Strengths
- Poolside is building foundation models and software agents for software development.
- Poolside closed a $500M Series B and is valued at about $3B.
Diligence report
Poolside
Poolside Public-Source Startup Diligence Report
Treat Poolside as a high-upside/high-risk AI infrastructure diligence target: proceed only with rigorous private validation of model performance, customer ARR, compute unit economics, data rights, security posture, and competitive retention.
Company profile
Poolside Public-Source Startup Diligence Report
Poolside has strong public financing, founder, technical, and strategic-partner signals, but underwriting risk is high because model economics, customer conversion, benchmark superiority, and IP/data governance are not public.
- Website
- poolside.ai
- Sector
- Foundation models and AI software-development agents for enterprise and public-sector software engineering
- Geography
- United States headquartered with distributed US and Europe team footprint
- Stage
- Late-stage private AI unicorn
- Known aliases
- poolside, poolside.ai, Poolside AI
- Report version
- 1.0
- Timezone
- UTC
Executive summary
Risks
- Compute intensity and burn can outpace commercialization: Poolside trains proprietary models at large GPU scale; revenue, gross margin, committed compute, and runway are not public.
- Customer proof and ARR concentration are undisclosed: Public sources describe pilots and enterprises but do not name customers, ARR, renewal terms, or conversion rates.
- Crowded AI coding market pressures differentiation: Poolside competes with deeply funded incumbents and developer-tool leaders; public evidence lacks independent benchmark and win/loss proof.
- Data, IP, and model-governance liability: Training-data rights, code IP provenance, customer code handling, privacy, and public-sector controls are material and not publicly verifiable.
Gaps
- Poolside audited financial statements, revenue by product, gross margin, burn, runway, debt, and full cap table are not public.
- Poolside product usage, retention by module, pricing realization, roadmap, and SLA performance are not public.
- Poolside customer list, top-account concentration, ARR by customer, renewal rates, NRR, and reference calls are not public.
- Poolside win/loss data and competitive displacement evidence are not public.
- Poolside CAC, sales cycle, channel economics, partner-sourced pipeline, and marketing attribution are not public.
- Poolside R&D budget, model or platform benchmarks, security test results, and roadmap dependencies are not public.
Risk register
R-001: Compute intensity and burn can outpace commercialization
Poolside trains proprietary models at large GPU scale; revenue, gross margin, committed compute, and runway are not public.
Diligence request: Review cloud/GPU contracts, burn, gross margin, revenue pipeline, and capital plan.
R-003: Crowded AI coding market pressures differentiation
Poolside competes with deeply funded incumbents and developer-tool leaders; public evidence lacks independent benchmark and win/loss proof.
Diligence request: Request direct win/loss evidence, pricing benchmarks, and customer switching data.
R-002: Customer proof and ARR concentration are undisclosed
Public sources describe pilots and enterprises but do not name customers, ARR, renewal terms, or conversion rates.
Diligence request: Request signed contracts, ARR by account, pilot conversion, references, and services margin.
R-004: Data, IP, and model-governance liability
Training-data rights, code IP provenance, customer code handling, privacy, and public-sector controls are material and not publicly verifiable.
Diligence request: Review data lineage, IP assignments, OSS/license scans, model evals, DPAs, SOC reports, and incident history.
R-005: Strategic infrastructure dependency
AWS/Bedrock/Trainium and NVIDIA GPU access support distribution and training, but dependency economics and resilience are private.
Diligence request: Review cloud contracts, portability plan, capacity commitments, and fallback infrastructure.
01Financial Information
Funding and valuation are public; economics and burn are private.
I.A Public financing, valuation, and financial quality signals
partially verified confidence: medium
Funding and valuation are public; economics and burn are private.
Evidence gaps
- Poolside audited financial statements, revenue by product, gross margin, burn, runway, debt, and full cap table are not public.
Hidden risks
- Compute intensity and burn can outpace commercialization
Follow-up questions
- Request audited financials, board pack, cap table, financing documents, debt schedules, and monthly operating model.
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| CB Insights tracker | $3.00B valuation; joined 2024-10-02; US/San Francisco; Enterprise Tech. | verified | Confirm post-money valuation, preferences, and current cap table. |
| Series B | $500M Series B led by Bain Capital Ventures with strategic/corporate investors. | verified | Review financing documents and strategic investor rights. |
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Revenue/ARR | No ARR, bookings, or customer revenue public. | not_publicly_verifiable | Request ARR, bookings, pilot conversion, NRR/GRR, and services revenue. |
| Compute burn/runway | Foundation model training described as capital-intensive; 10,000 GPUs disclosed. | partially_verified | Request GPU/cloud commitments, COGS, burn, and runway. |
02Products
Product direction and deployment boundary claims are visible; pricing and adoption are not.
II.A Product portfolio, pricing, and technical packaging
partially verified confidence: medium
Product direction and deployment boundary claims are visible; pricing and adoption are not.
Evidence gaps
- Poolside product usage, retention by module, pricing realization, roadmap, and SLA performance are not public.
Follow-up questions
- Request product analytics, packaging/pricing history, churn cohorts, roadmap, uptime/SLA reports, and support metrics.
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Foundation models/software agents | Long-horizon work starting with software agents. | verified | Request product availability, roadmap, and benchmarks. |
| Enterprise boundary deployment | On-prem/VPC/workstation options, RBAC, connectors to repos/databases/private corpora. | partially_verified | Review deployment architecture and customer security reports. |
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Pricing | No public pricing or packaging surfaced. | not_publicly_verifiable | Request price book, discounting, usage economics, and margin by deployment type. |
| ROI/productivity | No transparent customer productivity study public. | not_publicly_verifiable | Request pilot KPIs and controlled productivity studies. |
03Customer Information
Enterprise/public-sector and AWS signals are credible; customer names and ARR are private.
III.A Customers, partners, and operational dependencies
partially verified confidence: medium
Enterprise/public-sector and AWS signals are credible; customer names and ARR are private.
Evidence gaps
- Poolside customer list, top-account concentration, ARR by customer, renewal rates, NRR, and reference calls are not public.
Hidden risks
- Customer proof and ARR concentration are undisclosed
- Strategic infrastructure dependency
Follow-up questions
- Request anonymized customer ARR schedule, top-20 account contracts, churn log, and reference-call access.
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Enterprise/public-sector pilots | Public claims of pilots with largest enterprises, Global 2000, public sector. | partially_verified | Request named customers, references, ARR, and conversion data. |
| Customer concentration | No named customers or concentration data public. | not_publicly_verifiable | Request top-account ARR and contract terms. |
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| AWS | Bedrock, Trainium, and combined GTM partnership described. | verified | Review AWS terms and marketplace economics. |
| Strategic investors/suppliers | NVIDIA, Citi, Capital One, HSBC, LG, eBay Ventures participated. | verified | Review commercial terms and strategic dependencies. |
04Competition
Differentiation claims are plausible but unproven against intense competition.
IV.A Competitive positioning and basis of competition
partially verified confidence: medium
Differentiation claims are plausible but unproven against intense competition.
Evidence gaps
- Poolside win/loss data and competitive displacement evidence are not public.
Hidden risks
- Crowded AI coding market pressures differentiation
Follow-up questions
- Request CRM win/loss exports, competitive battlecards, pipeline conversion by competitor, and gross retention by segment.
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| AI coding incumbents | GitHub Copilot, Cursor/Anysphere, Windsurf/Codeium, Cognition, OpenAI, Anthropic, Google overlap. | partially_verified | Request direct win/loss evidence, pricing benchmarks, and customer switching data. |
| Differentiation claims | Proprietary models, RLCEF, enterprise boundary deployment, AWS distribution. | partially_verified | Request benchmark and win/loss evidence. |
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Model ownership | Poolside trains own models point/malibu. | verified | Review evals and model cards. |
| Secure enterprise deployment | Boundary/privacy claims public; audits not public. | partially_verified | Request security audits and customer deployment evidence. |
05Marketing, Sales, and Distribution
Enterprise/public-sector GTM and AWS channel are visible; sales efficiency is not.
V.A Go-to-market channels and demand generation
partially verified confidence: medium
Enterprise/public-sector GTM and AWS channel are visible; sales efficiency is not.
Evidence gaps
- Poolside CAC, sales cycle, channel economics, partner-sourced pipeline, and marketing attribution are not public.
Hidden risks
- Customer proof and ARR concentration are undisclosed
Follow-up questions
- Request CAC payback, sales cycle, pipeline source, channel margin, partner agreements, and quota attainment.
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Forward deployed research | FDR engineers embed with teams. | verified | Review services margin and delivery capacity. |
| AWS channel | Bedrock first-party availability and combined GTM. | verified | Request marketplace conversion and AWS co-sell metrics. |
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| GTM leadership | GTM team led by Paul St. John, previously CRO at GitHub. | verified | Review quota plan and pipeline generation. |
| Sales efficiency | CAC/payback/pipeline not public. | not_publicly_verifiable | Request pipeline, bookings, CAC, and sales productivity. |
06Research and Development
Model/RLCEF/GPU signals are strong; evals and provenance need diligence.
VI.A R&D organization, roadmap, and technical defensibility
partially verified confidence: medium
Model/RLCEF/GPU signals are strong; evals and provenance need diligence.
Evidence gaps
- Poolside R&D budget, model or platform benchmarks, security test results, and roadmap dependencies are not public.
Hidden risks
- Compute intensity and burn can outpace commercialization
- Data, IP, and model-governance liability
- Strategic infrastructure dependency
Follow-up questions
- Request R&D roadmap, architecture review, model/platform benchmarks, vuln management evidence, and engineering productivity data.
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Models and RLCEF | point, malibu, RLCEF described by AWS. | verified | Request model evals, safety testing, and roadmap. |
| GPU cluster | 10,000 NVIDIA GPUs disclosed. | partially_verified | Review capacity, cost, utilization, and reliability. |
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Product roadmap | AI assistant and foundation model expansion described, but dates/KPIs not public. | partially_verified | Request roadmap, model eval cadence, and release plan. |
| Data/IP provenance | Not public. | not_publicly_verifiable | Request training-data lineage, code license scan, and IP assignment evidence. |
07Management and Personnel
Founders and team profile are visible; HR and governance metrics are private.
VII.A Leadership, hiring, and organizational signals
partially verified confidence: medium
Founders and team profile are visible; HR and governance metrics are private.
Evidence gaps
- Poolside headcount by function, attrition, compensation, hiring plan, board composition, and option refresh history are not public.
Follow-up questions
- Request HRIS exports, org chart, attrition report, compensation bands, board minutes, and hiring plan.
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Jason Warner | Founder/CEO; former CTO at GitHub. | verified | Review founder vesting and key-person planning. |
| Eiso Kant | Founder/CTO; founder of source{d}. | verified | Review founder IP and technical succession. |
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Team footprint | US headquartered, US/Europe team, researchers from DeepMind, Yandex, Amazon, Uber. | partially_verified | Request headcount and hiring plan by region/function. |
| Attrition/compensation | Not public. | not_publicly_verifiable | Request HRIS, attrition, compensation, and option pool data. |
08Legal and Related Matters
Privacy/security claims are material but require legal/security artifacts.
VIII.A Legal, regulatory, privacy, and IP matters
partially verified confidence: medium
Privacy/security claims are material but require legal/security artifacts.
Evidence gaps
- Poolside complete litigation searches, IP assignment chain, open-source license scan, DPIAs, DPAs, and regulatory correspondence are not public.
Hidden risks
- Data, IP, and model-governance liability
Follow-up questions
- Request counsel memo, litigation schedule, IP assignment records, open-source scan, DPAs, SOC/security reports, and privacy impact assessments.
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Privacy/security boundary claims | Data stays under customer control; privacy/security emphasized. | partially_verified | Review DPAs, SOC reports, security architecture, and incident history. |
| Public-sector/high-consequence use cases | Enterprise/public sector/high-consequence positioning. | partially_verified | Review procurement controls, export-control analysis, and compliance artifacts. |
| item | public evidence | verification status | diligence request |
|---|---|---|---|
| Training data/IP rights | Not public. | not_publicly_verifiable | Request data provenance, licensing, OSS scan, and indemnity terms. |
| Litigation/regulatory | No surfaced public matters in scoped research. | not_publicly_verifiable | Request counsel litigation schedule and regulatory correspondence. |
Evidence
| ID | Claim | Status | Sources |
|---|---|---|---|
| EC-001 | CB Insights lists Poolside at a $3.00B valuation. | verified medium | SRC-001 |
| EC-002 | Poolside announced a $500M Series B led by Bain Capital Ventures with corporate investors including NVIDIA, Citi Ventures, Capital One Ventures, HSBC Ventures, LG Technology Ventures, and eBay Ventures. | verified high | SRC-003 |
| EC-003 | Revenue, burn, runway, compute commitments, valuation terms, and customer ARR are not public. | not publicly verifiable high | SRC-003 |
| EC-004 | Poolside builds foundation models for long-horizon work starting with software agents. | verified high | SRC-002 |
| EC-005 | Poolside markets enterprise deployment inside customer security boundaries with repository/database/private-corpora connectors. | verified medium | SRC-002 |
| EC-006 | Public pricing, product GA maturity, retention, and measurable productivity ROI are not disclosed. | not publicly verifiable high | SRC-002 |
| EC-007 | Poolside says its products are being battle-tested with enterprises and pilots with large enterprises/public sector. | partially verified medium | SRC-002SRC-003 |
| EC-008 | AWS is a strategic infrastructure and GTM partner for Poolside models through Bedrock and Trainium. | verified high | SRC-006 |
| EC-009 | Customer names, ARR concentration, signed public-sector contracts, and renewal references are not public. | not publicly verifiable high | SRC-002SRC-003 |
| EC-010 | Poolside competes in the AI software development assistant/agent market against GitHub Copilot, Cursor/Anysphere, Codeium/Windsurf, Cognition, OpenAI, Anthropic, Google, and other model providers. | partially verified medium | SRC-002SRC-006 |
| EC-011 | Poolside differentiates on proprietary foundation models, RLCEF, enterprise deployment boundaries, and AWS distribution. | partially verified medium | SRC-003SRC-006 |
| EC-012 | Independent benchmark superiority and win/loss versus incumbents are not publicly proven. | not publicly verifiable high | SRC-002 |
| EC-013 | Poolside pursues enterprise and public-sector GTM with forward deployed research engineers and strategic cloud distribution. | partially verified medium | SRC-002SRC-006 |
| EC-014 | Funding announcement says Poolside has a strong go-to-market focus and a GTM team led by Paul St. John, previously CRO at GitHub. | verified medium | SRC-003 |
| EC-015 | Pipeline, bookings, CAC, sales cycle, services margin, and cloud marketplace conversion are not public. | not publicly verifiable high | SRC-002SRC-006 |
| EC-016 | Poolside trains proprietary models point and malibu and uses RLCEF. | verified high | SRC-006 |
| EC-017 | Poolside brought online 10,000 NVIDIA GPUs according to its funding announcement. | partially verified medium | SRC-003 |
| EC-018 | Model quality, safety evaluations, IP provenance, data licensing, and compute-cost efficiency are not publicly verifiable. | not publicly verifiable high | SRC-002SRC-006 |
| EC-019 | Poolside was founded in April 2023 by Jason Warner and Eiso Kant. | verified high | SRC-003SRC-006 |
| EC-020 | Poolside reports a distributed US and Europe team with applied researchers and engineers from DeepMind, Yandex, Amazon, Uber, and more. | partially verified medium | SRC-003 |
| EC-021 | Current headcount, attrition, compensation, hiring velocity, board composition, and founder vesting are not public. | not publicly verifiable high | SRC-003 |
| EC-022 | Poolside makes privacy/security boundary claims for enterprise deployments. | partially verified medium | SRC-002SRC-003SRC-007 |
| EC-023 | Poolside serves high-consequence enterprise and public-sector use cases that may trigger procurement, data-security, export-control, and IP scrutiny. | partially verified medium | SRC-002SRC-003 |
| EC-024 | Complete IP provenance, training-data rights, customer DPAs, SOC reports, export-control review, and litigation searches are not public. | not publicly verifiable high | SRC-008 |
| ID | Publisher | Title | Accessed |
|---|---|---|---|
| SRC-001 | CB Insights | The Complete List Of Unicorn Companies | 2026-05-29 |
| SRC-002 | Poolside | Poolside: Frontier research to operational intelligence | 2026-05-29 |
| SRC-003 | PRWeb / Poolside | poolside closes $500 Million Series B to build the world's most capable AI for software development | 2026-05-29 |
| SRC-004 | TechCrunch | AI coding startup Poolside raises $500M from eBay, Nvidia, and others | 2026-05-29 |
| SRC-005 | Bain Capital Ventures | Poolside is Scaling a World-Class Research | 2026-05-29 |
| SRC-006 | Amazon Web Services | How poolside pioneers AI assisted software development on AWS | 2026-05-29 |
| SRC-007 | Citi Ventures | Investing in poolside, the Company Building the World’s Most Capable AI Code Assistant | 2026-05-29 |
| SRC-008 | Poolside | Privacy Policy — Poolside | 2026-05-29 |
Disclaimer
This report is a public-evidence diligence snapshot, not investment advice. Important financial, legal, technical, and contractual facts remain non-public and should be verified directly with management and primary documents before any investment decision.