Alexandria Real Estate Equities (ARE) had a lot to celebrate as it marked its 30th anniversary in business earlier this year, including nearly 27 years as a publicly-traded real estate investment trust (REIT). Alexandria has evolved successfully from a garage startup to the largest and longest-tenured owner, operator, and developer specializing in life science mega-campuses in top-tier innovation cluster locations, including seven of the Top 10 U.S. Biopharma Clusters as ranked in GEN’s nationally-quoted annual A-List.
From $19 million in Series A capital, Alexandria has grown to a total equity capitalization of $21.8 billion and 73.5 million square feet of total assets in North America, including 23.9 million square feet of future development projects, as of December 31, 2023. From its $155.25 million initial public offering in May 1997, which generated net proceeds of more than $139 million, through December 31, 2023, Alexandria generated a total shareholder return that exceeded 1,500%.
The first weeks of 2024 have proven busy ones for Alexandria. Earlier this month, the REIT celebrated Takeda Pharmaceutical’s signing of a 10-year lease extension through March 31, 2040, for 220,361 square feet at 75/125 Binney Street within the Alexandria Center at the Kendall Square mega-campus in Cambridge, MA. However, in New York City, Alexandria sold one of its two properties in Long Island City, Queens, to movie production facility operator Cine Magic for $19.1 million, according to a public record—down from the approximately $25 million Alexandria spent to buy the site in 2019.
More impactful than those deals, Alexandria says, is how artificial intelligence (AI) has been shaping its lab/office space portfolio based on activity by several of its tenant life sciences companies.
For example, insitro—a privately-held machine learning (ML)- and AI-enabled drug discovery and development company that Alexandria first nurtured by investing in the company’s Series A financing in 2018, and later through participation in subsequent financings—is building a mini AI data center within Alexandria’s Alexandria Center® for Advanced Technologies-South San Francisco, where the company leases space. In addition, insitro said, it also has an extensive computational footprint of CPUs and GPUs with off-premises / cloud computational partners.
Alexandria’s executive chairman and founder Joel S. Marcus, joined by Whitney Snider, MD, MBA, vice president of science and technology for Alexandria Venture Investments, the venture capital arm of ARE, discussed the transformative impact of AI on the life sciences industry, and on the publicly traded real estate investment trust (REIT), in an exclusive interview with GEN Edge.
(This interview has been lightly edited for length and clarity; Part II will appear separately).
GEN Edge: During Alexandria’s three decades in business, you have seen numerous technologies disrupt life sciences real estate. How similarly transformative is AI?
Joel Marcus: You have to think of it in really two buckets. One is AI as it relates to drug discovery, the other is AI as it relates to clinical trials. AI is not new to the pharmaceutical/biotech/life science industry. We were involved in the early startup of one of the first AI biology companies, insitro in South San Francisco, back in 2018. A number of companies have emerged over the past few years combining the thinking, and certainly the algorithmic approach to drug discovery.
People think, ‘[AI] just happened. It’s new today.’ But it really isn’t. It’s evolving and becoming a much more perfected way of doing things.
Whitney Snider: Over time, AI technology has truly become more sophisticated, and so does its potential to positively impact R&D for biopharma. The power of AI, ML, and biotech is truly derived from the large volumes of high-quality experimental data that’s generated by scientists in the laboratory.
These algorithms are trained on large and complex lab drive datasets that uncover patterns and scientific insights that then informs the next set of laboratory experiments. I think that’s really important, because this iterative process—which one of our tenants, Genentech, likes to call “Lab in a Loop”—means that the AI models themselves will likely become more standardized over time. and the lab derived datasets for the model training are really what is going to generate the value for biopharma with AI.
How that translates into research and development and infrastructure, the complexity and volume of these datasets, really requires integrated research and development capabilities. This results in more specialized technical lab space requirements. As the first and foremost provider of life science laboratory space, Alexandria has worked with our portfolio of companies and our tenants and investments on the cutting edge of AI to meet the market for their requirements in this space.
As we think about efficiency in drug development and more shots on goal, we can think about AI as we do all other technologies that have driven significant increases in efficiency and productivity in the workplace; high-speed processing or supercomputing are obviously good examples. AI should make drug discovery and development faster, cheaper and more efficient, reducing the time and costs in the overall R&D process to develop a drug. In helping solve this R&D productivity issue, AI will increase the volume of inquiries into new drug candidates across various modalities for a broad range of disease indications, which will ultimately increase the demand for lab space.
Marcus: What we’re doing for insitro, one of our anchor tenants in South San Francisco in one of our key mega campuses there, is building a mini-data center—not a full building, but something in the range of 10–15,000 square feet that’ll be an integrated data center for that operation, which then combines AI as it applies to the biological work they’re doing to become discovers of new molecules.
This adds another dimension, which then makes tenants feel very comfortable. They don’t have to go outside and necessarily build it themselves, or find a third-party operator in Nevada or Arizona. They can actually have it embedded in their own operations. So, it’s another piece of the infrastructure puzzle that I think helps our business and certainly enables them.
GEN Edge: How much is the insitro mini data center a model for other life-sci companies looking to add AI going forward?
Marcus: I think it will absolutely be a model. I think you’ll see this replicated by other companies, for sure.
GEN Edge: insitro has been into AI for several years. How was Alexandra able to meet their needs?
Marcus: They’ve been a longtime tenant since they started back in 2018, so we are just expanding them and enhancing the campus. In our discussions, the data center came up. And we said, ‘We ought to just do it, embed it in the current campus.’ They felt that was a great solution to have an integrated solution on their campus as opposed to going outside and trying to figure that out. So it’s a really nice win-win for both sides.
The meeting that led to this embedding of a mini-data center in the campus took place last October. We’ve been working on it since then.
GEN Edge: Does this mini-data center expand their space? Or did this take the place of some other facility that they used to have?
Marcus: A bit of both. It’s a little like cell and gene therapy—specialized, very unique modalities that have emerged over the last handful of years. As part of that, they just have a very different integration with R&D, and manufacturing. You’d like to keep it all together on campus. You’d like to have those people interact together, as opposed to having some people out and some in. That’s the big trend we see; it’s trying to keep all these functions internally for proprietary confidentiality reasons, and so forth. It makes great sense.
GEN Edge: You spoke about AI services being more standardized over time for life science companies. How does Alexandria add value?
Snider: The models themselves potentially over time will become more standardized, and that is because the datasets are truly what are proprietary for biotech companies, and that high quality experimental data is generated by scientists in the laboratory. As the AI models make drug discovery clinical development more efficient, we see companies able to have more shots on goal, more opportunities for pipeline development, and that requires additional scientists and additional lab space.
This is not just commoditized lab space. This is really technical high-end space that allows for specialized experimentation. And this infrastructure is really what powers AI models. The AI models themselves have been around a long time. It’s the data generation that laboratory scientists do that really power these requirements. We have been working with companies on the cutting edge of AI, to meet their requirements. I think insitro is a great example.
We have other examples beyond drug discovery. Eikon Therapeutics is working on new biological discoveries, harnessing a novel imaging platform to visualize thousands of protein interactions within live cells. Their work generates 1.5 petabytes of data a week. This is an astounding amount of data. They apply AI to this vast amount of data to elucidate or show molecular interactions with scale and precision. And they have incredibly technical laboratory needs with this specialized imaging platform that they have. So, it’s more about the data that they’re generating that they apply AI to than the AI itself. That’s where I think we have been able to help these cutting-edge investments and tenants with their key works.
Marcus: Eikon is run by [president, CEO, and chairman] Roger Perlmutter [MD, PhD, who previously headed R&D at Amgen and later Merck & Co.]. Roger’s probably one of the most successful drug hunters in the industry, really renowned, and a great human being. We’ve been host to them in our labs, both in New York and in the Bay Area. We’re building a new campus for them in Millbrae, CA, just south of South San Francisco. At the moment, they have a gigantic data center over in Hayward, CA. Over time, maybe the issue might be to bring that on campus, and maybe miniaturize it to the extent that that can be done with new modalities in the computing area.
There’s a lot going on that is beyond what people see in the drug discovery world. Everybody has a bit of a different technology base and approach. But I think the answer is, we’re going to have better drugs, faster and more capable on the personalized medicine side of addressing individual DNA and mutational differences than we’ve ever had before. I think that’s going to be really good news.
GEN Edge: Has the Eikon facility been built yet?
Snider: It has. They have two programs headed into the clinic. It’s a renowned company with incredible leadership. And they’re moving into clinical trials, meaning there’s near-term potential for patients, which is exciting. The goal of this faster development is really to help bring more therapies to patients, in an efficient, cost-effective way. That’s why we see this as an incredibly exciting area.
AI applies to more than just drug discovery. We see massive potential in clinical trial recruitment, in predicting side effects of medicines. And this ultimately will allow biopharma companies to expand their pipelines and be able to more efficiently drive capital and time and human capital into selecting the right drugs for the right patient subsets.
Marcus: Imagine you’re doing a cardiovascular or neurodegenerative clinical trial, you may have thousands of patients, many of whom probably aren’t going to be useful to the outcome of the trial. To the extent you could use AI to eliminate non-responding patients, have fewer people, but those who respond much more targeted, that has the opportunity to save the industry both time and cost. That has nothing to do with space, of course, because those trials are done generally in a hospital setting.
Snider: We see the efficiency in clinical trials as really important, because there’s only so much capital that biopharma have to put into each program. The programs in a pipeline for a company in some respects take capital from each other. If we can make the process more efficient, select better patients, we see this as expanding the need for laboratory-derived science. An overall efficiency of the process will allow scientists and developers to focus their time and energy into novel science. That that will be really fruitful for the whole industry.
GEN Edge: Some companies have attempted to quantify savings in terms of time and money. Does Alexandria have its own estimates on that?
Marcus: I don’t know that you can. At this point, I think it’s company specific and drug program specific. I don’t think you can generalize yet, because I don’t think we’ve had enough experience to say, ‘Well, over the last five years, 15 companies have had 30 programs, and here’s what the outcome is.’ So, I don’t think we’re able at the moment to make any kind of averages or generalizations. I think we’re still too early.
GEN Edge: How is the demand for AI affecting the demand for wet lab space? Does that mean less wet lab space being developed, which presumably would save capital costs for Alexandria in terms of TI (tenant improvement allowances)?
Marcus: It’s actually the opposite, which is a good thing.
Snider: AI ultimately increases the demand for laboratory space, the reason being AI and ML models are run on large volumes of high-quality experimental research data. It’s data that has to be done by scientists in a laboratory. It’s things that cannot be done on a computer. As AI becomes more powerful and efficient, the way to do that in drug development is really to train on incredible laboratory datasets. We actually see an increased need for these companies to have broader research and development scientific teams. And they’re able to take this laboratory data and analyze it more efficiently.
Marcus: Essentially, what it means is, if you think about it in the simplest of terms, it’s going to give companies with the same technology abetted by AI to have more drug shots on goal. So, they’ll have more opportunities to create a wider variation of drugs for a wider set of patients. And that means more biology work and more lab space, because the FDA is going to require that. But that’s good.
Snider: There are lots of examples of this. A leader in AI applications in biopharma is Recursion Pharmaceuticals. They have both massive computational scale and massive experimental scale. They do millions of wet lab experiments weekly, because that feeds into their AI models. They are not a tenant, but they have 200,000 square feet of lab and office space in Salt Lake City, where its headquarters are. That’s just another example of wet lab work feeding into these AI models.
GEN Edge: What changes has the rise of AI necessitated in Alexandria’s service to tenants, or, for that matter in what you offer, whether through lab space or other efforts?
Marcus: We haven’t seen the full realm of that yet. I think you’ll see more need for biology wet lab space, and I think you’ll have embedded data centers in the R&D centers. I think those are probably the two big things.
Continue reading Part II of this exclusives interview.
