Author: Muhammad Yusuf , Researcher at Delphi Digital
Compiled by: Yuliya, PANews
Editor's Note: While BIO has experienced a surge in price over the past month, don't be misled by short-term wealth creation myths and token frenzies. While AI and open-source models have lowered the barriers to new drug development to rock bottom, true medical breakthroughs still cannot bypass the long and rigorous clinical trial cycle. Faced with the impatient speculation in the crypto world and the "funding winter" in traditional scientific research, DeSci has indeed encountered a breakthrough opportunity. However, what it needs is not a speculative game of switching tracks every few months, but rather a financing infrastructure strictly tied to clinical milestones and governed by professionals to ensure its sustainable development.
An AI agent called peptAI designed a novel ADHD (Attention Deficit Hyperactivity Disorder) peptide drug candidate from scratch in just 24 hours, passed it through eight validation processes, and produced a molecule ready for wet lab testing. The lab work only cost a few thousand dollars. BIO Protocol, the platform supporting the project, saw its token surge by 105%. Within hours, half the people on Crypto Twitter had added "DeSci" to their bios, just as they had added "AI" six months earlier.
Today, open-source protein folding models rival AlphaFold3 at zero licensing cost, public bioactivity databases cover 2.5 million compounds, and wet lab validation costs have fallen below $2,000. AI is drastically reducing the cost and time of drug development. For the past week, I've been trying to figure out exactly what's truly different this time.
Surviving Phase I clinical trials doesn't prove anything.
Current data suggests that AI-discovered drugs have an 80-90% success rate in Phase I clinical trials, compared to the traditional baseline of around 47%. However, what no one seems to clarify is that Phase I clinical trials test whether a drug will cause death, not whether it can cure the disease. Passing this stage only means your compound is safe enough to continue research, but it still needs to undergo multiple rounds of screening until it finally receives FDA approval.
Currently, fewer than 40 AI-discovered compounds have reported Phase II clinical data, and none have completed Phase III trials. Insilico Medicine's Rentosertib is the most advanced AI-discovered compound to date, reporting positive Phase IIa results for idiopathic pulmonary fibrosis in mid-2025 (published in *Nature Medicine*), and starting Phase III clinical enrollment in China in the fourth quarter of 2025. If all goes smoothly (enrollment completed in 2027, data retrieval in 2028, and FDA review in 2029), you'll need to wait at least three years for the best candidate in this pipeline of 173 drugs. Several compounds in this pipeline were shelved in 2025 for failing to meet endpoints in atopic dermatitis, schizophrenia, and cancer. Independent analysts believe there is a 60% chance the first AI-designed drug will receive FDA approval in 2027, but to date, no AI-designed drug has achieved this.
Is encrypted Twitter really suitable for true DeSci?
Now, keep these timelines in mind and look at the BIO Protocol chart again. The token fell from $0.89 to $0.018, then rebounded 105% on the back of the PeptAI news, reaching a trading volume of $720 million and a market capitalization of approximately $68 million. The entire DeSci fundraising logic was based on the assumption that token holders would patiently wait for clinical projects that could take seven to ten years. However, the reality was that even before the Phase 1 clinical data was unblinded, Crypto's Twitter account had already shifted to the next narrative.
Pump Science even leaked its private key on GitHub, giving rise to a bunch of fraudulent tokens, one of which was even called Cocaine. Moreover, the enforceability of IP-NFTs has never been tested in court.
The battle between open source and DeSci
If we don't fall into conditioned reflexes of self-deception and long-term speculative hype, open-source science might offer a glimmer of hope for DeSci.
In October 2025, the OpenFold Consortium released OpenFold3 under the Apache 2.0 license. It is fully trainable and commercially usable, built upon structures determined by over 300,000 experiments (unlike AlphaFold3, which was restricted by Google to academic use only). Boltz-2 from MIT and Recursion jointly predicts protein structures and binding affinities 1000 times faster than physics-based methods.
Baker Labs released RFdiffusion3 in December. ChEMBL, containing 2.5 million bioactive compounds with a complete ADMET distribution, is freely available to anyone with a laptop. Infrastructure that pharmaceutical companies used to spend millions of dollars building in-house is now hosted on GitHub under a permissive license, and five pharmaceutical companies are currently conducting federated training on their proprietary drug-protein libraries through the Federated OpenFold3 Initiative.
Nobody's talking about this on Crypto Twitter because there are no tokens to trade, and I highly doubt the core contributors to these codebases would be excited about issuing tokens.
Funding shortage in the scientific community
In 2025, over 7,800 NIH and NSF grants were terminated or suspended, freezing over $5 billion in funding. The NIH's budget remained unchanged because Congress continued to allocate funds, but the government still froze the funding pipeline. New competitive grant awards plummeted from 11,659 in fiscal year 2024 to 6,095 in fiscal year 2025, a 48% drop. The success rate for researchers applying for funding fell from 21% to 13%, resulting in a loss of $508 million for Fred Hutch and $945 million for Harvard University.
This funding gap is precisely why DeSci's pitch, if handled properly, could secure opportunities. In July 2025, Gero, funded by VitaDAO, signed a research and licensing agreement with Chugai Pharmaceutical (a Roche subsidiary with a market capitalization of approximately $100 billion), with milestone payments reaching $250 million. This marked the first time a DAO-funded project had yielded something valued at nine figures by a genuine pharmaceutical company. The fact that this process was completed smoothly without governance battles or absconding remains one of the most significant developments in the field.
Four long years
This year, 15 to 20 AI-discovered drugs have entered Phase III clinical trials, while data access for Rentosertib won't be available until 2028 at the earliest. This means that a definitive conclusion on whether all of this can be translated into drugs that work in humans is still several years away. Regardless of the existence of tokens, the open-source technology stack will continue to reduce costs, and the funding vacuum will continue to push researchers to anyone willing to write a check. Today, open-source protein folding models are comparable to AlphaFold3 at zero licensing costs, wet lab validation costs have dropped below $2,000, and the NIH has just announced its lowest grant success rate in twenty years.
Even if AI delivers on all its proponents' promises and halves the drug development timeline, the cycle from discovery to approval will still be four to five years, and that's in the optimistic scenario assuming improved success rates in Phase III clinical trials. In an industry where portfolio convictions rotate with quarterly earnings calls, four years is a long wait, and token holders even find six months of holding as unbearable as a life sentence.
The cost of drug discovery and innovation is declining every quarter, regardless of token popularity. The NIH's funding shortage is also a cause for concern. Perhaps there exists a viable model somewhere in between, where tokens fund specific trials and milestones, while governance is handled by professionals. Gero's collaboration with Chugai is the first proof that DAO-funded projects can produce nine-figure sums that pharmaceutical companies are willing to pay. Beyond debunking the hype, I also wonder if anyone will build a defensive funding infrastructure for truly decentralized science.
