One developer + AI vs a team of 10: the math
Ten mid-level devs cost $1.8M a year and ship in quarters. One senior vibe coder costs $300K and ships in days. Here is the spreadsheet
Ten mid-level developers at $180K each is $1.8M in salary. Add benefits, tooling and management overhead and you're at $2.5M per year. That team ships in quarters
The senior vibe coder alternative
A senior engineer with Claude Code, Cursor and ten years of taste. Fully loaded for $300K. Produces the same feature set in a fraction of the time because no stand-ups, coordination costs or design-by-commission are required
The 74% productivity number
According to published studies, the average productivity gain of AI-assisted seniors is 74%. The top 10% report 10x. A senior vibe coder does not do the work of one developer, he does the work of three to 10 developers
The coordination tax
A team of ten has 45 communication pairs. A team of one has zero. Brooks' Law is still the most expensive unlearned lesson in software. AI has not revoked it, it has made it worse because fast humans are now 10x faster and coordination is still at human speed
The hidden costs of ten
$1.8 million in salary is the headline. The real bill is bigger. Ten people need a manager. The manager needs a director. The director needs an HR partner. The HR partner needs a recruiter. Each new layer adds 15% in coordination costs and zero in shipped positions
Then you have the office, the laptops, the SaaS chairs per head, the offsites, the all-hands, the 1:1 calendars. Conservative estimate, $700K in overhead on top of salary. That's the $2.5 million figure
The hidden costs of one
Almost zero. One Cursor chair. One Claude Code subscription. One laptop. No stand-ups, no Jira tickets discussed for an hour, no draft-by-committee, no "let's align first." The senior decides, the senior dispatches, the senior moves on
That's not a productivity increase. That's a category change
The 81-day delta
81 days to fill one of those 10 spots. To staff the full team requires six to nine months of recruiting, and 83.9% of IT projects fail or overrun. Meanwhile, a senior vibe coder is available as early as Friday of week one
Time is the commodity that costs no one. Compute the numbers, including time to first delivery, and the team of ten not only loses, it never starts
The math
$2.5 million team, 12 positions per year = $208K per position
$300K vibe coder, 20 positions per year = $15K per position
The gap is not small. It's not even in the same order of magnitude
The objection you're about to make
"But one person is a single point of failure." Sure. So is your CTO, your founder, your AWS account. Single points of failure exist everywhere, the question is whether the output justifies the risk. At 10x the throughput and 1/8 the cost, the math says yes
Bus factor is important at 100 engineers, not in the seed-to-Series-A phase where you're trying to find a product-market fit before the runway is up. Ship first, scale the organization later
When you actually need ten
Real answer, almost never until you get past $10M ARR. Before then, every extra engineer is a burden on those who are really going to deliver. Look at the YC W25 group, 25% have 95% AI-generated codebases with teams of two or three. They are not waiting for a recruitment round