Discover more from Elad Blog
Enough With This "End Of Silicon Valley" Nonsense
It seems like every 2-3 months recently there has been a new post about the end of Silicon Valley, or related, about how software as a category is over and done with . I think the arguments made that Silicon Valley is not innovating are far off the mark.
The arguments being made (and why they are wrong) are:
Argument A. "Social, mobile, and other internet products are not creating real societal value."
The social media and mobile applications that are being poo-poo'd are actually dramatically transforming the way society buys things, discovers information, and organizes itself. This includes big changes in industries such as hotels (AirBnB), transportation (Uber), and news/communication (Twitter). To say that the founders of AirBnB should have been working on a nano-materials startup instead of changing the economics of un-utilized space is kind of vacuous.
Argument B. "Software is easy. People need to work on things that are "harder" that involves, like, you know, physics or advanced materials or something."
The whole point of technology is to make innovation faster and easier - software enables this. In the last 40 years, we have moved up the stack from hardware to the software abstraction layer. Software lets you innovate faster and drives down costs, disrupting multiple industries . Rapidly dropping cost curves are what drive large scale societal transformation (e.g. steam energy and the industrial revolution, processor & bandwidth costs and the internet revolution). Similar cost curves or other key market shifts are starting to happen in genomics and hardware.
Counter-argument B2:It is more important to work on something valuable than something that is "hard". "Valuable" means that someone, somewhere, is willing to pay for it. Strikingly, two of the most successful companies in clean tech to date are a software company using social mechanics (Opower) and a financial services company (SolarCity). The hard science companies in energy are largely failing (e.g. Solyndra) due to industry economics or an inability to bring product to market . Early stage startups exist to commercialize technology, not do massive research explorations.
Hacking society is actually quite hard. It is possible consumer internet companies have the highest failure rates of any technology category. This is due to both (a) a huge influx of entrepreneurs given the low cost and glamour of starting a consumer internet company and (b) getting traction for social products is hard. The people who win, win big (Wikipedia, Facebook, etc.). But there are lots of dead consumer companies.
Innovation in Silicon Valley is thriving. Internet and mobile still have a long way to go with their own innovation cycles. Additionally, most new waves of technology are built on the fringes and then emerge from the shadows with success. (How many article did you read about SpaceX before ~2010? The company was founded 10 years ago in 2002!). There are multiple startups taking on healthcare, education, food, consumer hardware, and other industries. These startups are not covered by techcrunch but will get lots of press in a few years as they get so large you can't ignore them.
Argument C. "People are working on a lot of stupid ideas instead of changing the world."
While I agree with the overall premise on this, I don't think entrepreneurial talent is zero sum or fully fungible. The random person building an Instagram clone for cat owners is not necessarily the same entrepreneur who would find success revolutionizing our energy infrastructure (their choice of "Instagram for cats" as their startup sort of says it all).
Intriguingly, a lot of the entrepreneurs I know who had a small success with a consumer startup are thinking big for their second startup. Personal financial stability and past experience can go a long way for people to take big bets.
Argument D. "Venture capitalists should be placing bets that may take 10-15 years to succeed."
I do believe most venture capitalists are generally risk averse herd-seeking 'me too' types. There are a handful of notable counterexamples to this. However, I dont agree with the premise articulated by Steve Blank in his post:
"If investors have a choice of investing in a blockbuster cancer drug that will pay them nothing for fifteen years or a social media application that can go big in a few years, which do you think they’re going to pick?"
A 15 year time horizon for an exit is simply not the venture capital model, and it never was. Even Genentech, a highly speculative company at the time, was founded in 1976 and went public in 1980 - just 4 years later. Public markets, as well as corporate customers or investors, often end up funding value-creating companies with longer time horizons. A more recent example is SpaceX, which is benefiting from a combination of Elon Musk's personal wealth, venture funding, customers paying up front for space delivery and government funding.
If you can not find a source of funding or liquidity for your idea 10 years down the line, you are not starting a company but rather doing research. This is not a failure in the venture capital model, this is a failure in your ability to chose something people will pay for.
To sum it all up, innovation in Silicon Valley is alive and well. The Internet and mobile will continue to transform multiple industries, and in parallel new technologies and companies will emerge from the shadows over time. Venture capital has its issues, but does not, and should not, exist to fund research projects with open-ended times to an exit.
You can follow me on Twitter here.
 Really this should read "the end of Silicon Valley or its venture model", since many of the posts spend a lot of time on the venture capital side of it.
 Ironically, most of the people arguing software is "too easy" made their fortunes and reputations off of the exact type of Internet or enterprise software businesses they now consider lower value. I can't think of a single pundit who made their name building a biotech, energy, or transportation company.
 This point was first raised to me by Naval Ravikant, co-founder of AngelList (and a bunch of others cos) and one of the smartest guys in the valley. If you look at the evolution of tech, it has been a long climb up the stack to try to abstract away the hardware and underlying software layers as much as possible. Why? Because at the higher abstraction layers (e.g. application & presentation layers) you can both iterate quickly, but that is really where the value lies for multiple industries (retail, finance, government, etc.).
 When I was at MIT for my PhD, I would frequently run into people trying to commercialize science projects that had no underlying customer need. Often these people were graduate students or professors who had created really amazing technology (e.g. complex MEMs devices) with no short term commercial application.
So, you ended up with lots and lots of university spin outs without a clearly defined customer need raising large amounts of capital to "commercialize" a hard technology. Much of this money went to R&D (with an emphasis on the "Research" part of R&D) and many of these companies flamed out and failed. Just because something is "hard" does not mean it meets a market need.
In general, I don't think basic scientists *should* be trying to spend time on customer visits or product/market fit - basic science has its own enormous long term societal value.
 By the way, there are all sorts of interesting commercial applications of MEMs (e.g. the accelerometer in your car's airbag which tells it when to deploy). I am just raising this as an example I have seen in the past of a technology looking for a market.
 Blank typically has one of the better startup blogs out there. On a side note, Blank also did an amazing job building the technical team at E.piphany. At Mixer Labs, I tried to recruit a number of ex-E.piphany folks who, as a class, were stellar.
 Obviously SpaceX is unique in that Elon Musk was able to start the company with his personal wealth versus needing initial funding sources.