Founder Fridays No. 199
Orbit Is The Product -- Your Vendor Controls You -- Big Teams Make You Dumb
Happy Friday.
Compute in Orbit
SpaceX’s biggest revenue line by 2030 won’t be rockets or satellites — it’ll be selling AI compute time from data centers in orbit, already under contract to Anthropic ($1.25B/month) and Google ($920M/month), from a business that didn’t exist before February. The energy bottleneck will kill the AI buildout before the chip shortage does — terrestrial solar requires permits that take years and land that’s politically radioactive, while orbital solar delivers 4–10x more power with no grid, no night cycle, and no NIMBYs; whoever controls cheap access to orbit controls the AI power supply. Run the “idiot index” on one critical cost line in your business this week: find the ratio between what you pay for something and the raw cost of its inputs, then ask who you’d have to fire or replace to close that gap — SpaceX cut launch costs 99% by refusing to accept supplier math. If you identify even one process where you’re paying a 10x markup because you accepted the first quote, you’ve found the wedge that could double a margin line — and the habit of questioning every requirement is what separates companies that hit escape velocity from those that optimize their way to irrelevance. A16Z (26 minutes)
Big Teams Make You Dumb
A programmer who stops being allowed to ship new ideas doesn’t just get frustrated — they stop having ideas at all, meaning every large org is running a slow lobotomy on its best people without anyone noticing it’s happening. The tree structure that every large company is forced to adopt means individual freedom shrinks in direct proportion to headcount — not because of bad managers or bad culture, but because the math of coordination requires it, so the only way to protect creative output is to stay small by design. Before your next hire, ask whether adding this person solves a problem or just compensates for a process you haven’t killed yet — Graham’s data is blunt: mediocre hires hurt you twice, once in output and once by making you bigger, which slows everyone else down. Founders who treat headcount as a last resort rather than a growth signal keep the small-team physics — fast decisions, full context, no tree — that are the only structural advantage a startup permanently holds over a well-funded incumbent. Paul Graham (8 minutes)
You’re Fixing The Wrong Leak
Nine out of ten founders who think they have a conversion problem actually have a demand problem — meaning every hour you spend tweaking your pitch, your pricing, or your close rate is covering for the fact that you’re simply not talking to enough people in the first place. Going from a 10% to 15% close rate is a 50% improvement that requires rebuilding your sales motion; going from 10 demos a month to 20 just requires more outbound — and right now AI can do the prospecting, scoring, sequencing, and intent-signal crawling that used to eat most of a sales team’s week, which means the bottleneck on demand is no longer labor, it’s whether you’ve pointed the system at the right ICP. Pull last month’s numbers and ask one question: did you miss because deals didn’t close, or because you didn’t have enough deals to begin with — if it’s the former, pick one intent signal (a job posting, a funding announcement, a blog post about an offsite) and rewrite your outbound sequence to open with that signal rather than a generic cold pitch this week. Founders who fix their demand diagnosis before hiring their first sales rep stop the most expensive mistake in early GTM — bringing in a salesperson to solve a product-market fit problem — and instead build a demand-rich pipeline where conversion improvements actually compound on something real. The Split (45 minutes)
Founder FAQ: Why Proper Documentation is Important for a Startup?
The moment a founder scrambles to find a signed employment agreement or a board consent during due diligence is the moment an investor starts discounting the valuation — your document chaos isn’t an admin problem, it’s a pricing problem. Investors and acquirers don’t just want clean docs — they’re using your document hygiene as a signal for how you run everything else; a disorganized data room says “this team cuts corners,” and that inference bleeds into every number on your cap table. Spend 90 minutes this week setting up a shared drive with ten folders — incorporation, board records, equity, executed agreements, templates, cap table, compliance, financials, IP, insurance — and drop every document you already have into the right place, even if some folders stay empty for now. When your next investor asks for a data room, you send a link in five minutes instead of spending three weeks in triage — and that composure alone signals the kind of operator people wire money to. Westaway (5 minutes)
Startup Funding Guides
I’ve put together a series of guides to equip founders to excel at fundraising. These guides break down the deal term-by-term and give you negotiation tips so that you can speak to investors with confidence.
Convertible Note: Guide / Video
Saving Time and Money on Legal
When we met this Series B startup, they were frustrated with their law firm’s slow turnaround and high fees. Contract reviews took four to six weeks, and they charged $250,000 annually for basic work. The startup wanted to reduce sales cycle times and legal spend. They switched to General Counsel at Westaway. In year one, we 1) saved them about $200,000 in legal fees; 2) shortened their sales cycle by about four weeks; and 3) our streamlined processes saved their ops team eight to ten hours per month previously spent managing legal. By switching to Westaway, they expedited deal closures, saved hundreds of thousands in legal bills and regained one day per month in productivity. If you’re curious if we could save you time and money, let’s talk.
The "big teams make you dumb" point tracks with what I've seen building lean AI teams: past a certain headcount, the coordination tax eats the AI productivity gains. I've watched a 4-person team ship faster than a 15-person team at a similar-stage company because the smaller group could actually absorb and act on model output without three layers of review. There's a real ceiling where more people means more consensus-seeking, which is the opposite of what fast iteration with AI tools requires. Curious if you think that ceiling is a fixed headcount or scales with how AI-native the team already is.