Sunday, September 28, 2008


Never having been one to shy away from a complicated deal structure, I am starting to learn why term sheets and VC deal structures are so vanilla--it’s a little frustrating ;-)

If I put my entrepreneur’s hat on, it seems to me that every entrepreneur wants the chance to raise as much money for as little dilution as possible. So why not structure deals that give the entrepreneur the chance to win by doing a lot for a little, but at the same time, give the VC a win if they don’t--i.e., underwrite the entire round, but put half the money in at the VC’s target pre-$ , and have a tranche and a milestone that if triggered pulls in the other half as a convertible into the next round (probably with a discount for risk); but if the milestone is missed the second half goes in at the same pre$. If the entrepreneur does a lot for a little, then more power to them! After floating this concept past several valley VCs, and wiping the vomit off my shoes, I think I am beginning to understand the problem ... so let’s see:

VCs worry a lot about alignment with the team. Often, especially in other places, you see deals where too little money is put in, and the team is basically set up to fail, but they got the deal done and are then at the mercy of the VCs when they run out of money. Clearly there's a fundamental misalignment here, and not a good working relationship.

A good investor will always want to fund the company with whatever it needs to reach the next valuation inflection, plus a buffer in case things go wrong. If the company does everything right but the market shifts, the VCs will still likely put more money in, and convert into the next round--if the company does not, they will likely put money in too, but at the same terms as the previous round--pretty simple, right?

The misalignment in the first structure is that the VCs will be hoping that the company misses the milestone so they can get to the valuation they need to make their fund economics work--when, instead, they should be cheering the company on to do as much as possible with as little as possible, to get the best valuation on the next round. They are also misaligned on the next round, because if they don’t get the percentage they need on the first round, they will want the new lead to set a lower valuation so they can buy up their percentage to reach their original target. The valuation sensitivity of the company and desire for funky structures also tells the VCs that in the next round, the company will be inclined to take the highest valuation regardless of where it comes from--letting finance dictate strategy, rather than making the best strategic decision to build long term value.

If the VCs don’t get their target percentage, then they will be spending a full partner on working with the company, but only getting a fraction of the reward for it--a good firm will simply not do the deal, a lesser firm might, but basically won’t give the company much time, so why do the deal anyway?

The other argument is, yes, but if you can put in less money you have less at risk, and if the company trips, you will still have an opportunity to buy more. The problem with this, is twofold--most early stage firms are in the business of putting 10M to work, putting 1M in a deal wont move the needle on the fund even if it is a 10x; second, if the company trips the VCs have to take a writedown which their limited partnerss hate ...

I would greatly welcome comment and debate around this issue, especially if someone has a great idea for a deal structure that resolves these issues so we could stop arguing about valuation, and get on with building great companies ;-)

If you prefer not to post, please email me at and put BLOG in the subject.

Sunday, September 14, 2008

Chip companies and why they are hard

At present its very, very hard to fund a chip company. Firms that would previously have been very bullish on the space have backed off completely, and one very well known firm for chip deals has done only one deal this year in that space. VCs have always been careful on fabless semi, because it typically takes $50M to build a chip-company--not for the weak of heart!

However, the economics and scale of the chip business keeps luring investors back. For example a GPS chip that might go into a cell phone, or a laser that might make a cell phone display ,get investors excited because of that wonderful "billion" prefix to the number of phones sold--even when multiplied by an ASP of only say $5 that turns into a very very large number.

Chips have large gross margins because they are hard to build and can have great IP barriers to entry. The downside is just about every design decision you make is fraught with risk and quickly followed by a $5M price tag on making your first serious chip run. The level of design team you need to get to tape out the chip means a million a month level burn rates, and pretty quickly your cash is gone.

If your market materializes and you are first in, then chances are you can be the Intel of that market and enjoy 80% share for a long time to come; but if you miss and your competitor gets in first, then you are fighting with the others for the scraps. The worst case is when the market does not take off when you thought--like femto cells, or RFID. Then you are stuck waiting for the market to happen and all your technological lead becomes worthless because your competitors have plenty of time to catch up while you are waiting. VCs get scared off by the lack of a market, and the company is in real trouble.

Volume is key to the chip-co; remember when DSP was first thrown at optical for dispersion compensation? It was like black magic for its ability to solve aberrations occurring in transit with DSP at the receiver. There was a massive wave of investment in that area because the technology offered such a compelling value proposition. Unfortunately, the volumes couldn’t support the chip economics, even in a high volume optical application like telcom (which for laser jocks is like consumer electronics is to PC people), you were still only selling say 50,000 transponders that needed EDC which simply couldn’t support the economics of a single chip-co let alone a flurry of them.

On the positive side, they are great because chips have the unique ability to integrate functionality on a scale never seen before. Remember how wireless started on a card or a USB plug-in and ended up as a feature on an Intel chip?

So once you are in, you can retain your gross margin and ASP simply by adding more and more functionality to your chip through integration of other hardware and software functions elsewhere in your customer’s product. From the customer perspective you are taking cost out of their product and driving up their gross margin, and at the same time you are supporting yours, albeit by eating someone else’s lunch ;-(.

There is a particularly interesting connection between CMOS and optical, at least for me, because the parallel processing potential of optics is so great, and the DSP capability of CMOS offers the ability to solve tough optical problems. The combination of the two often leads to great technologies, at least when there aren’t 20 companies funded at once to do it (circa 2000). Hey, maybe this is a good time to do a chip deal ;-)

Responses to Comments:

To Life Saving Beverage Maker:
It's true that at acquisition time many investors don’t seem to value the team, but I think it's all about Team--and a good investor should be part of that team, otherwise, sure, put the money in the bank instead. Man, we’ve met a few VCs who acted a lot more like bankers when it came to risk haven’t we?

To BJ:
Mate, I was actually thinking of Lightbit when I wrote this one; the CEO or Chairman should answer your question, but if you shoot me an email, I will forward to them. From the VC’s perspective it was a top quartile success, and 3x return in the year after IPO is great – but you’ve given me a great idea for a future blog, thanks.