AI infrastructure commitments you can defend to finance.
I’m Swetal Patel, a former Sr. Manager of Cloud Capacity at Google. I help companies spending $50K+ a month on AI infrastructure forecast demand, plan capacity and commit with confidence.
Free · about 2 minutes · instant score
Served Google Cloud’s largest customers, including Anthropic, Meta and Apple.
AI infrastructure is volatile and expensive. Planning is the lever.
Demand swings, prices move and hardware turns over faster than contracts do. Planning for new models, services and hardware platforms, and for efficiency gains, takes an end-to-end view of the investment and a strategy that expects accelerating innovation. ROI follows sound planning and forecasting.
Volatile demand
One launch or one agent workflow can multiply usage overnight. A plan built for steady growth breaks first.
Fast-moving supply
New models, services and hardware platforms change cost and performance every quarter, and today’s commitment has to survive them.
No end-to-end view
Engineering, finance and procurement each hold a piece. Nobody owns the whole investment, so nobody owns the ROI.
How we work together
Start free, get specific, then keep the plan current. Each step stands on its own.
FinOps assessment
Score your organization across 10 capabilities and see where your planning and forecasting gaps are.
- Instant maturity score and tier
- Level 1 vs. Level 10 benchmarks
- No call required
AI capacity & forecast review
Your usage, contracts and roadmap turned into a plan you can take to finance.
- Demand forecast with base, high and shock scenarios
- End-to-end investment view across providers and hardware
- Commitment recommendation: how much, which, and when
- ROI and efficiency model for each option
- Walkthrough video and live readout
Advisory
Keep the forecast current as demand, pricing and hardware change, with a senior sounding board for each big decision.
- Quarterly forecast refresh
- Review of every major commitment before you sign
- Monthly working session and async access
Who’s behind this?
I spent 12 years at Google working on cloud capacity: planning, buying and managing the infrastructure behind products at global scale. Now I help other companies do the same for AI.
Read the full storyWho this is for
Engineering, finance and FinOps leaders at companies spending $50K or more a month on AI and cloud infrastructure, whether you build AI products or buy them.
See if it’s a fitThe AI Capacity Brief
My working notes on AI infrastructure economics. Read and watch them before you ever talk to me. New episodes go out on YouTube.
AI capex and capacity
Where the AI buildout is heading, what it means for data center, GPU and power supply, and how it reaches your bill.
Pricing and commitments
How AI and cloud providers price capacity, what reserved and committed deals really cost, and where the traps are.
Forecasting and unit economics
Demand modeling, cost per token and cost per outcome: the numbers that decide whether AI spend pays off.
FinOps practice
What mature FinOps teams do differently, drawn from practitioner talks and my own work.
Get the brief by email
New analysis on AI capacity, pricing and forecasting, when there is something worth saying.