Valley Wonk: Lessons of the 3D Wars

Admit it: You love tech booms. They're the Monday morning NFL sports supplement of the tech industry. You just marvel that so much can be written about something where the fundamentals of the game seem so clear.

Tech booms aren't hard to spot. Wireless LAN chipsets are the obvious case – so obvious that even VCs won't touch them any more. WLAN switches and smart antennas are headed there. Even optical is getting a bit bubbly, with the promise of 10-Gbit/s transceivers and reconfigurable OADMs (see 10-GigE Transponders: Update and ROADMs Could Boost Components).

With the number of times we've gone through the boom/bust cycle, you'd think we'd get better at it. Especially having lived through the inflation and decimation of the Internet bubble. Can't our forefathers teach us something, those souls who braved the rise and fall of mainframes or pocket computing?

At the recent MicroVentures 2003 conference, where the lunchtime speaker was Nvidia Corp. founder Jen-Hsun Huang, I tried to absorb some wisdom at the knees of the greats. Nvidia was one of about 30 hopefuls developing 3D graphics chips in the mid-90s, and it's now considered the market and technology leader, with second place (and survivor status) going to ATI Technologies Inc. (Nasdaq: ATYT; Toronto: ATY).

Nvidia is a tech boom victor. No argument. No BCS computer confusion. So what went right? Could Huang offer some insights into surviving telecom's next bubble?

Huang's wisdom, it turned out, was recycled advice from his Day One investors. Most of it sounds obvious – but then again, so many companies miss the point by several miles. Here are the highlights:

  • Pick a big enough market. Nvidia's chips are gawdawful expensive, but the video game market provided enough to cover the tab. Contrast that with the 25 or so ROADM guys fighting for what might be a $20 million market next year.

  • Change quickly. Nvidia started by licensing its technology just to get revenues, but it killed that plan upon realizing it would take real chip sales to stay afloat. Other startups hang on to one idea too long; one VC at MicroVentures complained about an 802.11 play that refused to come out of stealth mode and missed the acquisition wave that benefited Radia Communications Inc. and Intersil Corp. (Nasdaq: ISIL) (see TI Snaps Up Radia and Intersil Exits WLAN).

  • Keep a laser-like focus. Don Valentine, a VC godfather in the startup world, bestowed this upon Huang, calling it his most important piece of advice. Must have been good, because arguably, lack of focus killed rival 3dfx Interactive Inc. As its chips fell behind Nvidia's, 3dfx made a desperate dive into the boards business. The pricey acquisition didn't save 3dfx, nor did it fix the problem of its slow chip development.
But not all of Nvidia's tricks ring true for telecom. For instance:

  • Build a killer product. Nvidia did this, but does a telecom product have to be "killer"? Nvidia was lucky enough to serve a market where technology matters above all else; gamers underwent whatever forklift upgrade it took to get the best graphics card. Several optical components startups assumed their market would be similarly hungry, and they got slapped down. The same could happen to some smart antenna startups eyeing the 802.11 market.

    It's not the product that has to be great. Plenty of Light Reading readers love to point out how Cisco Systems Inc.'s (Nasdaq: CSCO) reign is a triumph of marketing, not technology. Too many startups assume that lesson doesn't apply to them.

Finally, Nvidia had one thing telecom doesn't: an insane turnover rate. Even cell phones don't match it – novelties like push-to-talk don't compare to the mind-blowing, Toy-Story advances that 3D graphics made in the past decade. Nvidia cranked new chips every six months, and there was a ready market for each one. Try selling to RBOCs that way.

Huang made magic with his company, but even the tough 3D graphics world lacks telecom's special brand of brutality. Nvidia isn't the right role model for building the next telecom-equipment dynasty. But that's okay. Hopefully a startup will find some VCs with the wisdom to know the difference.

And, if you're looking for inspiration to get you through the next tech boom, you can't do better than a chat with Huang. He's a tech industry version of Andy Reid. And you've gotta like a graphics designer who still loves Battle Zone.

— Craig Matsumoto, Senior Editor, Light Reading

sevenbrooks 12/4/2012 | 11:09:59 PM
re: Valley Wonk: Lessons of the 3D Wars
I think that people's view of killer products as not being required by a startup is stunningly wrong.

Cisco built a pretty good router in the old days. In those same days, it was very hard to get consistent results out of Wellfleet Routers (even starting from a clean bootup and typing the exact same keystrokes on the craft ports of 2 identical routers).

Cisco parlayed the growth of IP routing into a huge money machine (aka Cisco Stock Price) and used this relatively cheap currency to purchase other promising technology leaders. Does anybody out there think that the Cerent 454 wasn't a leading edge product when it came out?

I think trying to say startups can come out with good marketing but mediocre products is a completely flawed model.

whyiswhy 12/4/2012 | 11:09:58 PM
re: Valley Wonk: Lessons of the 3D Wars The idea is to play to a big healthy market.

Telecom is big but not healthy. Too many "yet-to-die" out there. Too much motion to China and no really new products being created.

Optics and telecom: just re-spun five and ten year old ideas.
whyiswhy 12/4/2012 | 11:09:58 PM
re: Valley Wonk: Lessons of the 3D Wars Lessons learned the hard way:

Jes ain't no sech thin' as Killer product in Telecom...not in optics anyway. Not now.

Challenge: Name one.

New optical product equals:

1) Nobody knows about it.
2) It has not been Telcordia qualified.
3) 1 and 2 means no big company (read: customer) is gonna take a real chance on it.
4) Ergo, you manage to give (beg-away) away your samples, and you don't make even the lowest sales numbers you promised your investors.
5) You might get a wash second round if you manage to sell a few in a year to a big company with massive plans for their new product.
6) The big company buys a few more samples, then just waits for you and your investors to go under and get the IP (and you) for (almost) free in bankruptcy court.
7) You might get a job, your investors might get a framed thank you note from the big corporations VP of R&D for supporting his product development efforts.


sevenbrooks 12/4/2012 | 11:09:38 PM
re: Valley Wonk: Lessons of the 3D Wars

I think you and I basically agree on this. The only thing I would say is that there are specific product categories within telecom that could be moderately healthy.

The problem is that today's batch of startups are now aiming their same technology at the 2nd or 3rd customer base. It probably was very nice for the original customer base. Its somewhat off skew for today's customer base.

laserbrain 12/4/2012 | 11:09:31 PM
re: Valley Wonk: Lessons of the 3D Wars ahh yes both are critical. But when is change not a loss of focus? Only the truly great (or truly lucky) are able to tell and then convince the stakeholders (guys like Don Valentine screeming at you at the board meeting) that it's the right thing to do.

Do we look back and discover the Procket was brilliant for licensing their software or is it the final nail in the coffin?
Tony Li 12/4/2012 | 11:09:26 PM
re: Valley Wonk: Lessons of the 3D Wars
If you look back, you'll find that Procket hasn't licensed its software so far.

dwdm2 12/4/2012 | 11:09:25 PM
re: Valley Wonk: Lessons of the 3D Wars "Jes ain't no sech thin' as Killer product in Telecom...not in optics anyway. Not now.

Challenge: Name one."

Well, optical fiber itself is a killer product, everybody knows about it. BUT, the genie is kept in the bottle for all practical purposes...
particle_man 12/4/2012 | 11:09:17 PM
re: Valley Wonk: Lessons of the 3D Wars The only candidate I can come up with is the Titan digital cross connect. It created (and then dominated) a new category. This is a pretty old example.

I suppose you could argue that the original Ciena product was a killer product as it got them into Sprint and created a billion dollar market cap. It certainly didn't have the staying power of the Titan product line.
laserbrain 12/4/2012 | 11:09:13 PM
re: Valley Wonk: Lessons of the 3D Wars Tony -

My comment was triggered by Procket's web site which seems to be marketing software via the partners and products pages.

My gut says that the license market is tiny compared to your bullseye market for that company, but maybe it gets a foot in the door for "big brother" partnerships, OEM deals or something. Maybe it provides a steadier cash flow than waiting on BFR projects.

Or maybe it turns out to be a low-value, high-time distraction and you're lucky you haven't sold any yet.

best of luck
sevenbrooks 12/4/2012 | 11:08:45 PM
re: Valley Wonk: Lessons of the 3D Wars
Well, they don't all have to make new product categories. But here are some killer products that have made companies that I have thought of. Now some of these have been overcome that a company to become a long term company must overcome the single product life cycle of its original product and later failed but..... (I will start with your examples)....

Tellabs - TITN 500 DACS
Ciena - WDM Transport
Ciena - Core Director
Pairgain - HDSL
Redback - SMS 1000
Cerent - 454
Optilink - Litespan 2000
AFC - UMC 1000
Cisco - IOS

I am sure there are more examples. Each one of these products made at least a $100M business on its own at one point. Some were over $1B.

I think it turns out to be hard to make a winning product of those scales. Thus, the fallout rate of startups is high. Once a company hits a certain size, it can invest in a number of "singles" and get the occasional "home run".

Sign In