Deep tech conumdrum

As the whole Silicon Valley veered fast into "deep tech" or "anti-lean" investments – or at least lip-service chats – a couple of interesting behaviors surfaced.

One of the most intriguing behaviors is VCs talking about investment horizons that are beyond normal fund lives, e.g. 10~12 years. They talk about how they are willing to wait another 3 or 4 years before the team rolls out a really good deep tech product. They talk about real disruptions in energy generation, material sciences, etc, as opposed to yet another teenager social networking app. They talk about how the 10-year bet would either monopolize a market or perish.

So much is talking, especially that some of those VCs do not have backgrounds as entrepreneurs or even engineers in the said deep-tech fields.

Semiconductor used to be hot deep techs. It's still deep tech today – an outside would never be able to imagine the difficulty of crafting a 1B-transistor SoC that include complicated analog and radio functional blocks and bring it into mass production – just not hot anymore.

In fact, it stopped being hot almost 10 years ago.

The main reason semiconductor became cold and remains cold in terms of venture investments is the industry structure. The cost of developing chips in advanced technology nodes rises much faster than the benefit one receives from advanced nodes. Capex dominates the whole product development cycle. As a result any customized SoC has to sell tens of millions if not hundreds of millions to justify the fully-customized design.

The high-capex and the volume mandate lead to little room for product differentiation – every chip has to aim at a general and large enough market and cannot afford to miss the target. The result is complete competition – since every chip in the same category pretty much has the same feature/price matrices.

Complete competition leads to low margins, which voids the potential of even trying to do this in startups.

As an ex-semiconductor veteran, I sometimes have to snuff my laughs when a lean-startup VC bragged to me about the latest deep tech investment that he did without really understanding the tech inside out.

"This is gonna be HUGE~~" he might joke with a Trump tone, but it seems like a real joke to me however.

Deep tech is no doubt trendy now, but let's not get too carried away. Investments still need to generate returns. The more capital required, the more an investor has to analyze the various aspects of such a business in an early stage. Simply saying it's deep tech and dump money into it is as stupid, if not more, than investing in another SoC startup today. 

If you don't believe me, I could give you contacts of several prominent VCs who are old enough to have participated in the semiconductor startup boom of 1995~2005 and the bust thereafter. Just make sure that you don't tell them that your deep-tech portfolio company just completed the MVP (Minimum Viable Product) and is ready to iterate for the 1st roll-out.

The game of "No"

Some thoughts about hardware investment trends in 2017