What if the infrastructure protecting you from scale is the reason you can't afford to reach it?
Basecamp's founder publicly abandoned the cloud after calculating they'd spent $3.2 million annually on AWS. Their discovery? Actual usage didn't match what they were paying for. Today we sit down with Vidar Hokstad, founder of Hokstad Consulting, who's spent twenty-five years helping startups stop overengineering their infrastructure. Vidar reveals the uncomfortable math behind startup cloud spending: sixty to seventy percent is pure waste. We explore why engineering teams build for Google's traffic while sales teams are still making cold calls, and how the communication breakdown between pitch decks and development teams creates expensive solutions nobody needs.
The pattern starts innocently. A founder presents realistic market projections to VCs, estimating perhaps five percent market capture over three years. That vision passes to the CTO, who tells the head of development they need to build for scale. The development lead tells engineers to future-proof everything. By the time code ships, the system handles ten times the demand the market could ever produce. Nobody intended to waste money. The business context simply never reached the people making architecture decisions.
AWS made a genius move when they gave engineers direct API access to spin up servers. No purchase orders. No CFO approval. No budget committees. The people who love building systems and care little about costs now control the spending. Finance discovers the damage when invoices arrive. The fix isn't tighter controls. It's giving technical teams the business context they're missing. Engineers make different decisions when they understand the constraints.
Twitter's early architecture was famously fragile. The fail whale became a cultural icon. But that crappy system let them iterate fast enough to find product-market fit. When demand finally spiked, they ran two engineering teams in parallel: one keeping the broken system alive, one building the replacement. They survived because they hadn't spent their early capital on infrastructure they didn't need. The startup that builds bulletproof architecture before finding customers often runs out of money proving their servers can handle traffic that never comes.
Before approving any significant infrastructure spend, answer three questions: What's your realistic user projection for the funding period? How fast would demand need to grow to overwhelm a simple solution? What's the cost of being wrong in each direction? If you're growing ten percent month-over-month, your infrastructure can grow with you. That growth rate doesn't require multi-region redundancy or auto-scaling algorithms. It requires watching your metrics and adding capacity when you need it.
Watch the Full Episode on Cloud Infrastructure with expert Vidar Hokstad below:
Follow us to watch live on YouTube and LinkedIn or listen to episodes on Apple Podcasts and Spotify.