🎓 mental models, technology cost declines and new year new format
Welcome back - new year new newsletter name and format, plagiarised from my own Medium page. This year, I’ll be posting monthly-ish and looking at a few venture investing mental models - frameworks to think about technologies, founders or startups - and I’ll also be occassionally delving into product-market fit best practices, as well as the usual macro trends and articles. Hope you enjoy.
Framework #1 - technology cost declines
If you’ve ever read the now famous Roelef Botha memo on Sequoia’s investment in YouTube you might know that much of their thesis was predicated on the accelerating penetration of broadband connectivity, or another way of thinking about it are the specific cost declines of that technology. In this instance, it meant as broadband price per Mb got cheaper, more people came online, therefore more people could access streaming video.
First, for context, let’s have a look at a technology who’s cost declines birthed inumerable saw a multitude of breakout applications - the OG network effect, electricity - be it lighting, computers, appliances and transport
Let’s now look at a couple technologies where we’ve likely already absorbed multiple standard deviations of cost change already;
Lithium-ion batteries are key in the shift from combustion engines to EVs, the fight against climate change and off-grid energy storage
Currently many of the widespread uses are being hampered by costs which is currently at about $180 per kWh - it’s believed that prices need to fall to around $90 per kWh for EVs to be widely affordable and in turn for the demand for batteries to increase dramatically, in turn lowering the cost
The shift to cloud has turned compute and server capacity from being a Capex line item to being a variable cost, ie the cost base scales with demand
This has led to the barriers to startup creation lowering but also has enabled compute heavy technologies such as machine learning which in turn is powering a whole heap of apps such as NLP, voice computing, computer vision, automous systems and even your TikTok feed :)
Likely, the biggest change in pricing may well be in genes sequenced per $
In 2001, genome sequencing cost $100m, in 2015 it was $4000, today it costs around $1000
Many of these cost declines have been driven by a) more accurate scanning of part of the genome b) better handling of the data pipeline c) increase in monetisable opportunities for that data (see 23andMe & GSK drug development)
“NGS” or Next Generation Sequencing is giving birth to more effective clinical trials through enhanced selection, targeted treatments, liquid biopsies and in combination AI/ML is enabling synthetic biology and gene editing
Importantly, cost decline charts can be forward indicators of where and when breakout technologies might develop, and thus where founders and investors alike might decide to allocate resources.
If you’re looking at any other technologies, where you believe we’ll see significant unit cost reductions, let me know!
Articles + News
Visa buys Plaid for $5.3bn - Visa put together a presentation on their rationale, tl,dr: as the software layer of fintech has developed, they want to own the connection between consumers and application
7 Reasons Why Video Gaming Will Take Over - an incredibly interesting essay on gaming, some fascinating points; TV has peaked and its attention is being redistributed, how gaming has much higher content leverage (replayability), consistent changes in format (Arcade, PC, console) have provided additive growth to the industry,
Discord and the ‘Future of Work’ - How Discord could be used as a persistent meta-layer / "lobby" for workplaces, with features such as work allocation, remote collaboration, file sharing etc - great read
Thanks for reading!
If you enjoyed this, please share with a friend or colleague