TechCrunch · June 5, 2026
One of my engineers spent $40,000 on tokens last month, and I genuinely don't know whether I should stop him or should I go and tell everyone else to be like him.
Vitaly GordonCEO, Faros AI · TechCrunch ↗
AI value management · The year the AI bill came due
Your AI budget: spent. What did it buy?
When Uber exhausted its 2026 AI budget by April and its COO was asked what those tokens shipped, he answered: "that link is not there yet" (Fortune). COGScontrol draws that link — cost per interaction, per customer, per feature, measured against the business metrics that prove value — reconciled to your P&L.
Unified visibility across your inference supply chain
five LLM providers · three cloud providers · one ledger
OpenAI
Anthropic
AWS Bedrock
Azure OpenAI
Vertex AI
AWS
Google Cloud
Microsoft Azure
The public record
The value question
The value question
Everyone can see the bill. Almost no one can say what it bought.
In a single 2026 news cycle, Fortune, TechCrunch, Constellation Research, and Tom's Hardware converged on the same finding: AI spend is enormous, visible, and disconnected from any measure of value. What follows is the public record. None of it is testimonial; all of it is the question we build for.
Fortune · May 26, 2026
Uber · AI coding tools
Uber · AI coding tools
If you're not actually able to draw a direct line to how [many] useful features and functionality you're shipping to your users, that trade becomes harder to justify.— Andrew Macdonald, COO, Uber — after Uber exhausted its entire 2026 AI coding-tools budget by April · Fortune ↗
TechCrunch · June 5, 2026
Now the conversations are about, 'hey, we're spending so much. What visibility do you have?'
Alexander EmbiricosHead of Enterprise, OpenAI · TechCrunch ↗
TechCrunch · June 5, 2026
Token economics is fundamentally more abstract and opaque than anything we've managed at this scale before.
Nishant GuptaChief Availability Officer, Salesforce · TechCrunch ↗
Constellation Research · 2026 · the counterexample
AI reasoning is less than 10% of our cost to serve.
ServiceNow's CFO can answer the value question — because she measured it.
Gina MastantuonoCFO, ServiceNow · Constellation ↗
Constellation Research · 2026
A lot of customers are getting a little bit surprised on the tokenization of models and how that is surprising their budget landscape.
Bill McDermottCEO, ServiceNow · Constellation ↗
TechCrunch · June 5, 2026
It's like the crack-cocaine epidemic. They let you try it to get you hooked on it, and now you're kind of beholden to it.
Priceline saw a 4–5× cost increase at its Cursor renewal — with no shared measure of what the spend returns.
Chris ReedSenior Director of IT Finance, Priceline · TechCrunch ↗
Agentic token demand
24×
Goldman Sachs projection
Projected multiple on token use as agents replace chat — spend that scales faster than anyone's measure of what it returns.
Agent vs. chatbot tokens
1,000×
Tom's Hardware
One agentic workload can consume up to a thousand times the tokens of a single chat exchange. The denominator matters more than ever.
One engineer · one month
$40k
Value unknown
Spend that can't be judged — problem or pioneer? — because no value denominator exists. The cost is known to the cent.
Reported surprise bill
$500M
TechCrunch · reported
One company reportedly faced a nine-figure AI bill after usage limits were left off. Nobody could say what it was worth.
Quotes are news citations, not endorsements. No person or company above is a COGScontrol customer.
Why COGScontrol
The financial infrastructure to measure AI ROI — and scale profitably.
Traditional cloud cost tools weren't designed for AI workloads. COGScontrol gives you the financial intelligence required to manage unit economics across your entire inference supply chain — from token to P&L line item, with the resolution finance expects.
The discipline
A definition
A definition
What is AI Value Management?
AI Value Management is the financial discipline of measuring what AI spending actually returns. It attributes every AI and cloud cost to the initiative that incurred it, measures those costs against business metrics — revenue, users, transactions — and reconciles the result to the P&L as unit economics a CFO can defend.
FinOps and cost tools tell you what you spent. AI Value Management tells you what it bought — cost per interaction, cost per customer, AI gross margin — so leadership can decide where to scale and where to stop. COGScontrol is the platform built for that discipline: measure AI ROI and unit economics on fixed-price plans that never take a percentage of your AI spend.
Platform
The instrument panel
The instrument panel
Six primitives that turn spend into signal.
Inference supply chain visibility
Unified cost data across OpenAI, Anthropic, AWS Bedrock, Azure OpenAI, and Vertex AI. One view of your fully-loaded AI COGS.
Core
Intelligent classification
Automatically categorize costs by Cost Center, Product Line, Project, and more. Create rules that classify spending automatically.
Core
Dimension-based budgets
Set budgets on any dimension — by team, product, project, or environment. Get real-time alerts before you exceed limits.
Control
Unit economics engine
Map AI costs to revenue, users, and transactions. Calculate cost-per-interaction, cost-per-customer, and AI-attributed gross margin in real time.
Core
Custom dashboards
Build dashboards with 8+ chart types. Share with your team, pin favorites, and set up scheduled reports.
Reporting
Margin leakage protection
Get alerted when unit costs drift outside expected bounds. Catch margin erosion from model changes, prompt bloat, or usage spikes before they hit your P&L.
Intelligence
Workflow
How it works
How it works
From black-box AI spend to board-ready financial intelligence.
Step 01 · Connect
Connect your inference supply chain.
Securely ingest cost data from all your AI and cloud providers. Data syncs automatically every day — providers normalize to one schema and reconcile to invoice.
Step 02 · Classify & Attribute
Map every token to a business outcome.
By product, feature, customer, or team. Our classification engine builds your AI cost of goods sold automatically — rules apply retroactively, with a full audit trail.
Step 03 · Measure value, not just cost
Track unit economics in real time.
Run variance analysis and surface margin leakage before it reaches your P&L. The questions a CFO actually asks — answered with the math attached.
Margin notes
from the ledger
from the ledger
Uber's COO asked for a direct line from tokens spent to value shipped. That line — drawn daily, measured against the metrics that matter, reconciled to the P&L — is the product.
Use cases
The questions
The questions
Asked by leadership.
Answered by COGScontrol.
For the CFO
Our AI spend doubled this quarter. Which half was worth it — and could I defend the answer to the board?
A live value-per-dollar figure rolled up from per-MAU, per-resolution, and per-lead economics. Quarter over quarter. Reconciled to the books. Board-ready in one click.
For the head of product
Which AI features earn their tokens — and which ones can't draw the line to value?
Cost per active user, per feature, per cohort — overlaid with retention and revenue. Pause-or-double-down decisions, with the math.
For the CEO & board
If we 10× usage tomorrow, do unit economics improve, hold, or collapse?
Goldman Sachs expects agentic AI to multiply token demand 24×. Forecasted value-per-dollar at 1×, 2×, 10× current load — with and without model price changes — so scale is a decision, not a surprise.
Don't build the dashboard
And now you don't build the dashboard.
You ask the question.
Type it in plain English. COGScontrol answers with the figures, the attribution, and the query it ran — every response reconciled to invoice and fully auditable. Board-ready by default.
›
Did moving support to Haiku actually lower our cost per resolution last month, or did volume just drop?
AnswerReconciled to invoice · 3 sources
Yes — cost per resolution fell 22%, from $0.40 to $0.31, and it's a genuine efficiency gain, not a volume artifact. Resolutions rose 11% over the same window, so unit cost dropped while throughput grew.
$ / resolution
$0.31▼ 22%
Resolutions
48,210▲ 11%
Support spend
$14,945▼ 13%
Model mix · Haiku
71%▲ 54pt
Attribution path
Anthropic · Haiku
→
Support · resolution events
→
Zendesk · solved tickets
→
$ / resolution
Query runSQL · read-only · 0.4s
select date_trunc('month', ts) as mo, sum(cost) / nullif(count(*) filter (where resolved), 0) as cost_per_resolution, count(*) filter (where resolved) as resolutions from inference_events join support_resolutions using (trace_id) where ts >= now() - interval '60 days' group by mo order by mo;
Full audit trail · every figure traces to a line item
FAQ
Common questions
Common questions
Asked before the first sync.
What is COGScontrol?
COGScontrol is an AI Value Management platform for finance professionals at AI-first companies. It aggregates AI and cloud costs across OpenAI, Anthropic, AWS Bedrock, Azure OpenAI, Vertex AI, and your cloud providers, attributes them to products, teams, and initiatives, and measures them against business metrics — so you can see the ROI of every AI initiative, reconciled to the P&L.
What is AI Value Management?
AI Value Management is the financial discipline of measuring what AI spending returns: attribute every AI and cloud cost to an initiative, measure it against business metrics like revenue and active users, and reconcile the result to the P&L as unit economics. Cost tools stop at what you spent; AI Value Management answers what it bought.
How does COGScontrol measure AI ROI?
Three steps, refreshed daily. First, ingest and reconcile spend from every AI and cloud provider. Second, attribute each dollar to a product, feature, customer, or team with rule-based classification. Third, pair attributed costs with the business metrics you import — revenue, MAUs, transactions, resolved tickets — to compute cost per interaction, cost per customer, and AI-attributed gross margin. The result is an ROI figure that traces to invoice line items. See the full feature set.
Which AI providers does COGScontrol support?
Five LLM providers — OpenAI, Anthropic, AWS Bedrock, Azure OpenAI, and Google Vertex AI — plus AWS, Google Cloud, and Microsoft Azure for the cloud infrastructure behind your AI workloads. Provider data is normalized into one ledger and reconciled to invoice every 24 hours.
Is COGScontrol a FinOps or cost-management tool?
No. FinOps and cost-management tools tell you what you spent and help you spend less. COGScontrol measures what the spending returned — unit economics, margin, and ROI per initiative. The two are complementary; only one answers the question a board asks.
How much does COGScontrol cost?
Four fixed-price tiers: Pulse (free forever), Track ($399/mo), Scale ($1,499/mo), and Custom for enterprise — every one a fixed subscription, never a percentage of your AI spend. See pricing.
Get started in minutes · 14-day free trial · no credit card
Ready to measure the value of your AI investment?
Know what your tokens bought — before the board asks. COGScontrol gives you the financial intelligence to scale AI profitably.