Is it a good time to start a software company?

The only answer that’s guaranteed to be wrong is “always.” While many great companies (Microsoft, Facebook) have been started in ‘bad’ times, there are better and worse times to start a new company.

When platforms are rising, customers have money to spend, talent is cheap, and enough (but not too much) capital is available is the best time. The opposite of those conditions is the worst time.

While there are secular changes (open source, computing as a service) that makes it easier to start a software company than ever before, and so there will surely be more, what we don’t know is what these companies’ margins will look like, how large they will be, and to whom the benefits will accrue.

In other words, in 1890 it was a great time to be a steam engine, but not a steam engine company. Our history is full of boom time industries (steam engines, airlines, electricity) that dispersed and became boring. But boring is only an insult coming from the capitalist; another way of saying boring is pervasive.

Carlota Perez differentiates between the installation phase (technology enters market, development funded by financiers) and the deployment phase (technology is widely pervasive, development funded by in-market companies). In the latter phase, society has greater access to the benefits of the technology, but earnings are more concentrated amongst the winners.

If you’re starting an infrastructure company (say, doing machine learning as a service, or data storage as a service) today you’ll have to compete against war chests of Microsoft, Amazon, and Google. They will outspend you and out-bundle you. If you’re starting a video streaming service, you’ll be compete against the war chests of Snapchat and Facebook. They will out-distribute and outspend you. This is almost certainly a boon for the consumer, as a long there continues to be viable giant competitor for each giant. But it is almost certainly not a good time to compete against the giants.

Instead, new companies (building on top of the infrastructure and distribution of the giants) will go up the stack, allowing for the proliferation of hundreds of thousands of small businesses that are now easy to get started because of all the things they no longer have to do.

The magnitude of the successes will go down but the frequency will go up, which will require a shift in how venture operates. You may not be able to build a venture capital model that depends on multiple multi-billion dollar exits.

Here are the questions I am thinking about:

1. In this phase of software’s deployment, what will distribution of the winners look like?
2. What is the next ‘platform’ in software?
3. What is the next boom industry post-software?
4. What are the valuable assets to capture from the last boom and redirect towards the next one?

You may be surprised that my preliminary conclusion has lead me to begin investing in software companies again. My inbox is open.

—

To receive future posts via email, sign up here.