Discover more from unicorn chats
🦄 vol. 20
the final edition for 2021: matt and sash chat about alphafold and corporate innovation
20 editions, 40 different articles, 20 consecutive weeks.
look, we’re enjoying this, but we also want to take a dezemba break.
so yes, this is the last edition for 2021… but we’ll be back (with a bang) in 2022. 🤝
origami and antibiotics 💊
A handful of months ago, DeepMind, the AI research company owned by Google's parent company Alphabet, published an open-source database that detailed the predicted structure of almost every one of the 20,000 proteins that make up the human proteome (like genome, but for proteins).
Pff, big deal... 😅
Well, it actually is. At a granular level, proteins have a role to play in almost every structural and functional aspect of living beings: digesting food, generating an immune response, aiding in muscle contraction and transporting oxygen around your body. And their function is almost solely dependent on their structure.
How hard can it be to determine their structure? 🤷♂️
Proteins are made up of a combination of building blocks, called amino acids, of which there are only 20, and varying combinations of those amino acids give rise to the 20-odd thousand proteins that make up the human proteome.
While that's easy to imagine in a linear form, proteins spontaneously fold (and twist), hence origami 😜, to exist in a complex 3D conformational form. And it's their 3D structure that, almost single handedly, dictates what function they perform. Knowing the structure is so important that scientists have remarked that solving the protein folding problem is “one of the most important yet unsolved issues of modern science.”
Unsolved...until AlphaFold 🥳
Up until this point, the scientific community only knew the structure of 17% of the proteins that comprise the human proteome, which were determined essentially by means of trial-and-error. With AlphaFold, the deep learning algorithm can now predict close to 94% of the human proteome with some level of confidence using the sequence of amino acids.
Trained by showing it the sequences and structures of around 100,000 known proteins, the AI system can predict the shape of any protein in minutes (with scarily high accuracy).
Where to from here? ⏩
Scientific feats like this aren't of much use if their discoveries don't translate into real world impact; this one does.
Since Alexander Fleming's serendipitous discovery of penicillin in 1928, antibiotics have played a significant role in improving health outcomes the world over. Weirdly enough, most of the antibiotics we use today were discovered and developed in the "golden era" of antibiotics almost 70 years ago. (Yes, you're treating your tonsillitis with a 1960s Ford Mustang)
As with vaccine development, antibiotic development is a drawn-out, expensive and failure-prone exercise. In the last 7 years, we've only seen 14 new antibiotics approved. In a survey of over 186,000 clinical trials, the success rate of producing an efficacious antibiotic sits as low as 25%. And even if one manages to succeed, it often takes longer than a decade from conception to market.
AlphaFold and antibiotics 💊
The predictive power of AlphaFold is being utilised by a research team at University of Colorado Boulder to generate novel antibiotics. While still in its infancy, commentators see immense promise.
It has even found a role in the pandemic, with a group at University of California San Francisco using AlphaFold in a bid to better understand the biology of SARS-CoV-2 (COVID).
If you work in corporate, you've probably heard the “we must future proof our business” cliché. Or better yet, your CEO came off of mute in your team Zoom call and added "we need to disrupt our business from the inside".
The bottom line is that your CEO is kinda right. Exogenous shocks (i.e. a global pandemic) aside, we’re living in a world where the operating environment for businesses is changing faster than ever before. One only has to watch Warren Buffet’s presentation at the 2021 Berkshire Hathaway shareholders meeting to see just how rapidly the largest businesses (by market cap) in 1989 have virtually gone into extinction.
So while there’s no denying that your CEO is on the right track, it's definitely easier said than done.
Conducive environments? 🤔
Some corporates are more serious about innovation than others, and it’s evident that those that are, invest a lot more time and money into the (ongoing) process. Here are some simplified examples of popular methods used:
🔬 Corporate innovation labs: creating an environment where employees can submit their ideas and test them.
💰 Corporate venture capital (VC): investing capital and resources into young startups with the hope that they can be the next major disruptor in their industry.
⚡️ Corporate accelerator programmes: housing even earlier-stage startups and providing office space, mentorship and smaller sums of investment with the hope to seriously invest in / acquire the successful ones.
IBM South Africa partnered with the Tshimologong Digital Innovation Precinct to launch its own accelerator programme.
But how does Big Tech do it? 🤔
It’s no secret that Big Tech (Meta, Amazon, Apple, Microsoft and Google) continue to be Big Tech because of their uncanny ability to continue “innovating”. Facebook (now known as Meta) famously acquired Instagram in 2012, and had multiple (failed) attempts at trying to acquire Snap (before creating their own version of Snapchat on Instagram).
But personally, I find Alphabet’s approach to be the most interesting:
In 2015, Google made the decision to rename and reorganise the company. The decision resulted in Google’s core products (left of the diagram) being separated from its “other bets” (right of the diagram).
Google’s core products generate almost all of Alphabet’s revenue (Ads on its own makes up over 80% 🤯).
By “other bets”, Alphabet is referring to its “long-term plays”. You can think of these as riskier ventures that Alphabet is ploughing resources into, with the hope that a few take off.
By splitting these two up, Alphabet also allows its “other bets” to operate with a high degree of autonomy, with little influence from the Google companies.
Within “other bets” also lies its super cool semi-secret lab: X, the moonshot factory - “creating radical new technologies to solve some of the world’s biggest problems”.
It appears that Alphabet isn’t just disrupting Google from the inside, it’s disrupting Google from the outside too!
Corporate innovation is easier said than done. It definitely takes a clear commitment to consistently invest time and resources towards building an environment that is conducive to it.
But even then, innovation is never guaranteed.
claude was excited by vauban’s successful series a funding round
sash went down a vc rabbit hole and started listening to the a16z podcast