I don't know where wileycoyote is getting his/her FACT numbers from, but your speculation on Procket burn rate seems reasonable to me.
However ...
> 1) Nortel has already tried to do it once. It > did not go well.
What does this mean ? "Tried to" ? Did it not go well because they offered too little ?
> 2) Avici's hardware architecture is a generation > behind now. Its old.
What is an older architecture, a 3D toroidal mesh, or a shared memory crossbar ala Cisco, Juniper, and now Procket ? I could understand if you said their implementation was older, however it seems they introduced new hw this year.
Looking at websites (with a grain of salt) their port density looks to be better than Cisco and Juniper, and a factor of 2 off of Procket - the clear leader ... if they are real.
> 4) If a feeding frenzy for core routers starts, > Avici will likely be bought someone. They > could be bought for cash by Huawei for example. > But anyone who has seriously looked at avici > knows there are lots of problems there.
Do you have any details that lead you to this conclusion ?
> 5) Their contract at ATT isn't a long-term money > maker now. Its getting mature. And the ability > of that contract to get Avici into more places > is used up.
Hmm, who exactly has a long term contract with an IXC ? From my experience, they re-evaluate and make decisions on (at longest) a yearly basis.
re: Will Nortel Pick Procket?Procket has also forced a 10 % pay cut on all employes. They wouldn't do that if they had 100M in the bank. They must be worried about thier burn rate. ------------
If the rumors are true and their burn-rate was over 6m a month, they would have had to cut it even with 100M in the bank. At six million a month, a 100M only lasts around 16 months.
re: Will Nortel Pick Procket? Skeptics original posting on the subject assumed a burn rate of a constant (average) $5m over the life of the company. I'd suggest that this is not the case - it takes a while to build to those kind of numbers. The average monthly burn is probably less which, over say 42 months, would mean considerably more cash in the bank ($1m a month less means $42m more in the bank - big swing factor!).
Secondly, just because they had a 10% pay reduction doesn't mean they are running out of money. The rule is "cut early to live longer" - a 10% pay cut now might give them another 12 months of life based on headcount and cash in the bank ... a 10% pay cut when you only have two months money in the bank is meaningless.
Of course - the extra 12 months could be pretty depressing if you were stuck in the groove waiting for the money to run out .... :-)
re: Will Nortel Pick Procket?Skeptics original posting on the subject assumed a burn rate of a constant (average) $5m over the life of the company. I'd suggest that this is not the case - it takes a while to build to those kind of numbers. The average monthly burn is probably less which, over say 42 months, would mean considerably more cash in the bank ($1m a month less means $42m more in the bank - big swing factor!). -------------------- I also assumed:
- They started January 1, 2000. They may have started earlier.
- An average of 5m a month allows for periods when it was either higher or lower than 5m. And the burn-rate has been substantially higher than 5m.
I don't think their burn rate was less than 5m a month in 2001-2003. I guess it could have been in 2001, but it doesn't seem likely.
2000 and before gets difficult to calculate. Its difficult because nobody on the outside knows their hiring ramp from that period or what kind of costs were associated with doing all that custom VLSI.
I would put the swing factor down at 24 million or so rather than 42 million because its only the first year or so where the burn rate would be under 5m.
As a disclaimer (again), this is all an intellectual guessing exercise. If you want to know the real numbers, its always best to talk to the company directly.
re: Will Nortel Pick Procket?> or what kind of costs were associated with doing > all that custom VLSI.
This is fairly easy to estimate. I saw a presentation for 6 IC's all done in 0.13um.
Mask costs in their timeframe would be ~800K per. Additional NRE would probably be on the order of 200K per. EDA tool licenses are probably on the order of $1M per. This would be around $12M if they were really efficient (no respins, no expensive 3rd party IP, etc.).
The cost of protos alone are probably on the order of 300 (minimal run) * 6 devices (if only one instance) * $300 (average price) = ~$500K.
Assuming typical inefficiencies, my bet is if they were good they could have done it all for around ~$20M.
Of course, this dosen't take staffing into account and it's common knowledge they used a large number of consultants on the IC side.
re: Will Nortel Pick Procket?Considering the evolution of Core Router business and multitude of players, I think that the Procket and Nortel deal would be great. Both Nortel and Procket have a good reputation in the industry.
Given the fact that RBOCs seldom buy from start-ups, it is no reflection on Procket. But when Procket and Nortel merge, the market opens for them very widely.
re: Will Nortel Pick Procket?Given the fact that RBOCs seldom buy from start-ups, it is no reflection on Procket. But when Procket and Nortel merge, the market opens for them very widely.
I agree with BobbyMax, however, Nortel would still need some edge router products to build a decent solution for any carrier. I am not sure if any of the WellFleet products are still alive (or not obsolote).
I don't know where wileycoyote is getting his/her
FACT numbers from, but your speculation on Procket
burn rate seems reasonable to me.
However ...
> 1) Nortel has already tried to do it once. It
> did not go well.
What does this mean ?
"Tried to" ? Did it not go well because they
offered too little ?
> 2) Avici's hardware architecture is a generation
> behind now. Its old.
What is an older architecture, a 3D toroidal mesh,
or a shared memory crossbar ala Cisco, Juniper,
and now Procket ?
I could understand if you said their
implementation was older, however it seems they
introduced new hw this year.
Looking at websites
(with a grain of salt) their port density looks
to be better than Cisco and Juniper, and a factor
of 2 off of Procket - the clear leader ... if they
are real.
> 4) If a feeding frenzy for core routers starts,
> Avici will likely be bought someone. They
> could be bought for cash by Huawei for example.
> But anyone who has seriously looked at avici
> knows there are lots of problems there.
Do you have any details that lead you to this
conclusion ?
> 5) Their contract at ATT isn't a long-term money
> maker now. Its getting mature. And the ability
> of that contract to get Avici into more places
> is used up.
Hmm, who exactly has a long term contract with an
IXC ? From my experience, they re-evaluate and
make decisions on (at longest) a yearly basis.