Deep Science Ventures Year 3: A new paradigm for Applied Science.

Mark Hammond
Jan 8 · 8 min read

Part 1 of 3.

This is a series of posts summarising our journey so far in redesigning the way that science is applied in the world. The second will be published on the 19th Jan 2020.

Technology, team or opportunity first?

After nearly a decade in science commercialisation we knew that there was considerable room for new thinking. Everyone is frustrated by our sole reliance on technology transfer to derive commercial impact from science, from the practitioners to the academics. And whilst surrounding funds and accelerators can, in some surprisingly rare instances, make good returns by picking off the cream of the crop (both IP and talent), it’s clear that this probably isn’t making the most of the potential locked up in scientific research: on average tech transfer offices lose money. This is even more the case as corporations shift away from R&D and product complexity exceeds the breadth of any given lab.

The stock policy response is to go after the obvious gaps in the technology commercialisation pipeline, however, these approaches rarely fully consider the dynamics of how those pieces fit together. For example: it is frequently suggested that we simply need to add more capital for ‘growth’ once ventures are established. Or that we should create IP matchmaking services. Start ‘Challenge funds’. Add funding that can only be used in universities for ‘translation’. Provide ‘services’ that can only be used once companies have met certain proof points, and require a grant, in a university. Add in EIRs. Push academics through ‘accelerators’. The list goes on.

Existing model pivot around 1) people 2) specific problems (usually either extremely niche or extremely vague) or 3) a new technology with little space to consider a systems level perspective.

These approaches rarely get to the root cause issues that constrain the commercial success of tech transfer. Rather, they double down on the necessary tendency of researchers to focus on research questions: that is, relatively niche problems which have a high likelihood of returning new knowledge about the world. These are then dressed up as venture scale opportunities in subsequent commercialisation steps. Meanwhile anything that looks like it might be working gets bid up to unjustifiable valuations because so much capital sits at the post series A ‘safe’ end of the pipeline. This is a phenomenon we commented on when we participated in the House of Lords Review of the Life Science Industrial Strategy earlier in the year

This is a serious problem. It’s what largely underlies why we are still going after one target at a time in cancer, despite knowing very well that it’s the complexity of the disease that causes failures in late stage trials. It’s why many of the approaches in cleantech failed — pushing forward incremental improvements with little consideration of the wider market dynamics. This highly linear, fragmented and serendipity-driven commercialisation model also, in our opinion, feels like an incredible waste of the best scientific talent.

None of the existing approaches systematically identify how to go after some of our biggest challenges with an integrated market and scientific perspective. As soon as you lock this process into any given technical approach or skill set, you forgo the opportunity to find the best possible solution. None of the existing processes engineer solutions from across the scientific landscape and take a long term view on how a single new company might redefine and own a sector, focusing instead on process or product level innovation.

What can be learnt from existing models?

Our objective was to develop a model which could scalably, proactively and systematically address the most profound opportunities by creating highly profitable monopoly science companies.

We have learnt a lot from our own direct experience with each of the existing tech-transfer investment, accelerator, talent investor and build-in-house venture models, as well as the grand-challenge and corporate innovation models that surround these.

For example, accelerators represent best practice in finding product-market-fit as rapidly as possible but often end up in a local optima around technology and market for many years due to the lock on the initial technological idea, which is the one aspect of a business you don’t want to be locked into if the aim is to drive sector wide change. Note that we don’t say this lightly: we were the founders of Imperial College’s first ever accelerator programme, from which teams have raised over £50m, with exits to both Google and Facebook. But having witnessed the extreme wealth of scientific talent at Imperial, and the world-class quality of the facilities available, we know that this still leaves huge potential on the table.

The good ‘ol days launching Imperial’s accelerator.

Talent investors have demonstrated that less experienced, synthetic management teams with incredible founder characteristics can very quickly build venture-scale companies. This can drive great results and undeniably creates opportunity. But the need to focus on a starting point of technology or a given skill-set can end up driving the same challenges as accelerators when it comes to applied science; sacrificing our ability to proactively and uncompromisingly take on the greatest possible opportunities.

Build-in-house funds such as Flagship Pioneering and Atlas have shown the power of developing conviction around major scientific opportunities, with returns often exceeding those of the tier 1 ‘tech’ funds. However, such funds don’t scale well beyond their pool of experienced entrepreneurs (and those individuals’ bias to do something very similar again). And as with traditional funds the typical cherry-picking of singular assets is increasingly competitive as, in general, problems become more complex as low hanging fruit is addressed.

Designing a model from the ground up for Applied Science

How do you create a model that has the returns of the build-in-house funds, scales like talent investing and solves the enormously pressing problems that exist at a sector level such as net-zero carbon, new models of computation, order of magnitude increases in crop yield and the many diseases which have thwarted every approach to prevent them so far?

We started in 2017 by building on our understanding of what makes exceptional founder-type scientists and were first to market, worldwide, for a science-based, talent-first venture builder. But in our first year pursuing this model, during which we produced and invested in 13 companies, we noticed a minority pattern of founder-type scientists who diverged from the “pair up and sprint” approach we advocated.

Our early days of science focused talent investment. Yesh (left) went on to build Scalpel to tackle patient medical error — read his story here, and Ben (right) went on to build Neuroloom, living neural interfaces.

Instead, this minority of founders used our funding to step back and identify a sector scale opportunity through structured analysis of market dynamics and technical landscape, to get as close as possible to an optimal approach to address a given opportunity. This often involved interviewing hundreds of stakeholders to build a picture of the constraints underlying the lack of change in a given sector so far, prior to designing tens of varied potential approaches independent of any particular technology based starting point. We realised we had actually seen this before in now portfolio company Cytera Cellworks as they sought to understand the biggest impact they could make in accelerating biotech R&D. It was a process more akin to the design-thinking undertaken at IDEO than ‘acceleration’, but is fundamentally far more challenging, given that the subject matter is science (and is therefore quite esoteric) and it’s not clear what technical expertise is required until fairly late in the process.

Portfolio company Cytera CellWorks who we met at Imperial took a highly systematic approach to unpicking what slows down science, inspiring the model that we now run at DSV. Photo: Computer Aided Biology community.

We decided to risk everything and double down on this, investing an entire year into systematising this approach. Almost as soon as we made this decision, we realised quite how powerful it might be. It was an approach capable of delivering key scientific interventions that could drive massive systems change. However, in many instances systems-level interventions of this flavour clearly go against the grain of locally optimised, but comparatively smaller, wins.

Most people are well-intentioned, however whilst fierce competition and the need for financial returns drives innovation, it also leads to perverse incentives on an individual level vs. what’s best for the sector, problem space, or indeed the long term growth of a company. This includes researchers publishing dubious results, GPs investing for rapid uplift in book value over long term results, pharmaceuticals pursuing approval on flimsy statistics, corporates trapped in the innovator’s dilemma, all coming a long way before any improvement. And so central to the design of our new model was the need to create technologies, product and companies that were designed to align these fragmented incentives, between individual and company, between company and society.

Matt from portfolio company CC Bio winning the AstraZeneca partnership competition due to their highly systematic approach to finding ways for antibiotics make sense both commercially and medically.

The beginning of a new approach

This new model has allowed us to tackle some of the biggest challenges in driving impact from research, taking an unbiased systematic approach to joining up siloed branches of science and addressing the gap in early-stage venture funding. Even if we can create a much more effective model, there is a gap in very early stage funding, and this is still the major thing holding back the translational scientific potential in the UK.

Even the earliest stage Innovate-UK grants require data and commercial partners. This creates an enormous bias towards approaches that already have data, have previously secured funding and can list the right names, leading to a preponderance of more incremental approaches. Whilst entrepreneurial scientists spend hours (years?) writing grants, industry has both the capital and the appetite, but often lacks effective mechanisms to work effectively with universities or early stage companies.

We seem to have found an approach that unlocks this industry engagement. For example our partnership with OGTC provides £100k non-dilutive upfront, a further £500k in non-dilutive proof of concept funding alongside energy industry partners, on top of our investment of up to a further £500k. Similarly a partnership in the computational hardware sector, to be announced later in the year, is backed by an internal £250m fund that will work alongside us from day one.

Nuno, founder of materials research acceleration platform HolyGrail.ai (now a YC company) at the OGTC partner day.

The outcome of this longer term, committed funding is that DSV is able to co-create the sort of contrarian, audacious companies that the most ambitious founders and scientists are dying to build, but that conservative research and science venture funders have been reticent to back. We can move with conviction in challenging sectors, until these companies have enough proof to raise with risk profiles that the wider market is currently more comfortable with.

In the last 2 years we’ve systematically deconstructed sectors from oncology (e.g. ‘why are drugs still not stratified according to resistance profile?’) to energy storage (e.g. ‘why are we still using battery technology from the 70s’)?

In part 2 of this post we’ll take a deep dive into the early results and the journey so far. In part 3, we cover what we’re building next.

If you’d like to find out which opportunities we’re currently working on and roles available you can find out more here.

Mark Hammond

Written by

Founder at @deepsciventures creating a new paradigm for applied science. Ex-neuropharmacologist & AI researcher.

Deep Science Ventures

A new paradigm for applied science. We bring together teams of scientists to seize crucial opportunities and redefine industries.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade