Boston’s AI-biotech story is usually told through startups.
Generate is using generative AI to design proteins. Lila Sciences wants AI systems to run parts of the scientific process itself. Dyno Therapeutics is using machine learning to design better gene-delivery vectors. Nabla Bio is applying AI to antibody design. Cellarity is using computational models to understand disease through changes in cell behavior.
But there is another side of the story that matters just as much.
Who is willing to pay for all of this?
Building an AI software startup can already be expensive. Building an AI company that also needs scientists, wet labs, biological data, robotics, clinical studies and years of drug development is on another level.
That changes the kind of investor these companies need.
A normal software investor may be comfortable funding engineers, cloud bills and sales teams for 18 months. An AI-biotech investor may have to accept years of laboratory work before meaningful revenue appears. If the company develops its own medicines, the investor may also need to understand clinical trials, regulatory risk, manufacturing and the possibility that a promising drug simply fails.
Boston has developed an unusually deep pool of capital willing to take that risk.
Our analysis of ten Boston-area companies at the intersection of computation, AI and biology found roughly $2.40 billion in disclosed private financing before counting public offerings. More important, the investor list is not dominated by one kind of fund.
The money comes from Boston biotech builders such as Flagship Pioneering and Third Rock Ventures. It comes from traditional technology investors such as Bessemer, Andreessen Horowitz, Lux Capital and Khosla Ventures. It comes from healthcare specialists including RA Capital, BVF Partners, Perceptive Advisors and Casdin Capital. It comes from giant financial institutions such as Fidelity, BlackRock and T. Rowe Price. It increasingly comes from strategic investors such as NVIDIA, Amgen, Eli Lilly and other large corporations.
That combination tells us something important.
Boston’s AI-biotech boom is becoming large enough that it no longer belongs to a single venture-capital category.
The strongest companies are pulling together several capital markets at once.
And that may become one of Boston’s biggest advantages.
The Short Version: Boston Has Built a Four-Layer AI-Biotech Capital Stack
Boston does not have one group of investors funding computational biology.
It has several.
The first layer consists of company builders and specialist biotech funds. These investors understand how to create biotech companies, finance laboratory work and survive long development cycles.
The second layer consists of AI and technology investors. They understand foundation models, software platforms, data advantages, compute economics and the possibility that a scientific company can scale more like a technology platform than a conventional biotech.
The third layer consists of large crossover and institutional investors. They tend to arrive once companies need hundreds of millions of dollars rather than tens of millions.
The fourth layer is strategic capital. NVIDIA, Amgen, Eli Lilly, Analog Devices and other corporations are appearing in financing syndicates because computational biology affects their own businesses.
That creates a very different funding environment from the Boston biotech market of twenty years ago.
Our ten-company sample produced three particularly important findings.
First, capital is extremely concentrated. Generate, Lila Sciences and Relay Therapeutics represent about 73.6% of all private financing in the sample.
Second, Flagship Pioneering is the clearest repeat company-building investor, appearing across four companies in our core dataset: Generate, Lila, Cellarity and Montai.
Third, NVIDIA’s venture arm has become one of the most interesting new connectors between AI infrastructure and biology. NVentures appears in Generate, Lila and Superluminal Medicines.
This suggests Boston is not simply developing more biotechnology companies that happen to use AI.
It is building a financing system specifically suited to companies where software, models, experiments and medicine development increasingly belong inside the same business.
How BostonBizScene Built This Analysis
Because phrases such as “AI biotech” and “computational biology” can mean almost anything, we did not simply search funding databases for every Boston biotech that mentions artificial intelligence.
That would create a misleading dataset.
Instead, we built a focused sample of companies where computation or AI is central to how the company discovers, designs or evaluates biology.
The ten companies in our core sample are Generate, Lila Sciences, Relay Therapeutics, Cellarity, Dyno Therapeutics, Superluminal Medicines, Montai Health, Manifold Bio, Nabla Bio and Converge Bio.
We included disclosed private financing through October 7, 2026. We excluded IPO proceeds from the main venture analysis because public offerings represent a different source of capital. We also excluded rumored or proposed financings that had not been publicly confirmed as closed.
That distinction matters.
For example, Generate subsequently raised substantial public-market capital, but including that money in a study of venture investors would distort the question we are trying to answer. Relay also completed a $400 million IPO in 2020, but its public offering is not counted in our private-capital total.
The numbers should therefore be understood as a disclosed private-capital floor, not a claim that every dollar ever invested in every company has been captured perfectly. Some companies have undisclosed early financing, and investor disclosures are not equally complete across every round.
The aim is not false precision.
The aim is to understand who repeatedly funds this market, which types of investors appear at different stages, and how concentrated Boston’s AI-biotech capital has become.
Our Boston AI-Biotech Funding Sample
| Company | Boston-area base | Disclosed private capital used in our analysis | Core computational angle |
| Generate | Somerville | ~$693M | Generative protein design |
| Lila Sciences | Cambridge | $550M | AI-driven autonomous science |
| Relay Therapeutics | Cambridge | $520M | Computational protein motion and drug design |
| Cellarity | Greater Boston | ~$274M | Computational modeling of cell states |
| Dyno Therapeutics | Greater Boston | ~$109M | AI-designed gene-delivery vectors |
| Superluminal Medicines | Boston | $93M | AI/ML and structure-based drug discovery |
| Montai Health | Cambridge | $50M | Computational discovery from natural chemistry |
| Manifold Bio | Boston | $40M | Data-rich biologics design and in-vivo testing |
| Nabla Bio | Boston/Cambridge | ~$37M | AI-first antibody design |
| Converge Bio | Boston | $30M | Generative AI tools for biotech R&D |
| Total | ~$2.396B |
Generate announced a $370 million Series B in 2021 and a $273 million Series C in 2023. The Series C brought total equity financing since 2020 to nearly $700 million.
Lila closed a $350 million Series A in October 2025 after an earlier $200 million seed financing, taking total capital raised to $550 million.
Relay reported that it had raised $520 million privately by March 2019 after completing a $400 million Series C.
Cellarity said its $121 million Series C brought total funding at the time to $274 million.
Dyno has disclosed roughly $109 million, including a $100 million Series A led by Andreessen Horowitz.
Superluminal launched with $33 million in 2023 and added a $60 million Series B in September 2026, producing $93 million from those two disclosed rounds.
Nabla announced an $11 million seed round and later a $26 million Series A, putting those disclosed rounds at about $37 million.
Converge Bio’s January 2026 Series A brought its reported total to $30 million.
Original Finding #1: Three Companies Represent Nearly Three-Quarters of the Capital
The first finding is hard to miss.
Boston has many computational-biology startups, but its capital is not evenly distributed among them.
Chart: Share of Our $2.396 Billion Private-Capital Sample
| Company | Approx. share |
| Generate | 28.9% |
| Lila Sciences | 23.0% |
| Relay Therapeutics | 21.7% |
| Cellarity | 11.4% |
| Dyno Therapeutics | 4.5% |
| Superluminal Medicines | 3.9% |
| Montai Health | 2.1% |
| Manifold Bio | 1.7% |
| Nabla Bio | 1.5% |
| Converge Bio | 1.3% |
Generate, Lila and Relay together account for approximately $1.763 billion, or 73.6%, of the entire sample.
Add Cellarity and Dyno, and the five largest companies represent about 89.6%.
That means fewer than half of the companies capture almost nine out of every ten dollars represented in the dataset.
The average amount raised is about $239.6 million, while the median is only about $101 million. The large difference between those figures is another signal that a handful of unusually well-funded companies are pulling the average upward.
This matters for founders because the headline “billions are flowing into AI biotech” can create the wrong impression.
The money is available.
It is not evenly available.
Investors appear willing to commit extraordinary amounts of capital once they believe a company has moved beyond an interesting model into a repeatable scientific platform with enough proof to support a large pipeline.
Why Generate Could Raise Hundreds of Millions
Generate is probably the clearest example.
Its $370 million Series B included Flagship Pioneering and a group of major financial and life-sciences investors including ADIA, the Alaska Permanent Fund, Altitude Life Science Ventures, ARCH Venture Partners, Fidelity, Morningside and T. Rowe Price.
Two years later, Generate raised another $273 million.
The Series C brought in new investors including Amgen, NVIDIA’s NVentures, MAPS Capital and Pictet Alternative Advisors. Flagship and the major Series B investors participated again.
This is not what the cap table of an ordinary early biotech looks like.
It is closer to a coalition.
Generate attracted a biotech company builder, institutional money, major asset managers, a chip-company venture arm and a pharmaceutical strategic investor.
Why?
Because the company was no longer asking investors to finance one scientific idea.
It was asking them to finance a machine for producing many programs.
That distinction is becoming central to AI-biotech financing.
Original Finding #2: Boston AI Biotech Has Four Distinct Types of Capital
Looking only at company funding totals hides one of the most interesting patterns in the data.
The investor groups are unusually diverse.
We can roughly divide them into four categories.
| Capital type | Examples appearing in Boston AI-biotech rounds | What they bring |
| Biotech builders and specialists | Flagship, Third Rock, RA Capital, BVF, Perceptive, Casdin, Foresite | Drug-development expertise and sector knowledge |
| AI and technology investors | Bessemer, a16z, Lux, Khosla, Radical, Zetta, General Catalyst | Software, AI, models, platforms and scaling knowledge |
| Large institutional/crossover capital | Fidelity, T. Rowe Price, BlackRock, ADIA, Pictet, Baupost | Large checks and later-stage financial capacity |
| Corporate/strategic capital | NVIDIA, Amgen, Lilly, Kyowa Kirin, Analog Devices | Industry access, infrastructure and strategic validation |
The categories overlap.
That is the point.
The most ambitious Boston companies increasingly need investors that understand several different parts of the business.
A computational biology company may need the mentality of a software startup during model development, the infrastructure discipline of a cloud company, the scientific discipline of a biotech and the capital structure of a clinical-stage drug developer.
Very few investors can cover all of those areas alone.
So syndicates are becoming more important.
Flagship Pioneering Is Still the Most Important Local Company-Building Machine
If there is one investor whose fingerprints are visible across the Boston AI-biotech landscape, it is Flagship Pioneering.
In our ten-company core dataset, Flagship appears across four companies: Generate, Lila, Cellarity and Montai.
That is not simply because Flagship bought shares after those companies became successful.
Flagship helped create them.
Generate was launched by Flagship. Cellarity was developed inside Flagship Labs. Montai was unveiled with an initial $50 million commitment from Flagship. Lila also originated from Flagship and received major funding from it.
This gives Flagship a different role from most traditional venture firms.
It Is Building the Supply of Companies
A normal VC may wait for founders to pitch.
Flagship’s model often starts earlier. It develops scientific ideas, forms companies around them, recruits leadership, supplies initial capital and then brings outside investors into later rounds.
That makes Flagship important to Boston’s ecosystem for a reason beyond its checkbook.
It manufactures investable companies.
That is especially powerful in AI biotech because many of the best opportunities begin as scientific questions rather than polished startup ideas.
Someone has to fund the years between “this might work” and “this deserves a $200 million financing.”
Flagship is designed to operate in that gap.
Lila Shows How Far the Model Can Be Pushed
Lila Sciences may be the most ambitious example yet.
The company raised $200 million at seed and then closed a $350 million Series A, reaching $550 million in total capital. Its investor group included Flagship, General Catalyst, March Capital, ARK Venture Fund and Altitude Life Science Ventures, while Braidwell and Collective Global co-led part of the Series A. NVentures later joined alongside Analog Devices, IQT, Catalio Capital Management and others.
Think about what that investor list says.
This is not only life-sciences capital.
It includes AI investors, financial institutions, strategic technology interests and investors that can support extremely large infrastructure projects.
That makes sense because Lila is not presenting itself as a normal therapeutic startup. It is trying to build AI systems and highly automated scientific infrastructure capable of running experiments across life science, chemistry and materials.
The investor base therefore begins to look less like the syndicate behind one drug company and more like the financing behind a new scientific computing platform.
Original Finding #3: NVentures Has Quietly Become One of the Most Important Bridges Between AI and Boston Biotech
One of the most interesting names in our network analysis is NVentures, NVIDIA’s venture-capital arm.
Within the ten-company dataset, NVentures appears in three companies: Generate, Lila and Superluminal.
Only Flagship appears across more of the sampled companies.
That deserves attention.
NVIDIA is not a traditional drug-development investor. Its importance comes from the infrastructure beneath modern AI.
As biological models become larger and more computationally demanding, the relationship between biotechnology and compute becomes much tighter.
Generate added NVentures in its Series C. Lila included NVentures among the new supporters disclosed when its $350 million Series A closed. Superluminal has also counted NVIDIA among its investors since its original $33 million financing.
This is a small sample, so it would be wrong to conclude that NVIDIA controls Boston AI biotech.
But the direction is worth watching.
The companies building the infrastructure for AI increasingly have strategic reasons to help create the workloads that will consume that infrastructure.
Biology could become one of the largest of those workloads.
Why AI Infrastructure Companies Care About Biology
A frontier biological model can require large amounts of compute during training.
It can also require substantial compute during everyday scientific use.
If pharmaceutical companies eventually run protein models, molecular models, cell models and scientific agents continuously across research programs, life science becomes an important computing market.
That means a company such as NVIDIA can benefit even when it never develops a drug.
More AI-based scientific work means more demand for accelerated computing.
The investment logic therefore runs in both directions.
Boston startups need compute infrastructure.
Compute companies need valuable new AI workloads.
Biology offers one.
Original Finding #4: Repeat Investors Reveal the Real Network Better Than One-Off Mega-Rounds
Large fundraising announcements get attention, but repetition can tell us more about an investor’s conviction.
We reviewed the disclosed investor lists across the ten companies and grouped investors at the organization level where appropriate.
Investors Appearing Across Multiple Companies in the Sample
| Investor | Companies represented in sample |
| Flagship Pioneering | 4 |
| NVentures/NVIDIA | 3 |
| March Capital | 2 |
| ADIA | 2 |
| Altitude Life Science Ventures | 2 |
| Pictet-related investment entities | 2 |
| Catalio Capital Management | 2 |
| BVF Partners | 2 |
| Perceptive Advisors | 2 |
| Casdin Capital | 2 |
| GV | 2 |
| Fifty Years | 2 |
This list reveals an important feature of Boston’s market.
There is no single investor clique.
Some repeat investors come from traditional biotech. Others come from AI. Others manage enormous pools of global capital.
That creates a network with multiple entry points.
The Local Ecosystem Is Importing Capital
Boston supplies many of the companies and scientists, but the money does not need to come entirely from Massachusetts.
Generate pulled capital from global institutions.
Dyno’s $100 million Series A was led by Andreessen Horowitz and included Casdin Capital, GV, Obvious Ventures, Lux Capital, Polaris, CRV and KdT Ventures.
Nabla’s $11 million seed round was co-led by Khosla Ventures and Zetta Venture Partners, while its later $26 million Series A was led by Radical Ventures.
Manifold Bio’s $40 million Series A was led by Triatomic Capital and included Section 32, FPV Ventures, Horizons Ventures, Tencent, Playground Global, Fifty Years and FAST by GETTYLAB.
Converge’s $25 million 2026 Series A was led by Bessemer with TLV Partners, Vintage Investment Partners and Saras Capital participating.
Boston therefore does not need to finance its AI-biotech boom entirely from Boston.
Its bigger advantage may be that companies founded around Boston science are capable of pulling international capital into the region.
Third Rock Shows the Older Version of the Company-Building Model
Flagship is not the only Boston-area firm with a long history of forming biotechnology companies.
Third Rock Ventures played a foundational role in Relay Therapeutics.
Relay was launched in 2016 with financing from Third Rock and an affiliate of D. E. Shaw Research. By the time the company completed its enormous $400 million Series C in 2018, it had built a much broader syndicate.
The round was led by SoftBank Vision Fund, with new investment from Foresite Capital, Perceptive Advisors and Tavistock Group. Existing investors included GV, Casdin Capital, BVF Partners, EcoR1 Capital, Alexandria Venture Investments and D. E. Shaw Research.
Relay is useful because it demonstrates that computation and drug discovery were already converging before the current generative-AI wave.
The difference today is scale.
AI is becoming much more central to the company narrative, the computing requirements are larger, and technology-focused investors are participating much earlier.
RA Capital, BVF and Perceptive Matter When AI Has to Become Medicine
At some stage, an AI-biotech startup faces a difficult transition.
The model stops being the whole story.
A drug has to work.
That changes which investors become particularly useful.
Boston-based RA Capital is one example. It led Superluminal’s $33 million launch financing. That round also included Insight Partners, NVIDIA and Gaingels.
By September 2026, Superluminal was ready for another step.
The company raised an oversubscribed $60 million Series B led by BVF Partners, with Deep Track Capital and Perceptive Advisors joining alongside existing investors including RA Capital, Insight, NVIDIA, Catalio, Eli Lilly, Cooley and Gaingels. The financing is intended in part to move its lead program into Phase 1 testing.
Look at how the investor group evolves as the company evolves.
Early funding supported the platform and pipeline.
Later financing has to support clinical execution.
That is exactly why specialist healthcare investors remain important even in an AI-heavy market.
AI expertise cannot replace drug-development judgment.
Original Finding #5: The Investor Mix Changes When a Company Crosses From Platform Risk to Clinical Risk
This is one of the most important patterns for founders to understand.
An AI-biotech company does not face one kind of risk.
It moves through several.
Stage 1: Can the Technology Work?
At the earliest stage, investors are deciding whether the model, dataset, experimental system or computational approach can produce something valuable.
AI investors can be especially valuable here because they understand technical architecture and model improvement.
Stage 2: Can the Platform Produce Repeated Results?
Once the basic technology works, investors want to know whether it works repeatedly.
Can the system create several good molecules rather than one?
Can it improve after each experimental cycle?
Can it generate data competitors do not have?
Can scientists use it in normal workflows?
That is where platform-focused venture firms become especially important.
Stage 3: Can the Science Survive the Real World?
This is where biology becomes unforgiving.
A model may predict something correctly.
A molecule may work in a laboratory assay.
That does not guarantee it works in animals or humans.
Healthcare-specialist investors become valuable because they understand the attrition.
Stage 4: Can the Company Finance Clinical Scale?
Clinical studies are expensive.
If a company owns several programs, capital requirements can rise quickly.
That is where large funds, crossover investors and institutional pools can become necessary.
Generate’s history demonstrates that progression.
Relay’s history does as well.
Superluminal is moving through it now.
Bessemer’s Converge Investment Shows Another AI-Biotech Business Model
Not every computational biology company needs hundreds of millions of dollars.
Converge Bio provides a useful contrast.
The Boston- and Tel Aviv-based company raised $25 million in January 2026, bringing total reported funding to $30 million. Bessemer Venture Partners led the financing, with TLV Partners, Vintage Investment Partners and Saras Capital participating.
Converge is selling AI tools to biotechnology and pharmaceutical companies rather than relying only on a wholly owned pipeline.
At the time of the financing, the company said more than a dozen pharma and biotech customers were using its systems for areas such as target discovery, antibody design and protein-manufacturing optimization.
That changes the capital model.
A software-like AI-biotech company may reach customers and revenue earlier than a company that has to finance clinical trials.
The tradeoff is that the upside may also be structured differently.
An internal drug program can eventually become worth billions if successful.
Software revenue may be more predictable, but a platform vendor captures a different share of the economics.
Investors are therefore financing several business models under the AI-biotech label.
Nabla Shows That Huge Funding Is Not Always Required
Nabla Bio makes another important point.
The company raised $11 million in a seed financing co-led by Khosla Ventures and Zetta Venture Partners, with Fifty Years and Cantos participating. It later raised a $26 million Series A led by Radical Ventures, while existing investors including Khosla and Zetta participated again.
Its total disclosed capital remains tiny next to Generate or Lila.
But financing size alone does not tell us whether a platform has commercial value.
Nabla has also announced relationships with major pharmaceutical companies. Zetta reported that agreements with AstraZeneca, Bristol Myers Squibb and Takeda carried more than $550 million in potential upfront and milestone economics plus royalties, although those headline values should not be confused with guaranteed revenue.
This creates an important alternative funding path.
Instead of raising hundreds of millions before proving commercial demand, a company can use partnerships to validate the platform and potentially shift part of the financing burden toward customers.
The Best AI-Biotech Financing May Mix Equity With Partnerships
For founders, this matters enormously.
Equity is expensive.
Every financing dilutes ownership.
Biotechnology companies often need enough capital that repeated equity rounds can significantly change who owns the company.
Strategic partnerships can provide another source of economics.
The best computational biology companies may therefore build a financing model using several channels at once:
venture capital to create the platform, strategic partnerships to validate and partly fund applications, later-stage capital to finance owned programs, and eventually public-market capital if the company becomes mature enough.
Generate has also pursued large pharmaceutical collaborations alongside its equity financing.
That is not an accident.
A platform capable of producing many programs should have more than one way to monetize them.
Corporate Venture Capital Is Becoming More Strategic
Corporate participation deserves its own category because companies invest for reasons that may differ from ordinary venture funds.
A VC ultimately wants financial returns.
A corporate investor may want financial returns too, but it can also care about access, learning, infrastructure demand or strategic positioning.
Generate’s investors have included Amgen and NVentures.
Superluminal’s investor base includes NVIDIA and Eli Lilly.
Cellarity’s Series C brought in strategic investors including Kyowa Kirin and Hanwha Impact Partners.
Lila has brought in NVentures and Analog Devices alongside a much broader set of financial investors.
These are not identical relationships.
But together they suggest that large companies increasingly see computational biology as important enough to monitor from the inside.
An investment gives them a seat closer to the technology.
The NVIDIA Pattern Is Especially Different
NVIDIA’s role deserves to be separated from pharmaceutical strategic investors.
A pharmaceutical company may invest because it wants better drugs.
NVIDIA benefits when more science becomes computational.
That gives it exposure across categories.
Protein models need compute.
Chemistry models need compute.
Autonomous laboratories need AI systems.
Scientific agents need inference infrastructure.
Large pharmaceutical research organizations running thousands of model-driven experiments could eventually become significant computing customers.
So NVIDIA does not have to pick the ultimate winning drug.
It benefits from the broader computationalization of biology.
That is a powerful strategic position.
Boston’s Old Biotech Strength Is Pulling New AI Capital Into the City
This may be the most important ecosystem-level story.
Boston already had almost everything required to build biotechnology companies: universities, hospitals, research institutes, pharmaceutical offices, wet-lab infrastructure, specialist lawyers, experienced executives and venture firms that understood life science.
What it historically lacked was a reason for mainstream frontier-AI capital to care about biotechnology at the same level it cared about software.
That is changing.
AI investors now see biology as a model-building problem.
Biotech investors see computation as a way to increase discovery productivity.
Large institutions see companies maturing enough to absorb very large checks.
Strategic technology companies see a massive future computing market.
Those forces meet unusually well in Boston.
Why General Catalyst Matters
General Catalyst illustrates the convergence.
The firm invested in Lila and publicly framed the investment around the idea of AI changing the scientific process itself.
Its thesis was not simply that AI might improve one pharmaceutical task.
It was that scientific discovery could increasingly involve AI systems capable of processing information, generating hypotheses and helping drive experimentation.
That is a much larger market thesis.
And it helps explain why major generalist technology funds are increasingly willing to enter an area historically left to specialized biotech investors.
If AI changes the entire way scientific work gets done, computational biology starts looking less like a niche inside biotechnology.
It begins to look like an enormous technology platform opportunity.
Original Finding #6: Boston’s Funding Advantage Is Not Only Local Money—It Is Local Credibility
It would be easy to measure Boston’s strength by counting Boston-headquartered venture firms.
That misses the more valuable advantage.
Boston companies can attract investors from almost anywhere because the region already carries scientific credibility.
Consider the investor geography represented across our sample.
There are Silicon Valley firms.
There are Boston investors.
There are Canadian investors.
There are Asian corporations.
There are Middle Eastern institutional pools.
There are global asset managers.
There are corporate venture groups.
Capital is mobile.
Scientists, laboratories, company-building experience and dense research networks are less mobile.
Boston’s advantage may therefore be its ability to make global money feel comfortable financing local science.
That is stronger than simply having a lot of local VC firms.
What Investors Actually Seem to Be Paying For
Strip away the AI language and a clear pattern appears.
Investors are not merely financing better prediction.
They are financing closed learning systems.
The most attractive companies increasingly connect several stages:
data enters the system, a model proposes an answer, an experiment tests it, new data comes back, and the system becomes better at the next decision.
That loop can exist in protein design.
It can exist in antibody engineering.
It can exist in gene delivery.
It can exist in cell-state biology.
It can exist in small-molecule discovery.
And eventually it can exist across automated laboratories.
The faster and cheaper that loop becomes, the more experiments a company can run.
The more experiments it runs, the more proprietary data it can produce.
The more proprietary data it produces, the harder the system can become to copy.
That is the flywheel investors want.
The Dataset Matters More Than the Demo
Founders entering this market should pay attention to one distinction.
A beautiful AI demo is not the same thing as a valuable scientific platform.
Investors will increasingly ask where the proprietary data comes from.
If the model relies mainly on public biological datasets, competitors may have access to much of the same information.
If the startup owns a laboratory that generates unique measurements, the situation changes.
Each experiment can create new training information unavailable elsewhere.
That is one reason the Boston ecosystem is so interesting.
Wet labs are not separate from the AI moat.
They can be part of the AI moat.
Why Manifold Bio Fits This Thesis
Manifold Bio’s financing illustrates the idea.
The company raised a $40 million Series A led by Triatomic Capital, with Section 32, FPV Ventures, Horizons Ventures and Tencent among the new investors. Existing investors included Playground Global, Fifty Years and FAST by GETTYLAB.
Its platform is designed around increasing the amount of biological information that can be collected during drug development.
That matters because machine learning improves when the feedback loop becomes richer.
The company that can measure more biological outcomes does not merely test more candidates.
It can potentially build a better learning system.
Investors are increasingly funding the measurement infrastructure behind AI, not only the AI model itself.
Why Dyno Attracted a16z, Lux and GV
Dyno offers another clean example of how technology capital enters biology.
Its $100 million Series A was led by Andreessen Horowitz. Other investors included Casdin Capital, GV, Obvious Ventures and Lux Capital, while founding investors Polaris Partners, CRV and KdT Ventures also participated.
Dyno applies machine learning to the design of AAV vectors used in gene therapy.
The core economic idea is easy to understand even without knowing the biology.
Better delivery can increase the number of diseases gene therapy can address.
If AI helps search a huge design space faster than older laboratory methods, the platform becomes strategically valuable to many drug developers rather than one program.
Again, platform breadth attracts technology investors because the potential market is larger than a single medicine.
Original Finding #7: Boston’s AI-Biotech Funding Market Is Splitting Into Two Business Models
One useful way to understand the next five years is to divide companies into two broad groups.
Model A: AI as a Better Drug Company
These businesses use computation to create their own therapeutic pipeline.
Generate, Relay, Superluminal, Cellarity and others fit much of this pattern.
AI may improve discovery, but eventually the company has to behave like a drug company.
It needs toxicology.
It needs manufacturing.
It needs clinical operations.
It needs regulatory strategy.
It needs patients.
The capital requirement can become enormous.
Model B: AI as Scientific Infrastructure
Other companies can sell models, tools, laboratory systems or discovery capabilities to many organizations.
Converge is closer to this model.
Lila could develop into an especially ambitious version of it if autonomous scientific infrastructure becomes broadly usable by outside customers.
Infrastructure businesses may be able to monetize earlier.
But the strongest companies could combine both models.
They might sell platform access to partners while retaining selected internal programs where they want the full economic upside.
That hybrid model could become common.
What This Means for Founders Raising Money in Boston
The biggest mistake a founder can make is choosing investors based only on who offers the highest valuation.
AI-biotech companies have unusual financing needs.
A useful investor needs to help the company survive several completely different stages.
Choose Investors for the Next Risk
If your immediate problem is proving the technical system works, an investor with real AI knowledge may add more value than a giant pharmaceutical fund.
If your platform already works and the next risk is entering the clinic, an investor who has financed dozens of clinical programs may be much more valuable.
If your company is approaching a $100 million financing, you need investors capable of helping assemble large syndicates.
The right investor depends on the next bottleneck.
Do Not Raise a Clinical-Stage Budget Before You Have Earned It
The giant Boston rounds can create unhealthy expectations.
A founder sees Lila raising $550 million or Generate raising hundreds of millions and concludes that bigger must be better.
That is dangerous.
Capital is useful only when the company can turn it into proof.
A startup that raises $150 million before discovering its true product can build an expensive organization around assumptions that later prove wrong.
The better question is not, “How much can we raise?”
It is, “What evidence should the next dollar buy?”
The answer might be a better model.
It might be a proprietary dataset.
It might be ten pharmaceutical customers.
It might be an IND.
It might be a successful Phase 1 result.
Raise enough capital to reach a materially stronger financing position.
Build the Financing Strategy Around the Scientific Feedback Loop
For an AI-biotech company, the fundraising plan should be linked directly to the learning loop.
Suppose $10 million can expand laboratory throughput enough to generate ten times more useful biological measurements.
That could be a rational investment.
Suppose another $20 million simply doubles team size without increasing experimental velocity.
That may not be.
Founders should know how each major spending category affects the rate at which the company learns.
Investors increasingly will ask the same question.
Show Why Your Data Gets Better With Scale
The best AI-biotech pitch should explain why the company becomes harder to compete with every year.
More customers may create more data.
More laboratory experiments may generate better training sets.
More drug programs may expose the model to more biological situations.
More automation may increase experimental throughput.
If those pieces reinforce one another, explain the mechanism clearly.
Do not simply say you have a “data moat.”
Show how the moat grows.
What This Means for Boston Investors
The opportunity is also changing for local venture firms.
Historically, a strong Boston biotech investor could specialize heavily in clinical science and drug development.
That is no longer enough for every deal.
Investment teams increasingly need to evaluate questions such as model architecture, compute costs, training data, inference workflows, automation and software distribution.
Technology VCs face the opposite problem.
They may understand AI perfectly while underestimating how slowly biological risk disappears.
The winning investment teams will combine both disciplines.
The New Competitive Advantage Is Cross-Disciplinary Judgment
Evaluating these companies requires an unusual question:
Where does the actual value come from?
Is it the model?
Is it proprietary experimental data?
Is it the laboratory infrastructure?
Is it a biological insight?
Is it a drug pipeline?
Is it customer workflow integration?
Or is it the feedback loop connecting all of them?
The answer differs company by company.
Investors who treat every AI-biotech pitch as “AI plus drugs” will miss those differences.
Boston’s Next Capital Gap May Be the Middle
Boston appears capable of producing early scientific capital.
It can also attract enormous later rounds once a company becomes a clear category leader.
The harder part may be the middle.
Companies that have proven technical capability but have not yet reached strong clinical or commercial validation can require large amounts of money while still carrying substantial risk.
This is where disciplined Series A, B and C investing becomes important.
The market needs funds that understand both computational platform risk and biological development risk.
There may be a major opportunity for investors willing to specialize precisely in that transition.
Watch the $30 Million to $100 Million Range
Several interesting Boston companies sit in this zone.
They are no longer academic projects.
But they are nowhere near Generate-sized capital bases.
Converge has reported $30 million.
Nabla has disclosed roughly $37 million from its two headline venture rounds.
Manifold raised $40 million in its Series A.
Montai began with a $50 million Flagship commitment.
Superluminal moved from $33 million at launch to $93 million across its disclosed seed and Series B financings.
These companies will help reveal whether Boston can repeatedly turn promising computational platforms into mature businesses rather than merely produce a few giant winners.
What Investors Should Measure Beyond Funding
Funding is not success.
It is fuel.
The better questions are what happens after the financing.
How many experimental cycles can the company run?
How much proprietary data is generated?
How many programs reach meaningful biological validation?
How many pharmaceutical partners renew or expand relationships?
How much time does the platform remove from research workflows?
How often do computational predictions survive experimental testing?
How much does the next validated program cost compared with the previous one?
Those are the numbers that tell us whether AI is changing the economics of biotechnology.
Capital Efficiency Could Become a Major Differentiator
One of the most interesting long-term tests will be whether AI-biotech companies become more capital-efficient than traditional drug developers.
They do not necessarily start that way.
Training models, building automated laboratories and generating proprietary biological data can require heavy upfront spending.
But the promise is that the platform should eventually allow better decisions earlier.
A company that kills weak programs before expensive clinical development can save enormous amounts of money.
A company that designs stronger candidates with fewer cycles can save time.
A company that runs thousands of experiments automatically can increase output without expanding headcount at the same rate.
If those benefits become measurable, investors may eventually value scientific throughput in a similar way to how software investors value productivity or recurring revenue.
The Biggest Question Is Whether the Learning Curves Are Real
This is where the entire AI-biotech investment thesis will ultimately be tested.
Does every new experiment make the platform meaningfully better?
Does each program improve the next?
Does greater scale lower the cost of finding a successful medicine?
Can companies show increasing prediction accuracy or experimental success?
If yes, the economics of drug discovery could change.
If no, AI risks becoming another expensive tool layered on top of an already expensive process.
The difference is enormous.
Why Boston Is a Natural Place to Find Out
Boston has the right ingredients to test the thesis faster than most cities.
A computational scientist can work near pharmaceutical researchers.
A machine-learning founder can recruit biology talent from universities and hospitals.
A startup can access specialized wet-lab infrastructure without recreating an entire regional industry.
Experienced biotechnology executives already understand how to move therapies into clinical development.
Specialist investors already understand the risks.
Now technology capital is joining them.
That combination allows Boston to run more experiments at the company level as well as the laboratory level.
The Broader Capital Flywheel
The ecosystem can become self-reinforcing.
Successful companies create experienced employees.
Experienced employees become founders.
Successful exits produce angel investors.
VC firms develop better pattern recognition.
Pharma companies form more partnerships.
Infrastructure providers build more labs.
More investors become comfortable with computational biology.
That attracts more founders.
Boston has already seen this flywheel in biotechnology.
AI can accelerate it.
What Could Break the Boom?
The capital story is strong, but it is not risk-free.
A major clinical failure from a highly visible AI-designed drug could cool investor enthusiasm.
So could evidence that expensive AI platforms do not improve R&D productivity enough to justify their costs.
Funding markets could tighten.
Large pharmaceutical companies could decide to build more systems internally rather than partner with startups.
Models could also become commoditized, weakening companies whose only advantage is access to algorithms that competitors can reproduce.
The companies most protected from those risks will probably own something harder to copy.
That could be experimental data.
It could be automated laboratory infrastructure.
It could be deep workflow integration.
It could be proprietary therapeutic assets.
Most likely, the strongest companies will own several of those things together.
The Investors to Watch Next
The most interesting question for 2027 is not simply which fund writes the next check.
It is which investors begin appearing repeatedly.
Flagship already has a clear company-building footprint.
NVentures is emerging as a cross-company technology investor.
Healthcare specialists such as RA Capital, BVF and Perceptive remain important when platforms move toward clinical development.
AI-focused investors such as Bessemer, Radical, Lux, Khosla, Zetta and Andreessen Horowitz have demonstrated willingness to finance biology when the computational thesis is strong.
Global institutional capital becomes increasingly important when successful platforms need several hundred million dollars.
The next phase will show whether these groups remain separate or start competing directly for the same deals.
Boston Could Produce a New Type of Venture Firm
There is also room for an investment strategy built specifically around the gap between AI and biology.
Such a firm would need genuine depth in machine learning.
It would also need scientists capable of evaluating experiments.
It would need drug-development expertise.
And it would need the patience to support longer timelines than normal enterprise software.
Boston may be one of the few places where enough people with those backgrounds already work within a few miles of one another.
That could eventually change the venture industry itself.
A Practical Framework for Evaluating an AI-Biotech Startup
For investors looking at this market, the simplest framework is to examine four connected assets.
1. The Computational Engine
Does the technology actually outperform simpler approaches?
A company should be able to explain what its models do better, where the improvement has been measured and why competitors cannot simply reproduce it.
2. The Experimental Engine
How quickly can predictions be tested?
The value of an AI system drops if every useful answer then waits months for laboratory validation.
The model and experimental operation should increasingly be evaluated as one system.
3. The Data Engine
Does each experiment create information that improves future decisions?
If yes, the company may possess a compounding asset.
If no, it may simply have an efficient tool.
4. The Economic Engine
Who pays?
That question sounds obvious, but it is often ignored.
Does the company sell software?
License molecules?
Sign research partnerships?
Develop its own therapies?
Charge for scientific infrastructure?
Or combine several models?
An extraordinary scientific platform with no clear path to value capture can still become a poor venture investment.
Original Finding #8: The Most Valuable Investor May Be the One That Helps the Company Change Categories
The investor network suggests one final insight.
AI-biotech companies do not stay the same kind of company for very long.
A startup may begin looking like an AI research lab.
Then it becomes a wet-lab platform.
Then it becomes a pharmaceutical partner.
Then it may become a clinical-stage biotech.
Eventually it might become a public company with several therapeutic programs.
The best investor relationship may therefore be one that helps the company move from one identity to another.
Flagship can help create the company.
An AI investor can sharpen the platform.
A healthcare specialist can help judge drug-development risk.
A strategic partner can validate market demand.
An institutional investor can supply scale capital.
No single investor necessarily needs to do everything.
The syndicate as a whole does.
Final Takeaway: Boston Is Building a Capital System for Computational Biology
The biggest finding from our analysis is not that Boston AI-biotech companies have raised roughly $2.4 billion in the ten-company private-capital sample.
The more important finding is who supplied it.
Boston’s computational biology ecosystem is pulling together groups of investors that historically occupied different worlds.
Flagship Pioneering and Third Rock bring company-building experience.
RA Capital, BVF, Perceptive, Casdin and other healthcare specialists understand the brutal realities of developing medicines.
Bessemer, Andreessen Horowitz, Lux, Khosla, Radical, Zetta and other technology-oriented investors understand AI platforms and computational scale.
Fidelity, T. Rowe Price, BlackRock, Pictet, Baupost, ADIA and other large pools of capital provide the financial capacity required when ambitious platforms grow.
NVIDIA, Amgen, Eli Lilly and other strategic companies have their own reasons for wanting exposure to the market.
Together, they are funding something larger than a collection of drug startups.
They are funding a new scientific production model.
Our sample also shows how uneven that market remains. Generate, Lila and Relay alone account for roughly 73.6% of tracked private capital. The top five companies account for almost 90%.
So this is not easy money.
Investors are concentrating capital around the companies they believe can build repeatable scientific engines rather than one-off AI demonstrations.
That may be the most useful signal for founders.
The winning Boston AI-biotech company will probably not be the company with the cleverest model, the biggest funding announcement or the most fashionable language.
It will be the company that can repeatedly convert computation into experiments, experiments into proprietary data, data into better decisions and better decisions into products that matter.
The investors funding Boston’s boom appear to be betting on exactly that loop.
And because Boston already has deep biology, world-class research, experienced biotech capital, wet-lab infrastructure and a growing pool of AI talent, few places are better positioned to find out whether the bet works.
