Dataiku

1.1 Identity, product, and operating model

Dataiku was founded in Paris in 2013, and retained current descriptions place its operating base in New York City. Its platform spans data preparation, analytics, machine learning, deployment, monitoring, and AI governance for both technical and business users. That breadth is consistent across independent profiles, a CIO executive demonstration, review material, and company-issued marketplace copy. The commercial model is described as enterprise subscription software, with annual or multi-year contracts supplemented by implementation and training. The evidence therefore supports a one-line identity as a governed enterprise AI orchestration platform, not a foundation-model developer or cloud infrastructure provider. Exact pricing, services mix, and contract economics remain private.[CO001, CO002, CO003, CO004]

1.2 Leadership, governance, and key-person dependence

Capital.com identifies Florian Douetteau, Clément Stenac, Marc Batty, and Thomas Cabrol as the four founders. SiliconANGLE's March 2026 reporting identifies Douetteau as co-founder and chief executive and Stenac as co-founder and chief technology officer, tying both founders directly to the latest product strategy. That continuity is a strength because the company's move from collaborative data science toward governed agent orchestration has clear founder sponsorship. It is also a concentration risk: the retained pool does not provide a complete current executive roster, board committee structure, succession plan, or delegated decision-rights map. Governance quality and operating depth therefore remain only partially observable.[CO010, CO012]

1.3 Capital history, scale, and stakeholder structure

The late-stage financing is directionally clear but not identical across sources. Tracxn records a $204 million Series F on November 25, 2022 at a $3.7 billion post-money valuation, while Reuters describes a $200 million Series F in December 2022 at the same valuation. These should be treated as conflicting representations of one round until primary closing documents reconcile amount and date. Tracxn also reports $1.04 billion of total funding across nine rounds, a database figure that is not equivalent to cash currently available. Reuters reported more than 700 customer organizations and about $300 million of annualized recurring revenue in January 2025; Sacra later estimated $342.5 million of ARR in September 2025. Neither figure is audited revenue.[CO005, CO006, CO008, CO009, CO013, CO015, CO016]

1.4 Milestones, public-market option, and diligence limits

Reuters reported in October 2025 that Dataiku had hired Morgan Stanley, Citigroup, and other banks to prepare for a possible first-half 2026 IPO, while explicitly warning that timing could change. The retained pool does not establish that an offering occurred or that a public filing is active as of the run date. In March 2026, SiliconANGLE reported a Platform for AI Success expansion built around Agent Management, Cobuild, and Reasoning Systems, reinforcing the shift toward governed agent orchestration. Independent adverse analysis provides an important market constraint: enterprise agent adoption depends on data cleanup, integration, governance, and organizational change and may mature more slowly than the product cycle implies. Current audited revenue, margins, retention, cash, burn, customer concentration, and complete legal or regulatory status are not established by the assigned evidence.[CO007, CO011, CO014]

1.5 Exhibits

2.1 Market boundary and substitutes

The relevant market is enterprise AI and analytics infrastructure that helps organizations prepare data, build and deploy models or agents, and govern those workflows across cloud, on-premises, and multi-vendor environments. Dataiku's own positioning describes an orchestration layer for building, deploying, and governing AI, while AWS Marketplace describes a single governed system spanning analytics, machine learning, and AI agents (SM004, SM005). Independent product coverage similarly describes an end-to-end data and AI platform with visual and coded workflows, deployment, monitoring, and governance (SM007). This boundary includes platform software and directly attached implementation or managed-service spend, but excludes foundation-model consumption, general cloud infrastructure, standalone BI visualization, and bespoke consulting unless they are purchased as part of operating the platform. The practical substitutes are fragmented stacks: spreadsheets and SQL for analysts, notebooks and Databricks for data scientists, Alteryx or ETL tools for preparation, and Tableau or Power BI for visualization. A CIO demonstration reports that teams otherwise work in siloed tools and email spreadsheets between analysts, engineers, and data scientists (SM006). The implication is that Dataiku competes for consolidation and governance budgets, not only for a narrowly defined MLOps line item.[CM001, CM002, CM003, CM004]

2.2 Sizing lenses and market momentum

Public sizing is directionally positive but not directly additive. MarketsandMarkets lists an AI Studio market of $32.7 billion by 2029 at a 38.4% CAGR, an AI governance market of $5.78 billion by 2029 at a 45.3% CAGR, and an AI platform market rising from about $18.22 billion in 2025 to more than $94.31 billion in 2030 at about 38.9% CAGR (SM010). These categories overlap in platform, governance, MLOps, analytics, and application-development spend, so they are upper-bound reference lenses rather than a defensible Dataiku TAM. A narrower adoption lens is more decision-useful: Deloitte's poll found only 13.5% of finance and accounting respondents said their organizations were already using agentic AI, while 33.6% were building or planning adoption (SM011). Academic evidence is also early-stage: a study of 16 practitioners from 12 companies found reported production practices at maturity levels 1-3 and recurring barriers around context, proprietary data, qualification, and confidentiality (SM008). Dataiku's addressable opportunity therefore expands with governance and orchestration needs, but conversion depends on enterprises moving from pilots into repeatable, measured production workflows.[CM005, CM006, CM007, CM008, CM009]

2.3 Buyer segments, adoption path, and constraints

The primary buyer is an enterprise data or AI leader who owns platform standardization and governance; data engineering and data science are the core users, while business analysts consume governed outputs and domain leaders own the operating budget. The buyer map is consequently organized around workflow maturity rather than company size. Early adopters can justify a shared platform when siloed tooling, audit requirements, or repeated deployment work creates measurable friction. The adoption trigger is strongest where a workflow has a clear baseline, high volume, and defined human escalation path. Dataiku's platform coverage and the FirstMark launch narrative emphasize connecting data, models, agents, and business rules in a governed environment (SM004, SM001). The principal constraints are trust, integration, talent, and unclear economics: Deloitte identifies trust as the leading barrier at 21.3%, followed by integration at 20.1% and skilled personnel at 13.5% (SM011). A separate deployment analysis says early costs are higher and less predictable than projected, with recurring connector, monitoring, and incident-response costs (SM009). This makes ROI proof, interoperability, and governance evidence central to sales cycles and valuation durability; broad market growth alone does not establish Dataiku's SAM or SOM.[CM010, CM011, CM012, CM013, CM014]

2.4 Exhibits

3.1 Competitive landscape and buyer alternatives

Dataiku is not confined to a single peer set. Its broadest direct comparison is with end-to-end enterprise AI platforms, while Databricks competes from a data-platform base and H2O.ai and DataRobot represent more specialist AI and AutoML alternatives. Alteryx and manual combinations of spreadsheets, SQL, notebooks, and point tools remain substitutes for narrower analytics workflows. Internal builds using open-source orchestration frameworks are another credible option for technically mature buyers. Dataiku's own field demonstration names Excel, Alteryx, Tableau Data Prep, SQL scripts, notebooks, and Databricks as the fragmented status quo it seeks to consolidate, so the relevant market includes both vendor displacement and toolchain consolidation rather than only head-to-head platform replacement.[CP001, CP002, CP006, CP007, CP008, CP009, CP010, CP011, CP012]

3.2 Capability and positioning

Dataiku's strongest supported position is between business accessibility and enterprise control. The platform combines visual workflows with Python, R, and SQL, works over existing data locations, and promotes centralized governance across analytics, models, and agents. Independent review evidence broadly supports the accessibility thesis, but also records a learning curve, workflow complexity, performance variability, and implementation friction. Databricks brings a much larger integrated data and AI estate, while H2O.ai combines an open-source framework with proprietary applications. The positioning map is an analyst ordinal synthesis of these qualitative facts, not a benchmark or measured product score; cells lacking direct evidence are marked unknown.[CP001, CP002, CP003, CP004, CP007, CP009, CP012, CP013]

3.3 Pricing, switching costs, and distribution

Public pricing evidence is weakest where enterprise buyers most need comparability. The AWS Marketplace listing uses one contract-based Units dimension but does not define a public unit mapping or price, while Databricks uses consumption-based Databricks Units tied to workload usage. No current, comparable public price was available in the assigned evidence for H2O.ai, DataRobot, or Alteryx. Dataiku can gain switching cost when visual flows, governance policy, deployment processes, and mixed-user collaboration become embedded; a G2 reviewer explicitly described moving workflows from Alteryx. Conversely, Dataiku's push-down model leaves underlying data in customer infrastructure, which reduces data-residency lock-in even as workflow and governance configuration create process lock-in.[CP002, CP003, CP005, CP008, CP012, CP013]

3.4 Moat durability and displacement risk

The moat is therefore workflow and governance integration rather than exclusive ownership of data or models. Dataiku's 2026 roadmap extends the control plane into agent monitoring, business-led agent creation, and reasoning systems, which could deepen enterprise standardization if these products become operationally central. The adverse case is platform convergence: Databricks is adding governed agents and business-user interfaces, hyperscalers already own infrastructure distribution, and Dataiku itself describes capable open-source orchestration frameworks. Customer evidence also shows that cost, setup complexity, access controls, large-workflow performance, and onboarding can erode the benefit of consolidation. Durable differentiation must be tested through win/loss data, realized implementation time, renewal behavior, and migration cost rather than feature breadth alone.[CP003, CP004, CP009, CP013, CP014, CP015]

3.5 Exhibits

4.1 Monetization and revenue quality

Dataiku's public monetization evidence points to annual subscription access to Data Science Studio and related capabilities, supplemented by professional services, ecosystem partnerships, and marketplace add-ons. Sacra estimates $342.5M of ARR in September 2025, up from $300M at the end of 2024; that is an analyst estimate, not audited revenue. Sacra separately reports roughly 500 customers and states approximately $500K of average ARR per customer at the end of 2023, a historical metric that should not be treated as a current cohort denominator. AWS Marketplace confirms a contract-based “Units” dimension but says neither the unit definition nor price is published. Revenue recognition, renewal rates, services mix, implementation revenue, and concentration remain unverified.[CI001, CI002, CI003, CI004, CI005]

4.2 Pricing architecture and sales efficiency

Dataiku does not publish a standard paid price list. A 2026 pricing review identifies a no-cost self-hosted Free Edition and a 14-day hosted trial, while paid deployments route to a sales conversation and are shaped by deployment model, role-based seats, governance, LLM Mesh access, organization size, integrations, and support. AWS Marketplace independently describes pricing as based on contract duration and terms, with additional AWS infrastructure costs possible, but does not define a unit price. These facts imply a high-touch enterprise sales motion with scope-based quoting and potentially material implementation or cloud-cost pass-through. Public sources do not establish sales-cycle length, CAC, payback, win rates, discounting, partner-sourced pipeline, or realized net price, so a sales-efficiency conclusion would be speculative.[CI006, CI007, CI008, CI009, CI010]

4.3 Cost structure and capital adequacy

The available evidence describes a software platform that can run above customer infrastructure, with compute pushed to external engines or customer environments and additional cloud charges possible. That architecture may reduce owned-infrastructure intensity, but it does not establish Dataiku's gross margin because hosting, support, implementation labor, research and development, stock compensation, and third-party model costs are undisclosed. Capital.com reports that Dataiku began preparing for a potential U.S. IPO after hiring banks, but IPO preparation is not cash and does not establish financing availability. No assigned source reports cash on hand, burn, debt, liquidation preferences, or a next-round trigger. Capital adequacy is therefore a material diligence blocker rather than a quantified conclusion.[CI008, CI012, CI013, CI014, CI015]

4.4 Financial underwriting verdict

Dataiku has the profile of a scaled enterprise software asset: recurring subscription mechanics, large-account positioning, a broad governed platform, and estimated ARR growth through September 2025. The underwriting case remains constrained by information asymmetry. Current ARR is an estimate, the latest customer denominator is historical, and negotiated contracts leave units, discounts, renewal economics, and realized price private. A peer data point from DataRobot also relies on third-party estimates with nonmatching revenue and headcount dates, so it cannot substitute for Dataiku unit economics. The financial stance remains track pending an ARR bridge, retention cohorts, gross-margin bridge, services split, cash and burn schedules, debt terms, and concentration.[CI002, CI003, CI009, CI013, CI014, CI015, CI016, CI018]

4.5 Exhibits

5.1 Product workflow and user surface

Dataiku is delivered as an enterprise platform for moving from data access and preparation through model development, deployment, and monitoring. A CIO demonstration describes access for coders and non-coders across cloud, on-premises, and other locations, while stating that the platform supports analysis, machine learning, and generative AI in one governed user interface. The same demonstration says Dataiku operates as a UI layer over existing data sources rather than requiring large datasets to be moved, and describes a browser-based experience hosted in the cloud, a VPC, or on-premises. That is a coherent workflow thesis for enterprises with heterogeneous estates, but the public material does not quantify implementation time, deployment success rates, or realized customer outcomes.[CE001, CE002, CE003, CE004]

5.2 Architecture, deployment, and operating model

The current DSS 15 developer guide groups five capabilities—agentic AI, machine learning, modern analytics, orchestration, and governance—inside one enterprise platform. It describes visual and code agents, an LLM Mesh for provider routing and quotas, a unified Flow and deployment automation, model registries, lineage, explainability, and monitoring. AWS Marketplace materials add concrete infrastructure dependencies: elastic compute powered by Amazon EKS for Python, R, and Spark, integrations with SageMaker and Comprehend, and an LLM Mesh connection to Amazon Bedrock. Dataiku's documentation site also exposes a developer guide, API reference, Academy, and community, while the public Python client repository confirms a public DSS API client. These are observable integration surfaces, not proof of uptime, scale limits, or customer-specific architecture.[CE005, CE006, CE007, CE008, CE009]

5.3 Differentiation, trust, and technical diligence

Dataiku's differentiation is the combination of a shared workflow for technical and business users with centralized governance across data, models, LLMs, and agents. Dataiku Govern describes registries for bundles, models, and LLMs, documented approval workflows, sign-off rules, audit timelines, and notifications. Independent marketplace and review material supports a broad multi-vendor and hybrid deployment proposition, but also leaves commercial and operational diligence open: AWS lists unit-based billing whose mapping is set by quote, and marketplace review text includes a complaint that the product's unique capabilities can be confusing compared with similar platforms. A synthesized review page also reports slow performance for large datasets and complex scenarios. Those signals make workload benchmarking, reference calls, access-control testing, and a full pricing/ implementation schedule necessary before treating the architecture as proven at scale.[CE010, CE011, CE012, CE013, CE014]

5.4 Exhibits

6.1 Buyer, user, and deployment segments

Dataiku's customer model is enterprise-wide rather than confined to one technical persona. The platform is described as serving data scientists, engineers, analysts, business users, and domain experts, with use spanning predictive maintenance, fraud detection, demand forecasting, and other operational analytics. This supports a segmentation thesis in which the economic buyer is likely an enterprise data or technology organization, while adoption depends on analysts and domain teams being able to use the same governed workspace. The evidence does not disclose customer-level revenue bands, seat counts, geography mix, or payer-versus-user splits, so scale should not be inferred from logos or a broad organization count.[CU001, CU002, CU003, CU004]

6.2 Adoption trajectory and named customer proof

Public evidence indicates a sizable enterprise footprint, but it does not establish an auditable active-account or deployment trajectory. Reuters reports more than 700 organizations and names Johnson & Johnson, Toyota, General Electric, and BNP Paribas; Capital.com repeats the greater-than-700 figure and gives the same examples. Dataiku's own surface adds a different scale signal—"more than 600 companies" and tens of thousands of people with measurable return on AI—while Forbes reports more than 700 organizations and lists LVMH, Morgan Stanley, Michelin, and Novartis. These are useful reference points, but differing dates, definitions, and issuers prevent treating them as one reconciled customer KPI.[CU005, CU006, CU007, CU008, CU009]

6.3 Durability, expansion, and procurement friction

Dataiku's product positioning is compatible with land-and-expand behavior: it can begin with a team or use case and extend across analytics, machine learning, generative AI, deployment, monitoring, and governance. AWS Marketplace describes contract-based Units pricing with a negotiated quantity and term, while reviews show both cross-functional value and friction. One reviewer says the platform lets data scientists, business users, and analysts work on one visual platform; the same review reports that setup may be difficult and the product may be expensive. Other review excerpts mention comparisons with Alteryx, confusion about differentiated capabilities, and missing niche data sources. No public source in this pool reports NRR, GRR, churn, renewal rates, contract length, cohort expansion, or customer concentration.[CU010, CU011, CU012, CU013]

6.4 Adoption loop and unresolved customer diligence

The observable customer journey runs from connecting existing data, through collaborative preparation and modeling, to deployment, monitoring, and governance. That sequence is reinforced by independent product coverage, an AWS marketplace listing, customer-review excerpts, and Dataiku's own platform description. The commercial implication is positive if one governed workflow creates multiple internal stakeholders and recurring operational dependence; the risk is that implementation complexity, infrastructure requirements, services intensity, and procurement friction may slow expansion or make the platform heavier than a point solution. The available sources do not provide customer-level outcome baselines, renewal cohorts, implementation duration, services mix, or concentration data, so those remain material diligence asks.[CU014, CU015, CU016]

6.5 Exhibits

7.1 Regulatory and legal exposure

Dataiku sells governance into a market where the rules are becoming operational rather than aspirational. The European Commission says high-risk AI systems will require risk assessment, dataset quality, logging, documentation, human oversight, robustness, cybersecurity, and accuracy obligations, with high-risk use cases applying from December 2027 under the current implementation timeline. That creates demand for controls but also raises the cost of proving that a customer's particular deployment is compliant. Dataiku's public product materials describe governance, lineage, monitoring, and agent oversight, yet the available evidence does not establish that these features satisfy every jurisdictional obligation or that Dataiku is itself the regulated provider or deployer in each customer arrangement. The most important legal diligence is therefore a product-to-obligation mapping by deployment model, jurisdiction, and use case, rather than assuming that a governance label is a compliance certification. The FTC legal library illustrates the breadth and pace of privacy, competition, and consumer-protection enforcement generally; it does not identify a Dataiku matter, so no company-specific enforcement conclusion is warranted.[CR001, CR002, CR003, CR004]

7.2 Operational, quality, and security exposure

The operating model is inherently complex: Dataiku presents a single interface over cloud, on-premises, databases, data warehouses, model environments, and third-party agent frameworks. That breadth can reduce data movement, but it also creates failure modes at connectors, permissions, model behavior, deployment orchestration, and monitoring boundaries. A company-hosted demonstration says Dataiku tracks users' access to data sources, models, and LLMs, while the AWS listing states that infrastructure costs may be additional and that AWS does not warrant vendor-supplied product content. Those statements support the existence of controls and a clear shared-responsibility boundary, not an absence of incidents. The new agent products are also a delivery risk: independent coverage describes Agent Management as not yet available at the time of reporting and places other capabilities on a 2026 launch schedule. The practical kill criteria are measurable: unresolved critical security findings, material data leakage, repeated model-drift or cost-overrun alerts without closure, or launch slippage that prevents referenceable production deployments.[CR005, CR006, CR007, CR008, CR009]

7.3 Partner and financial dependencies

Dataiku's value proposition depends on an ecosystem rather than a self-contained stack. AWS Marketplace describes integrations with S3, Redshift, RDS, EKS, SageMaker, Comprehend, Bedrock, and customer-controlled deployment choices, while current industry coverage describes monitoring across AWS, Snowflake, Databricks, and Google environments. This expands addressable use cases but increases exposure to platform APIs, cloud economics, procurement channels, and competing control planes. Commercial opacity compounds that risk: the AWS listing uses a contract-based Units measure, does not publish the unit definition or price, and says buyers must obtain a quote; additional AWS infrastructure costs may apply. Public review evidence is useful as a signal of user experience but is not a substitute for retention, concentration, margin, or cash data. A financing or IPO process was reported in 2025, and the prior 2022 round and valuation were reported by Reuters-syndicated coverage, but those facts do not show current liquidity or runway. Diligence should obtain partner-sourced pipeline, renewal and churn by deployment type, cloud-cost pass-through, cash and burn, and concentration by top customers and providers before treating ecosystem breadth as resilience.[CR010, CR011, CR012, CR013]

7.4 Exhibits

8.1 Recommendation and valuation stance

The public record supports a track recommendation with medium confidence and high risk. Dataiku reported $350M+ ARR in October 2025, more than 750 organizations, and more than 1,250 employees, while Sacra estimated $342.5M ARR in September 2025. Those figures show meaningful scale, but they are not audited financial statements and do not disclose gross margin, net retention, cash generation, or customer concentration. The last disclosed post-money valuation in the retained evidence is $3.7B from the December 2022 Series F; the $4.6B August 2021 Series E provides a clear precedent for multiple compression rather than a current mark. Comparing either historical valuation with later estimated ARR would mix periods and would not establish a current transaction multiple. The retained evidence is therefore insufficient to justify a buy without current financing terms, preference structure, and evidence that growth and margins support the price.[CV001, CV002, CV003, CV004]

8.2 Investment thesis and anti-thesis

The thesis is that Dataiku can monetize the coordination layer around enterprise AI: its official product page describes a governed system combining analytics, models, and agents, while an independent SiliconANGLE report describes a move toward an orchestration layer for enterprise-grade AI agents. This positioning is commercially relevant because the platform spans technical and business users and is designed to operate across existing data environments. The anti-thesis is that hyperscalers, data platforms, and specialist tools can absorb these functions, leaving a neutral layer with pricing pressure and high implementation requirements. Sacra's independent analysis describes subscription pricing based on users and compute and identifies competition from DataRobot, Alteryx, Databricks, and cloud ML platforms, but it also warns that public information is incomplete and not investment advice. The valuation therefore depends on proving durable expansion, efficient deployment, and a defensible position above infrastructure-native alternatives.[CV005, CV006, CV007, CV008]

8.3 Scenarios, entry discipline, and diligence

A scenario framework is more defensible than a single target because the public evidence does not establish a current price, audited revenue, margin profile, retention, or dilution and preference terms. The bull case requires verified ARR growth, strong retention, healthy gross margin, and evidence that agent governance expands wallet share without proportional services intensity. The base case assumes continued enterprise adoption but a fair private mark, with an entry that requires negotiated terms and a margin of safety. The bear case is a down-round or multiple reset if growth decelerates, cloud and platform vendors bundle comparable controls, or implementation economics remain heavy. Exit readiness should be treated as conditional: Sacra reports a potential U.S. IPO process, but that is a secondary report and not a filing or commitment. Before investing, request audited financials, a current cap table, liquidation preferences, ARR bridge, cohort retention, gross margin, services mix, customer concentration, cash and burn, and current financing terms.[CV009, CV010, CV011, CV012]

8.4 Exhibits