Axial - Observations #10
Life sciences reflections
More well thought out work can be found at — https://axial.substack.com/
Axial invests and partners in early-stage life sciences companies. If you or someone you know has a great idea or company in life sciences, Axial would be excited to get to know you and possibly invest in your vision and company — info@axialsprawl.com
Observations #10
A set of ideas and observations from a week’s worth of work analyzing businesses and technologies.
Business models for brain science
There are a wide range of companies developing products to measure brain activity non-invasively:
CTRL-Labs - https://www.ctrl-labs.com/
Neurable - https://www.neurable.com/
Neurosity - https://neurosity.co/
Kernel - https://www.kernel.co/
BrainCo - https://www.brainco.tech/
There is a plausible pathway to make a medical device company, but is there a pathway to build a business model focused on selling data derived from measuring brain activity to patients. Who would the customers be? Research labs - that’s not a big enough market? Advertisers - that’s probably crossing some major ethical lines? If the set of customers can be figured out then the business model for measuring the brain would be a pretty straight-forward consumables business: get high margin on data dn low margin on the device.
Neurable’s EEG device:
Choosing markets based on the toolkit not the size
Most businesses choose a market based on the size. That’s a useful method to choose where to focus. You can also choose initial markets based on the speed of the GTM then expand to larger markets. However, what kind of companies are choosing markets based on how well their technology can address the problem regardless of size/GTM?
One interesting business that is doing this is NGM Biopharmaceuticals - https://www.ngmbio.com/ They’ve been around for over a decade focusing on endocrine and metabolic disease with two phase II products for NASH. Their model is to find opportunities with a high margin of safety. NGM’s business model focused on two parts:
Is there a target that is the key driver for the disease?
Can NGM get to that target?
Then they figure out the degree of unmet clinical need and how large the market is. In other industries like healthcare and diagnostics, what are the equivalents? Maybe for the former, what diseases/behaviour can be modified through an application? For the latter, what molecules/data can be measured (i.e. now single-cell, microbiome) and where are they important?
Quantum computing for drug discovery
Quantum computing has a lot of potential to transform search in life sciences from searching valuable molecular interactions to new biomaterials. Quantum computing is a very difficult field to understand - my knowledge is based on a class I took in college and a former grad school roommate who was doing research on quantum computing at Lawrence Berkeley.
There is a large opportunity for quantum computing to improve small molecule manufacturing and medicinal chemistry instead of the commonly-held belief that drug discovery will be disrupted.
The current software products like Schrodinger, Accelrys, and Atomwise focus on using computational chemistry to model reactions and the interface between a drug and a target (i.e. docking). Quantum computing can help here but the competition is high and from my knowledgebase, this is not so much a search problem but a scale one.
During the process of virtually screening molecules, one has to also figure out if the candidates can be easily made by a medicinal chemist. This is a search problem where quantum computing can shine. There are at least 10^60 molecules with a molecule weight below 500 daltons (cutoff from Lipinski’s rules) - https://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291098-1128%28199601%2916%3A1%3C3%3A%3AAID-MED1%3E3.0.CO%3B2-6 Current software architectures can’t search this space; the current systems are pretty good at simulating individual molecules but searching 10^60 of anything is impossible right. Quantum computing offers the potential to search through spaces this large. Quantum products would search for molecules with feasible synthesis pathways, reactivity, and other features that can be scored a priori.
What is the market opportunity here? Molecular simulation companies might make $1B collectively. With over 1 million chemists and individual drug developers out there, the potential to build different types of businesses exists:
Use quantum computing to make your own drugs
Sell a quantum computing package to a partner and replicate the Schrodinger model
Sell directly to chemists the ability to search through their libraries; you would have to take a page from Abcam or Benchling
Building out the quantum computing clusters for others to plug into like what Rigetti Computing is doing
The biopharma R&D model
Do large biopharma companies need to do R&D? Overall, the performance of drug development is declining - R&D costs are growing faster than overall revenue despite major advances in immunology, computing, and sequencing. This productivity issue can’t just be technical given the breakthroughs we’ve seed and the declining results of the industry. Something must be a little broken with R&D - a level deeper would reveal that large biopharma companies (market cap above $10B) are doing pretty poorly at getting new drugs approved despite their massive R&D budgets. Whereas, small-to-mid-cap companies productivity has actually gone up. This observation validates that large biopharma are essentially banks for drug risk and small companies are pushing the field forward.
Other industries have pretty well-laid out product development pathways - invest in the technology early on and get to product-market fit. For drug development, what is product-market fit? It’s usually a successful phase II study (i.e. show efficacy in humans). As a result, drug companies’ product development pathways rely on experimentation and risk diversification. Feedback comes in from a clinical trial that take a few years to get to. Companies can take strategies to get to the clinic faster to get feedback more quickly.
With the complexities of human biology and current unpredictable nature of drug development as well as the productivity of smaller companies, biopharma R&D can make major leaps by redesigning who gets R&D dollars and building diverse pipelines to handle the risks:
Specialized venture funds by disease or modality - maximizing the number of PoC studies that if certain milestones are research will get acquired by biopharma (i.e. build-to-buy)
Accelerators and incubators to focus on de-risking the academic research
Discovery partnerships - pay a small company to develop risky research
R&D outsourcing to services companies like CROs
Superforecaster model like BridgeBio - https://axial.substack.com/p/axial-bridgebio
Citizen Medicine
What is required to bring more medicines and care direct-to-consumer? So who owns the patients - it’s the payors. They have the relationship with the consumer. The first step is developing businesses that aggregate patients in some way. Ideally, their products are paid for by the payors and in a way disrupt them over time by owning patient flow themselves.
So what are the opportunities:
Increasing engagement with patients and aligning their health with a business; most payors don’t have this
Changing patient behavior to improve things like drug adherence, diet, and so on. This is much harder than just prescribing a drug or something. Business models enabled by behavioural changes are often really unique and heavily misunderstood in the beginning.
Then combining all the data from owning patient flow into a single stream to go beyond the current Epic/Cerner EHR standard or Oracle-like monopoly
This type of stream would be longitudinal by nature and has the potential to transform patient care and drug/product development
