round.
Livingston: They put you down to other VCs?
Bhatia: They did. Of course, that was very early on and now everything is all
fine and dandy, but at that point in time . . . we had a term sheet for a much
higher valuation. But when we would talk to any other VC, the other VC would
call the guys at DFJ and they’d say, “No, don’t invest in them.”
Livingston: Were they helpful at all?
Bhatia: Yes. Steve Jurvetson was very helpful; he introduced us to a lot of people
and, on the whole, they’re a good VC firm in the sense that they try to put
deals together. But sometimes they don’t play by the rules.
Nobody knows this, but the round before the deal with Microsoft, they literally
put $5 million in the company just because they knew it was going to get
sold and that we needed some bridge money. This came at a very expensive valuation
with certain rights that should not have come with it—like participating
preferred, which is they first get their money out and then they participate in
the rest, which was OK for the earlier rounds, but not for the later ones. That
was just bridge money that we needed while we were negotiating with
Microsoft. They knew full well that we were going to get acquired; we were
negotiating about the final price.
Livingston: I’ll come back to the Microsoft negotiation in a moment. Did your
background in hardware help you in terms of building servers that could handle
massive loads?
Bhatia: It helped us because we knew what kind of hardware we would need to
be able to handle the kind of traffic to our site. Also, when you are hardware
designers, you have tremendously more discipline in writing and describing
software because in hardware you cannot get it wrong. Every turn of every chip
costs you millions of dollars, so when hardware designers design any piece of
software, they normally get it right. They use something called state machines
to describe the functioning of the software. When you do that, you are very
deterministic: if this is the input, then this will be the output.
So you write it in a very deterministic fashion and therefore you tend not to
make too many mistakes. Whereas the pure software writers—the way they
think and architect software is very creative. They put in lots of bells and whistles,
but they think, “No big deal. If there is a bug, we’ll fix it. Put in a patch.”
You can’t do that in hardware. There’s no patch. Once you ship a chip, it has to
work all the time. So in terms of being able to test it out, there is somewhat of a

Похожие записи:

• round.