Biotech in 2026
Excited about organoids and proteoforms
Castle Fund’s three pillar vertical sectors are biotech, energy and aerospace. Within those, we’re bullish on certain sub-sectors that have reached technical inflection points and investment momentum that we believe will drive them closer to consensus bets at Series A soon.
This post will take a look at two incredible new areas within biotech that are quickly moving from science fiction towards commercialization: organoids and proteoforms.
Organoids: Drug development has a prediction problem. Roughly 90 percent of drugs that enter clinical trials fail to reach patients, and a significant share of those failures are the preclinical models used to test them. Rodents and 2D human cell lines don’t accurately predict what happens in humans.
An organoid is a three-dimensional, self-organizing cluster of human cells grown in a laboratory with the architecture and similar function of a real organ. Unlike cell cultures, organoids develop spatial structure by forming layers, channels, and cell-type diversity that closely mirrors the tissue they’re derived from. Researchers have successfully grown organoids that model the gut, liver, kidney, brain, lung, pancreas, and other organs, using either adult stem cells taken from tissue biopsies or induced pluripotent stem cells (iPSCs) reprogrammed from a patient’s own blood or skin cells.
In target identification and validation, patient-derived organoids can be generated from individuals with specific genetic backgrounds or disease states, leading to drugs with better efficacy. In drug screening, organoid libraries from multiple donors can be used to test compounds across a genetically diverse population, aiding in identifying toxicity signals early.
Regulatory and legislative tailwinds have increased in the last five years:
The FDA Modernization Act 2.0 eliminated the statutory mandate for animal testing in December 2022, explicitly authorizing cell-based assays, microphysiological systems, and sophisticated computer models as equally valid evidence.
The FDA Modernization Act 3.0 was introduced in February 2024 to create a formal qualification pathway for non-animal methods.
The Senate passed FDA Modernization Act 3.0 by unanimous consent on December 16, 2025, directing the FDA to update its regulations within twelve months.
In parallel, the regulatory branch has taken up the charge with enthusiasm.
In September 2024, the FDA’s ISTAND Pilot Program admitted the first organ-on-a-chip submission — a liver model for predicting drug-induced liver injury — from Emulate (see below) signaling that complex microfluidic and organoid models can progress through the same evidentiary pipeline as traditional models.
In April 2025, the FDA announced a plan to phase out animal testing requirements for monoclonal antibodies, with organoid toxicity testing explicitly named as one of the New Approach Methodologies that can replace or reduce animal data in IND applications.
In July 2025, the NIH went further by barring funding for animal-only studies, and shortly after announced the Standardized Organoid Modeling Center with $87M in initial contracts.
And there are early but meaningful signals around startup funding and exits:
Emulate, the organ-on-chip company spun out of Harvard’s Wyss Institute, raised an $82 million Series E in April 2023 led by Khosla Ventures, and has commercial partnerships with Pfizer, J&J, and AstraZeneca.
CN Bio secured $21 million in Series B funding in April 2024 to expand its PhysioMimix multi-organ platform, which operates under a Cooperative Research and Development Agreement with the FDA.
Merck acquired HUB Organoids B.V. in December 2024. HUB holds foundational patents on adult stem cell-derived organoid technology and has run drug screening services for major pharma clients.
Precision Cell Systems acquired BennuBio in October 2025 to enter the 3D cell culture market.
Xilis is a Series B company that has raised $116M, backed by Mubadala, Google Ventures and Felicis, to develop its MicroOrganoSphere (MOS) platform. MOS generates patient-derived tumor organoids for precision oncology drug selection. Their cancer trials in 2023 and 2025 showed high concordance between MOS drug sensitivity testing and actual clinical responses.
In my Asymmetry Ventures portfolio, I have a few early stage organoid startups, including Volumetric (acquired in 2021 by 3D Systems for their IP around 3D-printed lungs), Frontier Bio (strongest in blood vessel organoids, also offering lung and brain, and aiming to move beyond clinical testing to organ implants) and Vitroscope, which can test and monitor organoids while physically perturbing them in ways that mimic the human body environment.
And now, zooming in from the organ all the way down to the protein...
Proteoforms: For decades, biotech operated on a simple equation: one gene makes one protein. The Human Genome Project, completed in 2003, was built assuming that a full sequence of the genome would give us all the proteins that run the human body. This was 2D sequence thinking. It turns out to be incomplete where it meets the chaotic 3D world of cells.
When proteins fold up into three dimensions, they can take on multiple shapes (and go through various modifications downstream in the cell) that change their function. For example, alternative splicing allows different segments of the gene’s messenger RNA to be included or excluded during translation into the amino acid protein sequence. And post-translational modifications (PTMs) add a further layer of diversity: after a protein is synthesized, it can be phosphorylated, acetylated, methylated, glycosylated, ubiquitinated, and changed in dozens of other ways at specific sites to effectively turn the protein “ON”, “OFF”, or modify function.
If you’ve read about AlphaFold and other similar AI tools for predicting protein folding, they have not solved proteoforms, although they are a step on the path. They are giving us the most statistically common or stable proteoform version of a canonical protein. It’s now believed that there are from 10X to 100X more proteoforms in the human body than there are genes.
Why this matters for curing disease: the same amino acid sequence can have a disease-promoting proteoform and a benign one. For example, tau protein, correlated with Alzheimer’s disease, is a well-studied example: different phosphorylation states of tau correspond to different stages of neurodegeneration and potentially different therapeutic targets. Similar dynamics appear in cancer, where aberrant splicing produces oncogenic forms of otherwise benign proteins.
Although the startups working on proteoforms previously were platforms to generate data or provide services to CROs and big pharma, we see that starting to shift towards internal therapeutics development. For example:
Protai, seed funded in 2022, has been demonstrating new proteoforms for years in collaboration with NVIDIA and Google’s AlphaFold, and has recently announced they are developing a therapeutic for breast cancer.
Post-Translational Medicines, which recently went through the IndieBio accelerator, has a team primarily from the very well-funded Altos Labs (longevity startup launched in 2022 with a record-breaking $3B in funding, backed by Jeff Bezos and ARCH Venture Partners), and has developed new methods for scaling the discovery of proteoforms, as well as a confidential internal drug pipeline.
The Human Proteoform Project at Northwestern is an academic effort to map proteoforms using “top-down” protein sequencing spectrometry that doesn’t break apart the protein (in combination with the lab leader’s protein analytics startup Immpro). This could become an accelerant for other companies to enter the field.
This follows closely the pattern from earlier proteomics ML startups that are bringing drugs to market, but are focused on known genetic mutations, combined protein/ligand folding prediction, or discovering hidden pockets for small molecules in “undruggable” proteins. The biggest bet is on David Baker’s diffusion models for protein design at Xaira Therapeutics – they emerged from stealth in 2024 with a whopping $1B in funding. Other examples include Charm Therapeutics ($80M Series B in 2025 for a leukemia lead drug candidate, backed by NEA and Khosla), and Think Bioscience ($55M Series A in 2026). Generate Biosciences just had a $400M IPO a few weeks ago that is sitting at a $1.5B market cap. They use generative AI to design novel proteins and antibodies, and have a Phase III trial going for an asthma drug.