Case Study
GPU Compute Startup
Pre-revenue compute company with technical depth and no go-to-market. I redesigned the entire business model, built a 4-stage market penetration strategy, and architected cash flow before a single customer signed. Then walked away when the fit wasn't right.
February 2026 · Engagement ended amicably
The Situation
The founder had deep infrastructure knowledge and a clear technical thesis: resell managed GPU compute to companies locked out of hyperscaler pricing. He could build the stack. What he couldn't see was the path from "I have compute" to "I have a business."
When I came in, the positioning was scattered. Half B2C ("the cooler AWS for young people"), half B2B, with no clear first customer. Website looked AI-generated. No pricing validation. No discovery calls. No understanding of payment terms with his own supplier. He was about to spend his last money buying infrastructure before talking to a single prospect.
What I Did
Killed the B2C Angle
Repositioned the entire business from vague "cool AWS for Gen Z" to a specific, defensible position: B2B infrastructure for emerging markets, starting with Nigeria. One country. One value proposition. Economic nationalism as the wedge: "Your data stays in Nigeria."
4-Stage Market Penetration
Designed a strategy where each stage funds the next. No outside capital needed at any point.
Stage 1. Resell compute at price parity. Win on nationalism, not price. Validate with 10 paying customers from warm network.
Stage 2. Managed deployment layer. "I push code and it works." 3-5x markup over raw compute. This is where real margin lives.
Stage 3. Identity layer. Emerges naturally from 30-50 companies authenticating through your stack. You don't build KYC. You observe the pattern and unify it. The identity graph becomes the moat.
Stage 4. App marketplace. Identity and payments solved. Third-party developers build on the stack because the hard problems are done. Transaction fees on every API call.
Cash Flow Architecture
The founder was about to pay his infrastructure provider upfront. I redesigned the model: charge customers on the 1st, pay the supplier at end of month. Positive float from day one. Then reframed the supplier relationship entirely. The supplier needs demand generators. We are their next market. Negotiate from that position, not as a small buyer begging for credit.
GTM Before Build
Stopped him from buying infrastructure before talking to a single prospect. Designed a discovery call sequence using his existing network. No cold outreach. Every call ends with "Who should I talk to next?" so the pipeline generates itself. The Mom Test applied: learn what they need, don't pitch what you have.
Operating System
Full EOS implementation: vision, core values, quarterly rocks, weekly accountability. Plus a branded proposal system, consulting agreement, and interactive execution documents. One URL, everything in it.
The Decision
After the alignment call, it became clear the engagement wouldn't work. Not because of capability on either side, but because of timing and priorities. The founder needed to resolve upstream questions before a strategic consulting layer would deliver value.
I told him directly. We parted on good terms. He wrote a LinkedIn recommendation the same day.
Walking away from revenue when you're pre-revenue yourself is not comfortable. But taking on an engagement where the prerequisites aren't met burns both parties. The consulting only works when the founder is ready to execute on the output.
What This Shows
Business model design, not advice
Repositioned the company, designed the revenue architecture, and sequenced 4 stages of market penetration in a single session.
Supplier as partner, not vendor
Reframed the infrastructure provider relationship from 'small buyer' to 'demand generator for your next market.' Changes the entire negotiation dynamic.
Revenue before product
Stopped a founder from spending his last money on infrastructure. Designed GTM that validates demand with zero capital deployed.
Honest assessment
The hardest consulting skill is telling a client they're not ready yet. Revenue lost, trust gained.
This engagement did not result in an ongoing contract. It's included here because the work was real, the frameworks were delivered, and the outcome (walking away) is as much a part of strategy as the wins.