Executive Summary

The convergence of artificial intelligence (AI) and biologics drug discovery has accelerated dramatically by 2026, signaling a major inflection point in pharmaceutical R&D. Rapid advances in AI – from protein-structure prediction (AlphaFold) to large language models for proteins – have enabled companies to design novel biologic therapeutics (antibodies, enzymes, cell therapies) at unprecedented speed. Venture funding and corporate investment have surged: for example, Earendil Labs, a new AI-driven biologics startup, raised $787 million in March 2026 to develop its AI platform for antibody and biologic design (^[1]). Major pharmaceutical firms have launched or expanded AI initiatives: Eli Lilly is teaming with NVIDIA to build a dedicated AI “supercomputer” (an “AI factory”) for drug discovery and manufacturing (^[2]) (^[3]), and Lilly also invested $250 million into an AI-focused research partnership with Purdue University (^[4]). Meanwhile, global investment trends show pharma investing heavily in external biotech innovation: a recent analysis found 11 big pharma companies (including Lilly, AstraZeneca, GSK) committed over $150 billion to license Chinese biotech assets in the last five years (^[5]), and Chinese-originated drug candidates now account for ~40% of licensing deals (^[6]). At the same time, venture capital is flowing strongly into AI: in the first half of 2025, 53% of global VC funding went to AI companies (^[7]), though funding for healthcare AI specifically has cooled from its 2021 peak of $22 billion to $10.5 billion in 2024 (^[8]) as investors demand clinical validation and solid business models.

This report provides an exhaustive analysis of AI in biologics discovery circa 2026. We cover historical context (e.g. Nobel-winning AlphaFold technology), discuss the current technological landscape (advanced protein modeling, generative design, lab automation), and detail leading initiatives and case studies. In particular, we examine the implications of the recent Earendil Labs funding round as emblematic of a new wave of AI-biotech ventures, and the Eli Lilly / NVIDIA collaboration as an example of Big Pharma scaling up AI infrastructure. We analyze pharma investment patterns, including massive R&D budgets, partnerships, and deal-making. We incorporate data on funding flows, quotes from industry leaders, and expert perspectives on challenges and future directions. The conclusion assesses the potential for AI to reshape biologics R&D and its broader implications for medicine.

Introduction and Background

Biologics – therapeutic agents derived from living organisms, typically large molecules such as proteins, antibodies, peptides or nucleic acids – constitute a rapidly growing segment of the pharmaceutical market. Advances in molecular biology have enabled blockbuster monoclonal antibodies (e.g. for cancer, autoimmune diseases) and fusion proteins, but traditional biologics discovery is slow and costly. By 2024, biologics accounted for a large share of novel drug approvals and sales; for example, over 100 biologics were expected to lose patent exclusivity by 2030 (^[9]), underscoring the need for new biologic innovation. However, the “biosimilar void” (only ~10% of expiring biologics had a biosimilar competitor in development (^[9])) highlights the high development costs and technical hurdles in this field. Manufacturing biologics often requires $100–300 million for development alone (^[10]) due to complex expression systems and cold-chain logistics. Thus, accelerating biologics discovery has long been a high priority.

AI offers the prospect of speeding up this process by leveraging large-scale data and computation. Shape prediction platforms like AlphaFold can compute 3D protein structures in silico; generative models can design novel protein sequences; and AI-driven lab automation can optimize experiments iteratively. These tools promise to augment human expertise, reduce trial-and-error, and explore vast "design spaces" of protein therapeutics. Indeed, in 2024 the Nobel Prize in Chemistry was awarded largely for AI-enabled protein modeling: Demis Hassabis and John Jumper (DeepMind) and David Baker (UW) were honored for AlphaFold and Rosetta, recognizing “groundbreaking techniques to decode and design new proteins using AI” (^[11]). Nobel Committee reviews note that solving protein structures from amino acid sequences—a problem once intractable—is now “easy for Mother Nature” thanks to these AI breakthroughs (^[12]).

Yet despite these technological leaps, biologics R&D remains challenging. AI models can predict protein folds with high accuracy up to ~90–100% (AlphaFold 2) and the latest AlphaFold 3 can model interactions between proteins, DNA, RNA and small molecules (^[13]). This marks a “first big step” toward simulating cellular molecular networks (^[14]). Nobel laureate Paul Nurse remarked that AlphaFold’s updates “enable increased accuracy in predicting structures of complexes between different macromolecules” (^[15]). In short, AI has begun to alleviate key bottlenecks in understanding biology. Generative models (for example, language-model-inspired systems like EvolutionaryScale’s ESM3 (^[16]) (^[17])) can propose entirely new protein sequences or functions. By pairing these designs with automated pipelines that rapidly synthesize and test candidates, companies can compress years of R&D into months. However, experts caution that many areas of biology remain poorly understood by data-driven models (^[18]), so the promise of "AI-designed therapeutics" is still in its formative years.

In the remainder of this report, we explore where the field stood in 2026: the key players, funding trends, real-world outcomes, and expert analyses of AI-driven biologics discovery. We emphasize examples like Earendil Labs (a record-setting funding round in biologics AI) and Lilly’s NVIDIA lab (an industry-scale partnership), and we contextualize these within broader pharma strategy and investment flows.

AI Technologies for Biologics Discovery

Protein Structure Prediction

AlphaFold and Rosetta have been transformative for proteins, the building blocks of most biologics. Google DeepMind’s AlphaFold system dramatically improved protein-fold prediction, solving the CASP14 challenge in 2020. The subsequent release of AlphaFold 3 (May 2024) extended this to multi-molecule interactions (^[19]) (^[13]).Time magazine reports that AlphaFold 3 “can predict the structure of and interactions between biological molecules including proteins, DNA, RNA, and small molecules that could function as drugs” (^[19]). By modelling how different biomolecules bind and fold together, this tool opens the door to designing, for example, multi-domain antibodies or protein complexes. As Demis Hassabis noted, AlphaFold 3 represents a “big step toward” understanding how biological functions arise from molecular interactions (^[20]). Notably, DeepMind made AlphaFold 3 available non-commercially on their public server (^[19]), ensuring that many researchers can leverage the tool for modeling new biologics.

The 2024 Nobel Prize also celebrated David Baker’s Rosetta platform, which used advanced computational modeling (later augmented with AI) to design new proteins from scratch. Indeed, as AP News reported: “Three scientists who discovered powerful techniques to decode and even design novel proteins… were awarded the Nobel Prize in chemistry”（Oct 2024） (^[11]). Thus, AI-based tools have reached the threshold of Nobel-level impact, and are now being applied to industrial drug pipelines.

Generative Protein Design

Beyond folding prediction, generative AI is emerging as a tool to create entirely new protein sequences with desired functions. Startups like Generate Biomedicines (NASDAQ:GBIO) are pioneering “generative biology”: using machine learning to propose novel proteins. Axios notes that in early 2026, Generate held a $425 million IPO focused on AI-driven protein design (^[21]). Similarly, EvolutionaryScale, cofounded by former Meta AI researchers, has developed a large protein language model (ESM3) that “can be prompted to design new proteins” (^[22]). In fact, Axios reports that EvolutionaryScale used its model to generate a novel green fluorescent protein, finding creative solutions “for which we can find no matching structure in nature” (^[23]). EvolutionaryScale has partnered with AWS and NVIDIA to make this model available to select customers (^[24]). These efforts illustrate how AI can engineer biology from first principles (^[25]).

AlphaFold3 and generative models together enable end-to-end biologic design: a company can specify a therapeutic target (e.g. a receptor), use AI to design a binding protein or antibody, and then validate the design. For example, Axios describes how one AI model, when prompted with functions, structures, or sequences, “finds creative solutions to complex combinations of prompts” (^[23]), essentially designing new proteins. Academic teams have already begun feeding AI with known protein data and mining its latent knowledge to suggest candidates. As Time magazine observes, these tools could “revolutionize drug discovery” in the coming years (^[26]).

Other Technological Enablers

AI in biologics is also supported by advances in automation and data. High-throughput lab robotics and improved “wet lab” instrumentation generate vast datasets (e.g. proteomics, single-cell assays). As Insitro CEO Daphne Koller explains, modern biosystems can be measured with “unprecedented fidelity” (^[27]), producing detailed images and omics data. However, only AI can process such voluminous data to find patterns: Koller notes that a human researcher’s “eyes will just glaze over” if given massive high-dimensional datasets (^[27]). By automating data analysis and even experiment planning (“scientific AI agents” (^[2])), companies aim to close a feedback loop of design-build-test that continuously improves biologic candidates.

High-performance computing (HPC) is another prerequisite. The envisioned Lilly–NVIDIA collaboration involves building a supercomputer capable of developing research models, advancing manufacturing techniques, and generating "scientific AI agents" (^[2]). Such a system would allow massive AI simulations (e.g., running millions of in silico experiments) in days rather than years. Beyond Lilly, NVIDIA with partners is also delivering large supercomputers (e.g. 2.2 exaFLOP OAK), and researchers worldwide are deploying GPU clusters for AI-driven bioinformatics. The growth of “laboratory informatics” (AI+lab) underscores that hardware and software advances together are propelling biologics discovery.

Industry Developments and Case Studies

The Rise of AI Biotech Ventures

The post-2020 funding climate has seen a new cohort of biotech startups explicitly leveraging AI. Axios reports noted significant rounds for AI-driven drug companies. For instance, Xaira Therapeutics (founded by former Genentech CSO Marc Tessier-Lavigne) launched in April 2024 with over $1 billion in funding (^[28]). Xaira’s platform combines machine learning, data generation, and product development. Similarly, Candid Therapeutics (a biotech developing bispecific T-cell engager antibodies for autoimmune disease) raised $370 million in late 2024 (^[29]). These examples, though not strictly all biologics, demonstrate investor appetite for AI-led drug R&D.

A particularly notable case is Earendil Labs, a San Francisco startup in AI-driven antibody discovery. On March 23, 2026, Earendil announced a $787 million private funding round (^[1]). The round was backed by industry heavyweights – notably Sanofi and a “Biotech Development Fund” (a Pfizer–Hillhouse initiative), as well as Dimension Capital, DST Global, and Luminous Ventures (^[1]). According to the company, this investment will accelerate its use of machine learning throughout the biologics R&D process (^[30]). Earendil reports that its AI engine has already generated over 40 programs, including a novel anti-TL1A antibody now moving toward Phase 2 trials (^[30]). Indeed, the company plans multiple IND submissions for AI-designed assets in 2026–27. This blowout funding round – on the order of a major pharma acquisition – underscores how AI-enabled biologics is perceived as a high-growth area.

The Earendil example also highlights pharma–startup partnerships. Sanofi not only invested in the round, but had earlier struck two major deals with Earendil. In April 2025, Sanofi paid $125M upfront (with up to $1.72B in milestones) for exclusive global rights to two Earendil-designed bispecific antibodies targeting autoimmune and IBD indications (^[31]). The agreement also included tiered royalties on future sales. Then in January 2026, Sanofi arranged a second deal: up to $160M upfront (and $2.56B in milestones) for additional autoimmune programs from Earendil’s platform (^[32]). These licensing deals – representing nearly $4 billion in total possible payments – signal both the confidence leading pharma placed in Earendil’s AI platform and the premium they are willing to pay for novel biologic candidates. As Dimension Capital’s Zavain Dar put it, “Earendil Labs stands out for its ability to translate AI innovation into real, scalable R&D execution… [showing] that AI can consistently generate high-quality biologics programs and advance them toward the clinic” (^[33]).

Other biotech real-world outcomes illustrate AI’s impact. Formation Bio, a startup using AI to streamline clinical trials, reported selling two drug candidates: one to Sanofi (€545M deal) and one (minority stake) to Eli Lilly (≈$2B) (^[34]). Though Formation Bio’s focus is administrative (accelerating trials rather than discovery), these deals – worth on the order of billions – reveal how new AI-led models can produce assets attractive to pharma. Insitro (founded 2018) has likewise partnered with Lilly and Bristol-Myers on metabolic and neurological disease programs (^[35]). These company examples show the breadth of AI’s reach across the drug pipeline.

For clarity, Table 1 summarizes a selection of recent major investments and partnerships in AI-driven biologics discovery (2024–2026). This non-exhaustive list highlights the scale and diversity of activity:

Table 1: Selected Investments and Deals in AI-driven Biologics (2024–2026). Sources indicate press reports and company announcements for each row.

Case Study: Eli Lilly – NVIDIA AI Lab

In late 2025, pharmaceutical giant Eli Lilly announced a major AI initiative with NVIDIA. In a speech at the World Economic Forum (January 2026 Davos), NVIDIA CEO Jensen Huang spotlighted Lilly as an example of pharma’s AI transformation (^[2]). Huang noted Lilly’s plan (announced in 2025) to build an AI supercomputer, codenamed their “AI factory,” that would generate “scientific AI agents” to plan experiments (^[2]). This lab will integrate NVIDIA’s latest GPU technology (such as the GH200 series) and possibly use software orchestration from NVIDIA’s acquisitions (e.g. Slurm) to simulate biologic research at scale. In the words of a company executive, Lilly’s forthcoming AI systems will “accelerate how fast new drugs reach patients” by shifting from bench labs to AI-driven platforms (^[37]).

Independent media and Spanish outlets elaborate on Lilly’s vision. Press reports in November 2025 described Lilly’s system as “the most powerful AI supercomputer operated by a pharmaceutical company” (^[3]). This supercomputer will underpin Lilly’s so-called “AI factory”, a specialized infrastructure for drug discovery. The factory aims to shorten R&D cycles by enabling rapid in silico iteration: for example, designing drug candidates, simulating their molecular interactions, and planning subsequent wet-lab experiments without human limbo. According to Lilly, the supercomputer will also be applied to manufacturing: it’ll “advance manufacturing techniques” for biologics, likely through modeling cell-culture processes and optimizing yields (^[2]). In essence, Lilly and NVIDIA are pursuing an end-to-end AI platform – from generative discovery to “augmented biology” manufacturing.

This Lilly–NVIDIA collaboration exemplifies how Big Pharma is channeling resources into AI. In the same year Lilly also invested $250 million in an eight-year AI research partnership with Purdue University (^[4]). This initiative embeds Lilly researchers at Purdue to co-develop AI methods for drug discovery, development, and biotech manufacturing. Purdue’s president hailed the Lilly deal as possibly “the largest private-industry-to-university agreement in history” (^[4]), underlining its scale amid federal R&D cutbacks (^[38]). Collectively, Lilly’s moves – multi-hundred-million investments in AI infrastructure and partnerships – demonstrate a strategic pivot. Lilly is not alone; other pharma companies are following suit. For example, Novartis announced a $23 billion commitment to bolster US health innovation (including advanced manufacturing and data science) over five years (^[39]), and Bayer highlighted $7 billion in US R&D investments (2019–2024) focusing on areas like advanced therapies, often enabled by AI (^[40]).

Other Pharma AI Collaborations

Beyond Lilly, multiple pharma–tech alliances are underway. At Supercomputing 2025 (Nov 2025), NVIDIA and partners unveiled several new AI supercomputer projects for science (^[41]), and at NVIDIA’s own GTC 2026 conference, consortiums were announced (the "Nemotron Coalition") to build open AI models (^[42]). While these are broad science initiatives, they signal that the ecosystem for AI-enabled R&D is expanding. In therapeutics specifically, industry leaders note AI’s promise: Bayer’s Sebastian Guth said, “What’s most exciting is…using AI at scale to develop medicines that would have otherwise likely not seen the light of day” (^[43]). AstraZeneca’s Head of Data Science, Jim Weatherall, similarly stated that AI and data science are “transforming R&D, helping us turn science into medicine more quickly…from target identification to clinical trials” (^[44]).

Some large-scale programs illustrate this merging of biology and AI: for instance, a recent Nature commentary found that each of 11 top pharma companies (including AstraZeneca, Bristol Myers Squibb, Eli Lilly, GSK) committed over $150 billion to license drugs from China’s biotech sector in the past five years (^[5]), many of which were discovered with advanced computational tools. And as regulatory pressures mount, we see pharma also directing AI toward safety and process: regulatory reports on AI suggest agencies are watching these developments closely.

Pharma Investment Trends and Funding Landscape

Venture Capital and M&A

The funding environment for both biotech and AI has seen dramatic shifts. After an AI-driven funding boom, investors in healthcare tech are becoming more selective. A PitchBook analysis (via Axios) noted that global VC funding for health-related AI peaked at $22 billion in 2021 but fell to $10.5 billion in 2024 (^[8]) as investors demand evidence of efficacy. This headwinds in biotech funding coincide with AI’s ascent: by mid-2025, AI startups overall captured 53% of all venture dollars globally (^[7]). The result is fewer new biotech startups and down rounds for some, but for AI-enabled companies that can show clinical progress, the capital is still available. Meanwhile, large corporate M&A remains active: for example, Sanofi’s $11.5 billion acquisition of Blueprint Medicines (June 2025) and Novartis’s announced $12 billion buyout of Avidity Biosciences (Oct 2025) signal that big pharmas are willing to spend on strategic assets, including biologics pipelines (^[45]) (^[46]).

Anecdotally, Pancreatic biotech stands reveal caution: Axios reports that new biotech IPOs and SPACs have been more muted in 2025 than AI IPOs, reflecting investor skittishness in biotech public markets. The drop in biotech stock indices (amid overall market gyrations) also shows that investors are wary without concrete product outputs. Some analysts even warn of an “AI bubble” where biotechnology has been neglected (^[47]). However, venture groups specializing in life sciences (Flagship, ARCH, etc.) continue to raise dedicated funds for next-gen therapeutics, and corporate VCs (e.g. Pfizer Ventures, GV) are selectively investing in promising AI-biotech startups.

Big Pharma R&D Budgets

On the corporate side, traditional R&D numbers remain massive. Bayer reported $7 billion spent on US R&D from 2019-2024, largely in advanced areas like cell/gene therapy (^[40]). GSK announced a $30 billion plan (2024–2030) to expand US R&D and manufacturing, explicitly including $1.2 billion for AI-enabled manufacturing upgrades (^[48]). AstraZeneca, with more modest public commentary, is known to invest multibillions per year globally on R&D, with growing emphasis on data science to boost pipeline productivity. These enormous budgets indicate that pharma can both pursue internal innovation and acquire/licence external technology.

However, industry leaders lament a funding squeeze in public science. At a May 2025 industry forum, Flagship Pioneering CEO Noubar Afeyan warned “it has never been as bad as this” for biomedical innovation, pointing to proposed NIH budget cuts and FDA resource constraints (^[49]). Another panelist noted that absent federal research support, pharma and private capital will need to pick up the slack (^[50]). This context partly explains Lilly’s Purdue deal: as federal grants shrank, Lilly’s $250M to Purdue (the largest private–university partnership on record) aimed to “accelerate drug discovery using AI and other technology-driven approaches” (^[4]). In short, while pharma investment dollars are large, external pressures are reshaping how risk is shared between government, academia, and industry.

Global and Strategic Factors

Globally, a major trend is the rise of China’s biotech sector. Axios reports that Chinese-originated drugs – often discovered with AI and genomics – are projected to comprise ~40% of all new licensing deals in the year (^[6]). In response, western pharma has poured capital into China: a recent analysis found they have licensed over $150 billion of Chinese biotech assets in 5 years (^[5]). This geopolitical dynamic reinforces competitive pressure to innovate with AI at home: U.S. leaders have even floated major cuts to NIH and FDA, making corporate R&D and licensing the primary engine for new drug discovery. In contrast, Asian governments likewise support biotech and AI initiatives, adding to R&D momentum.

Another strategic factor is manufacturing. The global biosimilars market (low-cost copies of biologics) has saved >$56 billion since 2015 (^[51]), but analysts note that very few complex biologics have biosimilars, partly due to the difficulty in development (^[9]). AI-driven process optimizations are now being explored to lower manufacturing costs or improve yields – for instance, using ML to refine cell culture conditions or fermentation. The Lilly–Nvidia supercomputer explicitly includes “manufacturing” in its remit (^[2]), reflecting a recognition that end-to-end AI can also apply to production pipelines (e.g. glyco-engineering of antibodies, continuous bioreactors). In summary, pharma investment trends in 2026 show a dual commitment: spending heavily on core R&D pipelines and aggressively integrating digital technologies into every phase, under the belief that overall return on cancer and other biologic R&D can be raised.

Case Studies and Real-World Examples

Earendil Labs: A Deep Dive

Earendil Labs, a stealth-mode AI-biotech startup, has become a bellwether for the field. According to a March 2026 report in Biospace/Patient Daily, Earendil “raises $787 million for AI-driven biologics development” (^[1]). This headline-grabbing round ranks among the largest-ever single financings in biotech. Earendil’s stated mission is to leverage machine learning throughout the biologics R&D process – from target discovery to candidate optimization (^[30]). Specifically, the company focuses on antibodies, including bispecifics, and other advanced modalities.

The Biospace piece details that Earendil’s AI engine has already generated “more than 40 programs” within a few years of founding (^[30]). One lead candidate is an anti-TL1A antibody, which is expected to enter Phase 2 trials soon. (TL1A is a target implicated in inflammatory bowel diseases.) The company plans IND filings for this and other AI-designed assets in 2026–27 (^[30]). These are tangible milestones: within roughly five years of inception, Earendil is moving at least one computationally-originated drug into human studies. In interviews, investors highlight that Earendil demonstrates “AI … translating into real, scalable R&D execution” (^[33]).

The structure of the investment round is also revealing. Earendil was backed by a mix of strategic and financial investors. Sanofi participated, which is no coincidence given its prior collaborations: in April 2025, Sanofi paid $125M up front (with ~1.72B in milestones) for exclusive rights to two of Earendil’s bispecific antibodies (autoimmune/IBD targets) (^[31]). Then in January 2026, Sanofi agreed to fund up to $160M up front (plus $2.56B in milestones) for access to additional autoimmune programs from Earendil’s platform (^[52]). In other words, major deals with Sanofi pre-dated the $787M round, suggesting strategic validation. When Sanofi partially exited at IPO or acquired companies, it comes full circle: now it is a lead investor in the founder.

Other investors include the “Biotech Development Fund” created by Pfizer and Hillhouse Capital, as well as VC firms Dimension, DST Global, and Luminous. This syndicate indicates both pharma and tech capital are keen to back AI platforms with big potential. For example, Dimension Capital (a biopharma VC) said “the team has shown that AI can consistently generate high-quality biologics programs and advance them toward the clinic” (^[33]), underscoring confidence that Earendil’s algorithms deliver credible candidates.

What makes Earendil’s case instructive is the transparency of outcomes. Unlike many AI projects whose results are proprietary or unpublished, Earendil is publicly naming targets, pipeline progress, and deal terms. This provides a benchmark for efficacy: if its 40+ AI programs truly converge on drug-like molecules with in vivo activity, that would validate the technology. Moreover, the licensing deals and funding numbers – nearly $4B of contingent payments from Sanofi plus hundreds of millions in VC – show that the market is willing to pay top dollar for AI-discovered biologics. In practical terms, Earendil exemplifies a new model: a pure-play AI biotech that conceives and develops antibodies entirely in silico, then passes them to pharmaceutical partners for clinical development.

Lilly–NVIDIA AI Lab

Lilly’s partnership with NVIDIA represents industrial-scale commitment to AI. Public statements by Jensen Huang and Spanish press detail the endeavor. As noted, Lilly will build the most powerful industry supercomputer yet for R&D (^[3]). This machine is intended to power a continuous AI-driven discovery cycle. In Jensen Huang’s words (from the WEF talk): the system will “develop research models, advance manufacturing techniques and generate ‘scientific AI agents’ that plan experiments” (^[2]). In practice, this could mean autonomous lab planning: the AI might design an antibody sequence, virtually screen it against targets, suggest mutations, simulate binding (using molecular dynamics or AlphaFold-like pipelines), and even schedule synthesis and screening tasks in high-throughput labs.

Importantly, the scope includes manufacturing. Lilly explicitly wants to “reimagine the factory” using AI. If successful, the supercomputer might optimize cell-line engineering, fermentation conditions, purification protocols and yield, as part of a “factory of the future.” While details are limited, industry sources describe an on-site AI lab adjacent to Lilly’s laboratories, feeding into both discovery and production teams. This “factory” vision harks to a future where R&D and manufacturing are tightly integrated by data flows.

From an industry perspective, the Lilly–NVIDIA lab can serve as a precedent for others. JP Morgan and other analysts noted the announcement alongside reports of Nvidia’s stock; the overall message was that pharma’s largest players are pivoting to AI-led R&D. Observers compared the Lilly-NVIDIA pact to tech partnerships in other sectors (e.g. airlines with IBM’s Watson). The scale of investment (workers, hardware, software) has not been fully disclosed, but it is surely on the order of hundreds of millions or more. This collaboration, slated to come online in 2026, will be one of the first big tests: can a pharma supercomputer genuinely accelerate biologics discovery?

Other Initiatives and Examples

Beyond Earendil and Lilly, numerous other projects illustrate the trend:

• Insitro (San Francisco) exemplifies applying ML to early discovery. As CEO Daphne Koller explains, Insitro integrates “quantitative biology” data (from cellular assays, single-cell sequencing, etc.) with machine learning models to tackle complex diseases (^[27]). Insitro has already partnered with Lilly and Bristol-Myers Squibb to use its platform for metabolic and neurological disorders (^[35]). This shows how established drug companies are outsourcing AI work: pharma supplies targets and resources, the AI company applies its algorithms.

• Generate Biomedicines. Having raised a record-breaking IPO, Generate is a leader in generative biology. The firm uses deep learning to propose novel proteins. One recent example: generating new scaffolds for growth factors to potentially treat diabetes or ophthalmological conditions (as noted in BioRxiv publications, but beyond our press scope). Generate’s public disclosure of its IPO (and a Nasdaq ticker) signals that AI-biotech is maturing to mainstream investment.

• Shiru. This startup catalogs and predicts functions of natural proteins (millions identified from environmental DNA). While Shiru has mostly targeted biotech partnerships (e.g. for food ingredients), its method – using AI to assign function to proteins – is adjacent to drug discovery. TIME magazine noted Shiru’s on-demand platform for ordering protein samples after AI-based functional prediction (^[53]). The implication is that if one can mine nature’s protein structures via AI, similar approaches could find novel therapeutic proteins. Shiru’s technology underlines that not all AI in biologics is about synthetic design; repurposing natural proteins is another avenue.

• EvolutionaryScale. The firm’s ESM3 protein language model is now available to select industry partners (^[24]). This model represents a “first step toward” designing biology from first principles (^[25]) and is being used by companies to ideate new enzyme or antibody sequences. Partnerships with AWS and NVIDIA mean some corporate customers will have early access, anticipating that the model’s outputs can feed into their R&D pipelines.

• Pharma M&A. Though not AI-native, big acquisitions reflect where pharma is investing. For example, Sanofi’s acquisition of Blueprint Medicines (rare-disease small molecules) for $9.5B (^[45]) and Novartis’s purchase of Avidity ($11B for RNA?) demonstrate that pharma is hungry for new modalities. If any part of the acquired pipelines used AI in discovery, those deals can indirectly boost AI’s credibility. When Novartis commits $23B to US health investment (^[39]), it includes expansion of manufacturing and digital, indirectly supporting AI infrastructure.

• Academic Initiatives. Universities are also partnering, as with Lilly-Purdue (^[4]). Such collaborations typically involve shared labs or talent exchange, training a new generation skilled in AI for pharma. Without formal press coverage, numerous NIH-funded projects are also exploring AI for antibody design or cell therapy optimization, though details are often behind scientific publications.

These examples demonstrate multiple models of AI integration: startups developing novel candidates, big pharma building AI centers, partnerships bridging academia and industry, and traditional drug development boosting AI in trials or production.

Data Analysis and Trends

Quantitative data on funding and outcomes highlights the rising tide in AI-biologics. The VC trends in healthcare AI show initial hype followed by rationalization: according to PitchBook (via Axios), 2021 investments hit $22B in AI-health, dropping to $10.5B in 2024 (^[8]). The lesson is that “buckets of capital” flowed into AI biotech, but investors are now ‘becoming more selective’ (^[54]), focusing on validated approaches. Nonetheless, AI companies remain prominent: in H1 2025, over half of VC dollars went to AI startups (^[7]), reinforcing that any sector not using AI risks being starved of capital.

On the corporate side, we compiled reported R&D figures to illustrate the scale of investment (see Table 2). Bayer alone has invested $7B in U.S. pharma R&D (2019-2024) (^[40]), and GSK plans $30B by 2030 (^[48]). In 2025, GSK emphasized that $1.2B of its new plan is earmarked for introducing AI in manufacturing processes (^[48]). These capital commitments dwarf typical startup rounds and reflect that developing a biologic is still often a multibillion-dollar endeavor. Notably, many of these budgets are increasingly tied to AI as an efficiency lever.

Table 2: Selected Pharma Company AI R&D Investments and Collaborations (2024–2026).

In terms of outputs, the data is just beginning to emerge. One early sign is the progression of AI-designed candidates: for instance, Shankar Vishwanath of Vertex Pharmaceuticals noted in 2025 that the first-ever AI-derived drugs are entering trials (Vertex itself worked with Recursion for an exa-sale small molecule, though not a biologic) (^[41]). The tangible evidence in biologics will likely be first clinical-stage filings, which firms like Earendil are targeting. Analysts are watching metrics such as “AI-derived candidates in clinical pipeline” as a new trend indicator, analogous to biotech pipeline counts. Another measurable trend is patent filings: an uptick in patents for “machine learning-designed antibodies” or similar could be quantified from patent databases. Finally, an interesting meta-trend: computational biology IPOs and drug approvals. Experts point out that historically, large pharma M&A has depended on proven assets; if AI-driven companies can produce assets that lead to acquisitions or validation deals (as seen with Sanofi-Earendil), that will spur further investment.

Discussion: Implications and Future Directions

The developments up to 2026 suggest several key implications for science and industry:

1. Transformation of R&D Paradigm. The shift toward “dry lab” AI platforms represents a fundamental change. Pharmaceutical R&D has traditionally been “wet lab”-driven – lab experiments generating data incrementally. Now, the front end of discovery can be largely computational. This could drastically shorten lead identification timelines. For example, generating tens of thousands of antibody variants in silico and filtering them computationally could replace whole months of lab work. If Earendil’s reported pace holds, we may soon see drugs entering trials within 3–5 years of concept rather than the usual 7–10.

2. Economic and Business Impact. The massive funding rounds indicate that biotech valuations will become increasingly tied to software and data capabilities. A startup’s worth now factor in not just bench results but the sophistication of its AI models. Cap tables have shifted: companies like Earendil count both VCs and tech-savvy investors on-board, unlike traditional biotechs which were funded mostly by pharma or biotech VCs. In markets, one consequence is re-evaluating biotech valuations: companies that are “AI-ready” get premiums, whereas those reliant on trial-and-error drug screening may lag.

3. Pharma Strategy. Big Pharma is retooling internally. Hiring patterns are changing: countless job postings now seek computational biologists, software engineers, and AI specialists within R&D organizations. We expect more “science is software” initiatives, akin to how Google sphere embraced the shift. The partnerships with universities (like Purdue) and tech firms (like NVIDIA) are both talent pipelines and knowledge transfer vehicles. Over time, this could blur traditional boundaries: we might see hybrid companies equally expert in biology and data science.

4. Regulatory and Ethical Considerations. The rise of AI in drug design brings novel regulatory questions. If a molecule is generated by an AI algorithm, how does one validate its safety profile? Regulators are not yet fully prepared for software-originated candidates. However, long regulatory lead times (8–12 years for a drug) mean agencies have time to adapt. The press has noted global AI safety concerns recently (e.g. Turing-winning Yoshua Bengio’s discussions on biosecurity (^[55])), suggesting that oversight will develop in parallel. Biologics themselves have heavy regulation (e.g. FDA’s Center for Biologics Evaluation and Research), and the appointment of figures like Vinay Prasad as head of that center (^[56]) may influence how AI-derived biologics are assessed.

5. Scientific Challenges. Despite optimism, many hurdles remain. AI models rely on data – but our experimental understanding of complex biology is incomplete. Cancer, for example, involves multi-gene interactions and tumor microenvironments; an AI-designed antibody against a single target may not suffice. Janssens et al. (FT) note that AI drug discovery has “been slower than hoped,” partly because “crucial areas of biology…aren’t sufficiently understood…to give AI the data they need” (^[18]). Moreover, AI outputs must still be validated in molecules and cells. Off-target effects, immunogenicity of designed proteins, and manufacturability are real concerns. As Nikoloz (Bayer) put it, humans still supply “three billion years of evolution” wisdom that AI alone cannot replace (^[43]). In response, hybrid approaches – combining AI design with directed evolution techniques and high-throughput screening – are likely to be common.

6. Future Outlook. By 2026, some predicted timelines may already be in motion. Many analysts now envision “clinically useful AI-designed drugs” emerging by the late 2020s. Earendil’s advancing candidates, along with the rumored pipelines of companies like Generate Biomedicines, suggest Phase 2 trials for AI-nominated biologics could occur around 2027–2028. If any such drug reaches even Phase 3 (or gets regulatory submission) by 2030, it will vindicate years of hype. On the technical front, continued synergy with other fields (CRISPR genome editing, quantum computing) could further accelerate the pace.

In closing, AI in biologics discovery is at a tipping point. The 2026 landscape shows robust investment, high-profile projects, and early indicator successes. All stakeholders – pharmaceutical companies, biotech startups, investors, regulators, and patients – must prepare for this transformation. If harnessed well, AI could indeed lead to “breakthrough medicines” more efficiently (^[57]). At the same time, vigilance is needed to ensure rigorous science underpins the AI-generated promise.

Conclusion

The integration of AI into biologics drug discovery is reshaping the pharmaceutical industry. A wave of investment – exemplified by Earendil’s $787 million funding and major corporate AI labs – reflects confidence that machine intelligence can accelerate the creation of novel therapeutics. Historical achievements like AlphaFold (Nobel Prize 2024 (^[11])) have unlocked new computational capabilities, while startups and pharma giants alike are now translating those capabilities into pipelines. Funding trends show both the appetite and caution of investors: biotech is no longer the only game in town, as a majority of VC dollars flow into general AI technologies (^[7]), but pharma’s multibillion-dollar R&D budgets are shifting to include AI-driven methods (^[44]) (^[4]).

Multiple case studies – from Earendil’s AI-discovered antibodies (^[30]) to Lilly’s NVIDIA-powered supercomputer (^[2]) – illustrate that the marriage of AI and biologics is moving from theory to practice. Evidence-based signs of success are emerging: discovery of dozens of candidate programs, licensing deals worth billions, and preclinical-to-clinic transitions. Yet, experts emphasize that underlying biology still poses fundamental questions. DeepMind’s Hassabis, Nobel laureates, and biotech CEOs all acknowledge that AI is not a magic wand but an accelerator (^[15]) (^[11]) (^[18]). The near future will test whether these AI-sourced leads can survive clinical trials and improve patient outcomes.

Looking ahead, the implications are profound. AI-driven biologics discovery promises faster cures for diseases that stymied traditional methods. It also forces an industry-wide pivot: skill sets, processes, and business models will evolve. Pharma companies will become as much data companies as biology companies, biotech valuations will depend on algorithms and data, and regulatory science must adapt to ensure safety. If realized, the convergence of AI and biologics could indeed catalyze a “quantum leap” in medicine – echoing pundits’ hopes that AI will transform drug R&D once and for all (^[58]). Our analysis of the current state (2026) suggests that we are at the threshold of that leap, with high stakes and high rewards ahead.

References

• Axios, “Nvidia CEO Jensen Huang: Drug research will be transformed” (Jan. 21, 2026) (^[2]).
• Axios, “Lilly investing $250 million in research at Purdue” (May 9, 2025) (^[4]).
• Axios (sponsored), “Bayer’s next act: From AI to breakthrough medicines” (Dec. 17, 2025) (^[40]) (^[43]).
• Axios, “How will AI impact the pharma industry?” (Moneyweek, Sept. 25, 2025) (^[44]) (^[59]).
• Axios, “Lilly y NVIDIA crearán el superordenador…” (Redacción Médica, Nov. 5, 2025) (^[3]).
• AP News, “Better drugs through AI? Insitro CEO…” (Dec. 2, 2024) (^[35]) (^[27]).
• Axios, “Ex-Stanford president’s AI biotech startup snares $1 billion” (Apr. 24, 2024) (^[28]).
• Axios, “Ex-Meta experts at AI-biotech startup offer tool to create new molecules” (Jun. 25, 2024) (^[22]) (^[17]).
• Axios, “Investors get pickier with AI-fueled health deals” (Jan. 9, 2025) (^[8]).
• Axios, “AI is eating venture capital, or at least its dollars” (Jul. 3, 2025) (^[7]).
• Axios, “Biopharma must pivot on funding sources amid federal research cuts” (May 9, 2025) (^[49]).
• Axios Vitals newsletter, “China’s biotech march” (Sept. 8, 2025) (^[6]) (^[5]).
• Axios Pro Rata, “Pharma …” (March–Nov 2024 summaries) – financial/deals context (^[34]).
• Time, “Google DeepMind’s Latest AI Model Is Poised to Revolutionize Drug Discovery” (May 8, 2024) (^[19]) (^[13]).
• Time, “Daphne Koller” (Sept. 5, 2024) (^[60]).
• Le Monde (trad. in English), “2024 Nobel Prize for Chemistry: AI garners more recognition” (Oct. 9, 2024) (^[12]).
• AP News, “Nobel Prize in chemistry honors 3 scientists who used AI to design proteins” (Oct. 9, 2024) (^[11]).
• Axios (sponsored), “Biosimilars, the healthcare hero in danger” (Jan. 5, 2026) (^[9]) (^[10]).
• Axios, “Candid Therapeutics raises $370M for bispecifics” (Sept. 10, 2024) (^[29]).
• Axios, “Startup uses AI to develop vaccines…” (Feb. 26, 2024) (^[36]).