Max Hodak: How Startups Build Speed

Max Hodak says infrastructure makes startups fast

Max Hodak argues that deep tech startups rarely fail on the technology and mostly fail on organization. Speed is the difference between success and failure, and speed comes from infrastructure such as purchasing, hiring, and internal software. Science built its own tools because no vendor sells what a fast company needs. Hodak says that action produces information.

Takeaways

• Deep tech startups fail on organization. Hodak says it is uncommon for startups to fail because they cannot get the technology to work. It is more common that they cannot organize the human organization to accomplish their goals. Learning to organize is a skill set learned by doing.
• Speed is the basis of startup success. Rate of iteration separates success from failure, and speed is determined by infrastructure. Purchasing and recruiting determine how fast a company can act, and internal software does too. Hodak says there are no general principles beyond this.
• Procurement breaks as headcount grows. Founders can use credit cards, but the seventeenth employee cannot easily buy things. A $3,000 power supply or a two-week procurement delay can stall work. Spending review must come earlier through budgeting, because attribution is hard when buying materials in bulk.
• Science built Helix to make buying fast. Almost everything in Helix is a button. Hodak says you cannot buy these infrastructure things. ERP systems like NetSuite exist, but no company loves their ERP system. A single wafer iteration costs $40,000. Science also built Warp Speed.
• Eigenreviews beat annual 360 reviews. Science has used eigenreviews for six or seven years. They ping employees every four to six weeks and use a PageRank-like graph with a thousand dropout iterations, with about a month lag. Hodak criticizes annual 360 reviews.
• Raise twice the experiment cost. A $20m Series A might fund four years and 20 people. About half the burn is headcount. The company pays for 20,000 square feet at $4 per square foot and a $3,000,000-a-year research budget. Keep waste to 20 to 30 percent and push for revenue.

Science's hiring process

1. Company-wide voting comes first. The system picks seven or eight current employees with similar backgrounds and pings them for votes, usually within twenty-four to forty-eight hours.
2. A phone screen looks for judgment and horsepower. It also tests for agency.
3. Homework is scorable to two or three numbers.
4. A full interview with at least 25% conversion to offer.