Building a blog platform cost $4 million in 2005 and $200 in 2025, and Garry Tan has done it three times

Garry Tan has built a full-featured blog platform three times. The first time cost roughly $4 million, required six or seven people, and took around a year and a half. The second time cost around $100,000, required two people, and took about three months. The third time cost $200 and took five days.

That compression, across three distinct builds of what Tan describes as the same class of product, is not an incremental efficiency story. It is a change in the order of magnitude of what a single person with the right tool can attempt. The $200 figure came from Tan’s Claude Code Max account. No team. No extended runway. No institutional capital standing behind the effort.

The comparison carries weight precisely because it is longitudinal. A single before-and-after data point can always be explained away by better tooling, a simpler scope, or accumulated domain knowledge. Three data points across what appears to be roughly two decades of practice are harder to dismiss. Tan is not comparing his current work to a hypothetical earlier version built by someone else, or to an industry average. He is comparing it to his own prior work, under his own direction, with his own co-founders.

The first time it took about, you know, $4 million and, you know, six or seven people and about a year and a half. And then the second time it, you know, took about, I don't know, 100 grand and two people, me and my co-founder Brett Gibson, who now runs Initialized. Um, and maybe like 3 months or so. And then in this case it took about $200, which was my Claude code Max account, and probably 5 days. Garry Tan

The second build is worth pausing on. Tan and his co-founder Brett Gibson, who now leads Initialized Capital, brought that project in at around $100,000 with a two-person team over roughly three months. That was already a dramatic reduction from the first attempt. The move from $100,000 to $200 is a further reduction of roughly 99.8 percent on top of a baseline that was itself already a fraction of the original. Each successive build compresses cost not just in absolute terms but relative to an already-compressed prior round.

What Tan’s account does not supply is a detailed breakdown of what changed between each attempt. Some portion of the gap almost certainly reflects accumulated knowledge of the problem domain, scope decisions made more tightly the third time, and tools that simply did not exist during the earlier builds. A single speaker’s retrospective comparison cannot control for all of those variables, and it would be wrong to read the $200 figure as a universal benchmark for blog platform development. The conditions Tan brought to that fifth day, including years of having built the product before, are not replicable on demand.

Even with those caveats, the directional claim is striking. The cost floor for serious software development has dropped to a point where the binding constraint is no longer capital or headcount. What once required institutional funding and a small team now fits inside a monthly subscription. That shift changes who can start, what they can attempt, and how long they need to sustain a burn rate before having something to show. Whether the pattern holds across more complex product categories, or whether it reflects something specific to Tan’s accumulated expertise and the nature of this particular build, is a question the evidence here cannot answer. What it can support is the narrower claim: for Tan, the same product that once required millions and a team now required neither.