Startups Then vs Now

Joe writes about the great expenses of hardware and software that were a part of the technology startup experience ten years ago. For Coconut, my first startup, founded in 1987, this was painfully true. I spent several months evaluating different hardware and software platforms for the server software: what was the cheapest, fastest, and most popular platform out there to develop for? Once I decided the server software had to be Unix based, the decision became, which hardware and which Unix? By 1988 it was time to start coding, and I decided the hardware would be x86-based, and the software we'd first support would be SCO Xenix (remember SCO, back when they were the good SCO?).

We needed a server and we bought one. At the time it was considered fast, relatively affordable, and a great platform for SCO Xenix: the AST Premium 386. It cost $10,000 and came with 4 megs of RAM (yes, megabytes), and -- gasp -- a 90-megabyte hard disk. The CPU? A 20-mHz 386. I remember splurging on a color graphics card for the machine (can't remember if VGA was out yet) and a Zenith "Flat-Panel" monitor (flat CRT). I remember taking the Zenith out of its box, putting it on top of the AST, plugging it in, turning it on, and watching a small mushroom cloud apear above the monitor as it blew up.

We soon needed more disk space and more RAM. We could only afford 3 more megabytes of RAM -- because that alone cost us around $1200. And then there was the disk drive. We splurged on a $2300 CDC WREN V drive with a massive capacity of ... 340 megabytes. Whoooo-ooo!

Then there was the expense of the SCO Xenix license. Then the other desktops and all of their Microsoft, Borland, and other software licenses. Then the licenses for other Unix platforms (and the costs of buying hardware for all those other platforms) that our customers wanted our software to run on. I remember considering joining the Go developer's program (remember Go? the handheld "Slate" computer?). Then I saw they wanted $7000 just to join their program.

Being a software company in the late 80s and early 90s was really an expensive proposition.

As Joe points out, today, development tools now cost nothing. And commodity hardware is dirt cheap. It's remarkable how things have changed.

Joe also brings up the global market for labor, and the fact that even the smallest startups now have access to it. I can see the benefits of outsourcing some specific development projects, but I don't see the benefits of outsourcing the whole development effort. Maybe I'm old-fashioned: I care about building a company culture as well as great products. I've not arrived at the point where I could be comfortable with a company where everything is virtual: everyone works in different regions of the world, and communicates only through phone, IRC, AIM, email, wikis, and other electronic means. That's fine for project matrix teams that form and break up and re-form in different groups for the next projects, but where's the fun in that for a company?

Posted by brian at September 20, 2004 10:40 AM