort week 13 (the week after incorporation). Routed within 22 hours to a Paris-based AWS Premier partner with three prior EF Paris cohort approvals and substantive Bedrock and SageMaker deployment experience. Discovery call scheduled cohort week 14 day 1. Partner scoped three ACE records: Portfolio for the application infrastructure, Build for Startups for the GDPR + French CNIL-aligned multi-account compliance build, Bedrock POC for the inference routing and evaluation infrastructure.

The filing narratives. Portfolio narrative described the production workload in eu-west-3 (Paris) and eu-west-1 (Ireland) for failover, with a service mix of EKS, RDS Aurora PostgreSQL, S3 with KMS, CloudFront, Cognito, Lambda, SQS, and OpenSearch for the vector search layer. Projected consumption $5K/month at month six, $9K/month at month twelve. Build narrative described the multi-account AWS Organization restructure with French CNIL-aligned data residency controls, customer-managed KMS hierarchy, CloudTrail multi-year retention in a dedicated log archive account, GuardDuty plus Security Hub plus Inspector for continuous monitoring. Bedrock POC narrative described the inference routing evaluation: Claude Sonnet 4 in the EU-resident variant for the primary reasoning workload, Llama 3.3 70B in eu-west-1 for the cost-sensitive volume workload, evaluation methodology N=400 representative inference queries from three early-customer pilots with human-reviewed gold-standard responses, projected Bedrock spend $4K/month at POC scale and $12K/month at production scale.

The outcomes. Portfolio approval landed at $100K on day 17 (the upper end of the Portfolio range, supported by the EF Paris standing arrangement plus the tight infrastructure-category narrative). Build for Startups approved at $25K on day 20. Bedrock POC approved at $40K on day 23 — above the standard $25K because the infrastructure-layer LLM inference category and the structured evaluation methodology supported the upper end of the Bedrock POC range. Total stack: $165K. Founder time across the credit application: roughly 11 hours including discovery call, use-case paragraph review, AWS account setup verification, and three rounds of narrative refinement with the partner.

The post-engagement runway. Production AWS infrastructure went live cohort week 18, three weeks before Launch ended. The $165K credit balance supported the cloud cost coverage through the eight-month post-cohort window during which the team ran customer pilots with three early European enterprise customers and closed a €1.8M post-cohort seed round led by a Paris-based VC with Berlin-based co-investor participation. The data room referenced the AWS credit balance and the Build-funded compliance posture as part of the technical due-diligence response; the partner-built readiness documentation supported the EU enterprise procurement conversations through the pilot-to-paid transitions.

XWhat EF portfolio founders typically get wrong about AWS credits

• Filing too early (pre-incorporation) — EF founders sometimes ask partners to file Portfolio during Form weeks 4–8 against a "company name to be confirmed" or against the cohort identifier. AWS reviewers reject these — Portfolio applications require a registered entity with verifiable registration. The right pattern is to wait until incorporation completes in cohort weeks 10–14 and start the filing the same week.
• Filing too late (post-Launch month six or later) — Other EF founders defer the credit application past Launch reasoning that "we will sort credits once we close the seed round." The deferral costs months of credit windowing during exactly the period the production AWS infrastructure is ramping. Filing immediately post-incorporation with the EF cohort context produces credits in the account by Launch week 18–20; deferring to seed-round-close pushes the credit availability to cohort month nine or later.
• Skipping Build for Startups because Portfolio "feels like enough" — Portfolio at $50K–$100K is the headline number, but Build for Startups ($25K) and Bedrock POC ($10K–$50K) are additive. EF post-cohort companies that file only Portfolio leave $35K–$75K on the table. The partner files all three in the same business week if the scoping conversation is done up-front during discovery.
• Treating Bedrock POC as a future-quarter concern for AI-native EF companies — Most EF companies are AI-native given the EF deep-tech orientation. The Bedrock POC funding maps to the AI workload that is typically already in scope at incorporation. Deferring the Bedrock POC application until "after we figure out the inference architecture properly" defers $25K–$50K of POC funding past the window when it would have been most useful for the inference architecture decisions.
• Picking the wrong region for the cohort location — Some EF Paris and Berlin companies default to us-east-1 because the engineering team has prior US-deployment experience. For European customer bases with GDPR commitments, this triggers reviewer clarifying questions and complicates the downstream enterprise sales conversations. The default region should match the cohort location: eu-west-2 (London EF), eu-west-3 (Paris EF), eu-central-1 (Berlin EF), ap-southeast-1 (Singapore EF), ap-south-1 (Bangalore EF), ca-central-1 (Toronto EF), ap-northeast-1 (Tokyo EF).
• Filing through a partner with no EF cohort track record — EF portfolio applications carry the standing-arrangement verification track, which the partner has to invoke correctly in the ACE narrative. Partners with prior EF approvals know exactly how to invoke the standing arrangement; partners new to EF take longer per application and sometimes file in a way that bypasses the standing-arrangement track, costing time on reviewer back-and-forth. CloudRoute routes specifically to partners with prior EF cohort approvals.
• Confusing EF cohort participants with EF post-cohort portfolio companies — The standing-arrangement Portfolio access covers post-cohort EF portfolio companies that completed Launch and received the EF investment-committee check. EF cohort participants who chose to step out of cohort during Form or who did not pass investment committee at the end of Launch are not covered by the standing arrangement — they have EF affiliation but file partner-filed Portfolio through the open ACE track. The ceiling for these EF alumni without investment-committee backing tends to land $50K–$75K rather than reaching $100K.

XIWhat a CloudRoute-routed AWS partner does for an EF portfolio company

The partner’s role for an EF portfolio company is structurally similar to other partner-filed engagements, with EF-specific patterns worth flagging.

They file the ACE records — Portfolio, Build for Startups, Bedrock POC if applicable — invoking the EF standing-arrangement Portfolio Sub-Program access in the narrative. Cohort city, cohort number, Form-to-Launch transition date, and EF investment-committee outcome go into the narrative explicitly so the reviewer can verify against the EF portfolio list quickly.

They vet the use case and EF cohort context before filing. EF post-cohort companies are typically still pre-revenue or in early pilots at the credit-application moment; the partner sanity-checks the projected consumption and use-case scope against what the reviewer is calibrated to approve for a cohort-stage portfolio company. The 12-month consumption projection in the $3K–$8K/month range is the typical landing for EF post-cohort filings.

They act as the named technical partner on the file. The reviewer queue recognizes partners with prior EF approvals — the standing-arrangement track plus partner track record together compresses the reviewer cycle.

They provide the actual AWS implementation if you want it: production account setup, multi-account AWS Organization structure for the compliance posture, IAM and VPC setup, KMS hierarchy, the migration off prototype-stage infrastructure (Vercel, Render, Supabase, managed Postgres), observability stack, and deployment pipelines. Most EF post-cohort companies want this because the post-cohort window is fundraising-intense and cloud-engineering bandwidth is constrained.

They provide a co-branded readiness assessment — a written document describing the partner-built infrastructure, the architectural trade-offs, the compliance posture (GDPR / PDPA / DPDP / PIPEDA / APPI depending on cohort city), and what to monitor. This deliverable is useful for the post-cohort seed-round data room.