# The Enterprise AI Cost Model: Per-Seat vs Token vs Owned > Enterprise · AI Course · ENT-3 > Source: https://ibl.ai/solutions/enterprise/course/enterprise-ai-cost-model > Last updated: 2026-08-25 **Model AI spend across pricing shapes at real headcount — where per-seat licensing breaks, what tokens actually cost, and when owning the stack wins.** ## The Short Answer **Per-seat AI licensing scales with headcount rather than usage, so a 50,000-employee organization pays for tens of thousands of seats that never log in. ibl.ai has no per-seat pricing — usage-based or self-hosted — and because you own all the code and the data, the switching cost that makes seat licences sticky largely disappears.** On ibl.ai you own all the code and the data, run it model-agnostic across any LLM, and pay with no per-seat pricing — so you can deploy anywhere, from your own cloud to a fully air-gapped network. [Request Access](https://ibl.ai/contact) · [Explore Enterprise](https://ibl.ai/solutions/enterprise) ## Course facts - **Level:** Foundational - **Duration:** 5.5 hours across 8 modules - **Format:** Workshop with spreadsheet modeling - **Modules:** 8 - **Catalog code:** ENT-3 - **Frameworks covered:** TCO analysis, NIST AI RMF ## What is this course about? Per-seat AI licensing scales with headcount regardless of use, so an organization above a few thousand employees pays many times over for a workload it could meter directly. This course builds the three-scenario model on the learner's own numbers, including the utilization problem, the hidden line items, and an honest account of when self-hosting loses. ## Who is this course for? - CFOs, finance business partners, and FP&A staff - CIOs and IT finance leads - Procurement and vendor management - Heads of AI and transformation programs ### What do I need before starting? - Bring headcount figures and any current AI vendor quotes - Comfort with a spreadsheet ## What will I be able to do afterwards? - Classify any AI quote by pricing shape and name its hidden assumption - Model per-seat cost at 1,000, 5,000, and 50,000 employees - Estimate token cost bottom-up from workload rather than headcount - Compute a defensible self-hosting break-even including operations - Present a three-year TCO with sensitivity analysis to a finance committee ## What does each module cover? ### Module 1 — What are the three pricing shapes? Per-seat, usage-based, and owned — and what each assumes about your organization. _(35 min)_ **Objectives** - Classify a quote by underlying pricing shape - Name the assumption each shape embeds - Recognize hybrid pricing designed to prevent comparison **Topics:** Per-seat · Usage-based · Owned and self-hosted · Hybrid obfuscation **Activity:** Classify three real quotes and state each one's hidden assumption. ### Module 2 — What does per-seat cost at 50,000 employees? The headcount multiplication that makes per-seat structurally wrong above a threshold. _(45 min)_ **Objectives** - Build the multiplied bill at three organization sizes - Identify the headcount at which the shape becomes indefensible - Model tiered and volume-discounted pricing accurately **Topics:** Headcount multiplication · Tier modeling · Volume discounts · Threshold identification **Activity:** Build the multiplied bill at 1,000, 5,000, and 50,000 seats for a real quote. ### Module 3 — How many seats never log in? The utilization problem, and the effective cost per active user it produces. _(45 min)_ **Objectives** - Measure real utilization from existing tool data - Compute effective cost per active user - Project utilization for a new deployment honestly **Topics:** Utilization measurement · Effective cost per active user · Adoption curves · Projection realism **Activity:** Compute effective cost per active user from a real deployment's login data. ### Module 4 — How do you model token cost from workload? Bottom-up cost estimation from what the work actually consumes. _(50 min)_ **Objectives** - Characterize representative workloads in tokens - Account for retrieval and context overhead - Model cost against volume growth **Topics:** Workload characterization · Context overhead · Input/output asymmetry · Volume modeling **Activity:** Measure token consumption for three real workflows and extrapolate annually. ### Module 5 — When does self-hosting actually break even? GPU capital, utilization, and operations staffing, modeled without optimism. _(50 min)_ **Objectives** - Model hardware capital and amortization - Include realistic operations headcount - Identify the conditions under which self-hosting loses **Topics:** GPU capital · Utilization economics · Operations staffing · Losing conditions **Activity:** Compute break-even volume and state the assumptions it depends on. ### Module 6 — Which costs does nobody budget? Integration, evaluation, monitoring, retraining, and exit. _(45 min)_ **Objectives** - Estimate integration and change management - Budget ongoing evaluation and monitoring - Quantify exit cost and its negotiating consequence **Topics:** Integration · Evaluation overhead · Monitoring · Exit and switching cost **Activity:** Add all five forgotten line items and observe how the comparison shifts. ### Module 7 — How do you present this to a finance committee? A business case built on risk and sensitivity rather than a single confident number. _(45 min)_ **Objectives** - Frame the case around risk and optionality - Present sensitivity rather than point estimates - Pre-answer the standard finance objections **Topics:** Risk framing · Sensitivity presentation · Objection handling · Staged commitment **Activity:** Present your model to a cohort playing a hostile finance committee. ### Module 8 — Building your three-scenario model The workshop module: a complete model on the learner's own figures. _(50 min)_ **Objectives** - Complete all three scenarios with real numbers - Run sensitivity on the decisive variables - Produce a one-page decision summary **Topics:** Model completion · Sensitivity analysis · Decision summary · Assumption documentation **Activity:** Finish the model and write the one-page summary. ## What is the capstone project? **Three-year enterprise AI TCO model.** Build a complete three-scenario cost model on your organization's real headcount, utilization, and workload data, including all hidden line items and a sensitivity analysis, with a one-page summary for an investment committee. _Deliverable:_ A working model with documented assumptions and a one-page decision summary. ## How are learners assessed? - Model reviewed for arithmetic and explicitly stated assumptions - Sensitivity analysis must identify the genuinely decisive variables - Committee presentation assessed on objection handling ## What ships with the course? - **Facilitator guide.** Session-by-session running order, discussion prompts, and the questions that reliably derail a room. - **Learner workbook.** Exercises, checklists, and the templates each module's activity produces. - **Hands-on lab environment.** A sandboxed ibl.ai deployment so exercises run against real agents, not screenshots. - **Assessment bank.** Scenario questions and rubric criteria mapped to each stated learning outcome. - **Source bibliography.** Every primary regulation and standard cited on this page, linked and dated. ## Which AI agents does this course use? - [Operations Agent](https://ibl.ai/solutions/enterprise/agent/operations-agent) - [Data Analysis Agent](https://ibl.ai/solutions/enterprise/agent/data-analysis-agent) - [Enterprise Assistant](https://ibl.ai/solutions/enterprise/agent/enterprise-assistant) - [IT Help Desk Agent](https://ibl.ai/solutions/enterprise/agent/it-help-desk-agent) ## Where does the course material come from? Every module is grounded in primary sources — the regulation, standard, or research itself, not a summary of it. Each was resolved at authoring time. - [AI Index Report](https://hai.stanford.edu/ai-index) — Stanford HAI. Inference cost and compute trend data underpinning token modeling. - [Ideas Made to Matter](https://mitsloan.mit.edu/ideas-made-to-matter) — MIT Sloan. Research on enterprise AI adoption economics and pilot outcomes. - [Transformers documentation](https://huggingface.co/docs/transformers/index) — Hugging Face. Technical reference for self-hosting requirements in Module 5. - [Embeddings guide](https://platform.openai.com/docs/guides/embeddings) — OpenAI. Token accounting mechanics for the bottom-up model. ## Delivery notes Binding guidance for anyone preparing and delivering this course: - Ship the spreadsheet with working formulas as the primary artifact. It is the most-reused deliverable in the enterprise catalog. - Module 5 must state honestly when self-hosting loses — low volume, no existing ops capability, spiky workloads. A model that always favors self-hosting will be discounted entirely by finance. - Date every price used. AI pricing moves quarterly and an undated model looks stale within months. - The hostile-committee role-play needs someone with real finance experience. A sympathetic facilitator produces a useless rehearsal. - Keep the per-seat framing consistent with the rest of the ibl.ai corpus, and cross-link the existing cost calculators rather than rebuilding them. ## Why run AI training on a platform you own? - **You own the course, not a licence to it.** Course content, learner data, and the platform run inside your perimeter — you own all the code and the data. - **Model-agnostic delivery.** Run the course's AI components on any LLM — Claude, GPT, Llama, Gemini, Command — and switch anytime. - **No per-seat training licences.** Usage-based or self-hosted, so cost tracks actual use rather than headcount. - **Deploy anywhere.** Cloud, private VPC, on-premise, or fully air-gapped — including for cohorts that cannot use public AI tools. ## Frequently asked questions ### What does the The Enterprise AI Cost Model: Per-Seat vs Token vs Owned course cover? Per-seat AI licensing scales with headcount regardless of use, so an organization above a few thousand employees pays many times over for a workload it could meter directly. This course builds the three-scenario model on the learner's own numbers, including the utilization problem, the hidden line items, and an honest account of when self-hosting loses. It runs 5.5 hours across 8 modules across 8 modules, at foundational level, and closes with a capstone: Three-year enterprise AI TCO model. ### Who should take The Enterprise AI Cost Model: Per-Seat vs Token vs Owned? It is written for CFOs, finance business partners, and FP&A staff, CIOs and IT finance leads, Procurement and vendor management, Heads of AI and transformation programs. Prerequisites: Bring headcount figures and any current AI vendor quotes; Comfort with a spreadsheet. ### Can we run this course on our own infrastructure? Yes. ibl.ai is model-agnostic and deploy-anywhere — cloud, private VPC, on-premise, or fully air-gapped — and you own all the code and the data. Cohort data, submissions, and any material learners upload stay inside your perimeter, which matters for enterprise teams that cannot send work to a public AI tool. ### How do we get access to The Enterprise AI Cost Model: Per-Seat vs Token vs Owned? Request access and we will set it up for your cohort — hosted by ibl.ai, or running against your own deployment. Tell us the group size and timing you need, and whether it should run inside your own perimeter. ### How much does AI training for enterprise cost on ibl.ai? There is no per-seat pricing — you pay for usage or self-host and pay only for the infrastructure, so a 5,000-person rollout does not cost 5,000 licences. 1.6M+ users across 400+ organizations run the platform this way, including NVIDIA, MIT, and Syracuse University. ## More Enterprise courses - [Agentic AI for the Enterprise: From Chatbot to Workforce](https://ibl.ai/solutions/enterprise/course/agentic-ai-for-the-enterprise): What separates an agent from a chatbot — tools, memory, autonomy — and the orchestration patterns that let agents finish multi-step work without supervision. - [RAG on Enterprise Knowledge: Architecture, Chunking, Evals](https://ibl.ai/solutions/enterprise/course/rag-on-enterprise-knowledge): Production retrieval over enterprise content — chunking strategy, hybrid search, permission-aware retrieval, and the eval harness that proves it works. - [AI Security: The OWASP LLM Top 10 in Production](https://ibl.ai/solutions/enterprise/course/ai-security-owasp-llm-top-10): Securing deployed LLM systems — prompt injection, data leakage, supply chain, and excessive agency — with the controls and tests for each. - [AI Governance in Practice: NIST AI RMF, ISO 42001, EU AI Act](https://ibl.ai/solutions/enterprise/course/ai-governance-nist-iso-eu-ai-act): Operationalize three overlapping frameworks into one governance program — inventory, risk classification, controls, and the evidence auditors ask for. - [Building an LLM Eval Harness That Ships](https://ibl.ai/solutions/enterprise/course/llm-eval-harness-that-ships): Move from vibes to measurement — task-specific eval design, LLM-as-judge and its limits, regression gates, and production monitoring. - [Model Context Protocol: Connecting Agents to Enterprise Systems](https://ibl.ai/solutions/enterprise/course/model-context-protocol-enterprise): MCP as the integration layer for enterprise agents — server design, authentication, authorization, and exposing internal systems without exposing them to everyone.