It seems amongst many scientists that the idea of starting a company is still looked down upon. Many seem to imagine companies as entities for which one must give up their principles and the possibility to pursue the most interesting work.
But done right, that doesn’t need to be the case at all. A good company can actually be a superior vector for pursuing the most impactful research.
If more scientists adopted the model of starting research companies, I think we would see many more transformational technologies come to exist than we currently do. I believe many kinds of basic research questions would be explored at a faster rate too.
A Better Way to Coordinate Teamwork
I like to think of companies as social organization tools first and foremost, where people can be optimally aligned towards working on a hard problem.
In academia, most scientists are incentivized to work on their own research in a relatively siloed fashion – publications are an all-important currency that is hard to be shared.
The culture seems to end up revolving more around the individual, heavily discouraging teams of multiple people from devoting all their attention towards working together on the same problem. 
On the other hand, with a company, it's much easier to collectively coordinate people towards the same goal. With a general de-emphasis on the individual, collective milestones can be set for everyone as all-important north stars.
This type of environment makes it much simpler for people to work together on different approaches to the same question, or even double up on the same sets of experiments to brute force more possibilities and get past bottlenecks.
With companies too, you’re much more likely to have support team members assisting with experiment setup and downstream analysis to free up other members of the team to focus on other parts of the problem. This makes work much more efficient.
A Greater Path for Funding
The other big advantage companies have is their ability to unlock resources.
The present period of where we are in the world is one awash with investment capital unlike a time that has ever existed. It is easier than ever for ambitious scientists with a plan to approach investors and ask for private funding to support a startup.
Good companies can mobilize a lot more capital in a much shorter time frame than what is possible through traditional research grants. Private capital is often less restrictive, and can be spent on large experiments that would otherwise not be feasible. And as a smaller company that is just getting started, private capital can still be complemented with grant funding – there are many non-dilutive grants (like SBIR and STTR US government grants, as well as private foundation grants) that private companies can obtain.
The idea of pitching to investors for funding might be scary, but it is becoming common for graduate students and postdocs to get hundreds of thousands to millions of dollars from investors to start their own companies. Often, they don’t have anything more than just an idea, a rough plan, and a commitment to pursue their work full time.
And as progress is made and momentum grows, it’s common for company founders to raise the tens of millions to hundreds of millions that become necessary to translate research – a much harder feat to accomplish from the academic side.
With fewer and fewer professorship tracks in academia, granted tenure positions, and conventional opportunities for ambitious early career scientists, it’s seriously worth considering this alternate path.
What it Takes
As much as I espouse that scientists should start companies, that doesn't mean it makes sense for everyone. For those that are interested in starting a company, there are important criteria to be met – In general, you want to be the right kind of person and have the right kind of idea.
On the idea side, you want to be working on something that has the potential to have a massive long-term impact – this is the type of realm that private investors will be eager to fund. This could be a cure or treatment to some disease, or some type of novel technology that can improve people's lives. It's actually very OK for this to be just an idea at the beginning and for a lot of discovery work to still have to be done. The main point is that there should be some type of discernible endpoint, which, even if it is challenging to reach, will make a big difference in people’s lives.
If you want to do something purely for basic research / mechanistic understanding, academia is still probably the better place for that. 
To be a successful company founder also requires a certain type of character and it's a realistic truth that not every scientist is cut out to start a company. Company creation can often be a lonely, uncertain process where it can be difficult to keep up motivation, it can be hard to know what to focus on, it can be hard to convince people to work with you, and it can be hard to convince investors to fund you.
It's important to have the right personality to do this type of work – ascribing to the theory that companies are good vectors for people and resource coordination, you or someone else on your team should have the aptitude to grow into being a leader.
If you don’t think you can do this yourself, that's fine—try to find someone to join you in getting started who you think can. Companies often do best when it’s two or three people of complementary skills coming together as co-founders to get started (and they don’t have to be business people either – often ‘business skills’ are of secondary importance at the start to getting the technology development plan right).
If you don’t know the right people to start a company with, consider joining a startup you think is ambitious and promising as an alternative – you’re likely to meet other like-minded people who you might be able to work with, and you can contribute to important and interesting work in the meantime.
If you indeed start a company, it will be challenging, but people will want to help you – people love to support those who have the courage to go after very hard but important things. Over time you will be able to assemble advisors, coaches, and investors that will assist with your growth.
I’m happy to be one to help if you want to reach out – I firmly believe more scientists pursuing companies will be an incredibly impactful shift over time, and one that will make the research world wider.
Thanks Alex Schubert and Lucas Harrington for reading drafts of this post.
 I find it odd how most academic labs tend to be named after the Professor – I think it's harder for people to feel like a super connected team when everything seems to revolve around the importance of the leader.
 Academia has an incredibly important place – beyond just the fact that starting companies isn't for everyone, there are many basic research topics that companies would be bad mechanisms to pursue – it's wonderful we have a system where many other kinds of research can be optimally supported. We should be doing more to help academic research flourish.