The 80% Revolution: How AI is Rewriting Drug Discovery
AI is flipping pharma's 90% failure rate on its head. After years tracking this space, I'm sharing what convinced me the 80% revolution is real—and why it matters to you.
Take your coffee—I've got something interesting to share with you today…
Where 90% Failure Becomes "Normal"?
I've spent years watching drugs fail spectacularly, but Pfizer's torcetrapib still haunts me. Nearly $800 million and over a decade of work [1]. The drug was supposed to prevent heart disease but didn't work...
Here's what keeps me up at night: we say "it's normal". We accept that 90% of drugs entering clinical trials will fail,[2] that only 10-14% make it from Phase I to FDA approval.[3,4] That's just how drug development works.
Well, I'm here to tell you that's changing now - dramatically!
The Numbers That Made Me Believe
When I first heard about 80-90% AI success rates in Phase I trials, I was really skeptical. I've seen too many hyped technologies fizzle. But then I dug into the 2024 Drug Discovery Today study analyzing AI-native biotech pipelines,[5] and the data speeks to me.
Let me share what really convinced me: Insilico Medicine's IPF drug, ISM001-055.
Using their Pharma.AI platform, they went from identifying a completely novel target (TNIK kinase) to Phase I trials in under 30 months.[6,7] They synthesized fewer than 80 compounds, not the hundreds of thousands you'd see in traditional screening. The drug sailed through Phase I with zero serious adverse events.[8]
But here's where it gets really exciting: their Phase IIa results from November 2024 showed the drug isn't just slowing IPF progression, it might actually stop or reverse it.[9] For a disease that's essentially a death sentence, that's transformative.
This isn't luck. This is what happens when AI can predict, with remarkable accuracy, which molecules will actually work in humans before you spend years and millions $ testing them.
How This Actually Works? Trust Me, It's Cooler Than It Sounds
I love explaining this because it's genuinely fascinating. Think of AI as having instantaneously reviewed every drug trial, every molecular interaction, every failure and success in pharmaceutical history, then using that knowledge to predict what will work.
Here's what I find most compelling:
Digital biology: We're creating virtual human tissues and disease pathways. You can test thousands of drug candidates in silico before touching a test tube. The predictive accuracy? It's legitimately impressive.
Smart screening: Instead of blindly testing compounds, AI narrows millions of possibilities down to the handful most likely to succeed. It's like having a molecular GPS that routes you around all the dead ends.
Pattern recognition: AI spots the subtle signals that separate drugs that work from those that fail. After tracking this space for years, I'm convinced this is where the real magic happens.
What This Means for All of Us?
For pharma leaders: If you're not integrating AI into your discovery pipeline, you're watching your competitors lap you. We're talking 60-70% timeline compression. Traditional discovery timelines of 10-15 years[10] are becoming 3-5 years. That competitive gap? It's existential.
For clinical and medical affairs teams: I know change is hard. You're suddenly working with data scientists, interpreting AI predictions, designing adaptive trials with AI-selected biomarkers. The results—higher success rates, faster patient recruitment, better outcomes—make it worth the learning curve.
For patients: This is personal for me. Rare diseases have been ignored because the economics didn't work. AI changes that equation completely. Lower costs and higher success rates mean conditions that affected too few people to matter financially now become viable targets. Kids diagnosed with rare genetic diseases in 2025 might see treatments by 2030 instead of never…
What I'm Watching in 2025-2030?
I'm closely tracking several companies that I believe are defining this space. The November 2024 Recursion-Exscientia merger created a massive AI drug discovery powerhouse with over 10 programs in development.[12] Insilico Medicine continues to impress me with their execution. BenevolentAI is doing fascinating work.
Meanwhile, every major pharma company—Boehringer-Ingelheim, Pfizer, Roche, Sanofi, Novartis—is either building AI capabilities or writing billion-dollar partnership checks.[13] Sanofi's $1.2 billion collaboration with Insilico? That's not experimental; that's strategic.
My prediction: by 2030, at least 40-50% of new clinical trials will involve AI-discovered molecules. The companies moving now will own this space. The ones waiting? They'll be acquisition targets.
Here's What You Need to Ask Yourself
I've been in enough boardrooms and conference calls to know the question on everyone's mind: How is your organization preparing for AI-first drug discovery?
Are you building internal capabilities? Partnering with AI platforms? Upskilling your teams? Or are you still in "wait and see" mode?
Because I'll tell you what I'm seeing: the 80% revolution isn't waiting. The gap between AI-enabled companies and traditional players is widening every quarter. The companies acting decisively right now - today - are the ones who'll define pharmaceutical innovation for the next decade.
Your coffee might be getting cold, but this opportunity? It's just heating up.
I'd love to hear your perspective on this—how is your organization approaching AI in drug discovery? Connect with me and let's talk about what you're seeing in your corner of the industry.
References
Jones PH. Torcetrapib development terminated. NCBI. 2006. https://pmc.ncbi.nlm.nih.gov/articles/PMC1702474/
Sun D, et al. Why 90% of clinical drug development fails and how to improve it? Acta Pharmaceutica Sinica B. 2022. https://pmc.ncbi.nlm.nih.gov/articles/PMC9293739/
Wong CH, et al. Estimation of clinical trial success rates. Biostatistics. 2019. https://pmc.ncbi.nlm.nih.gov/articles/PMC6409418/
Lowe D. Drug Failure and Approval Rates. Science | AAAS. 2019. https://www.science.org/content/blog-post/latest-drug-failure-and-approval-rates
Jayatunga MKP, et al. How successful are AI-discovered drugs in clinical trials? Drug Discovery Today. 2024. https://pubmed.ncbi.nlm.nih.gov/38692505/
Bio-IT World. Insilico Medicine Demonstrates Validity of AI Drug Discovery. 2024. https://www.bio-itworld.com/pressreleases/2024/03/15/new-paper-from-insilico-medicine-demonstrates-validity-of-ai-drug-discovery
Insilico Medicine. First AI drug enters Phase II trials. EurekAlert! 2023. https://www.eurekalert.org/news-releases/993844
Insilico Medicine. From Start to Phase 1 in 30 Months. 2022. https://insilico.com/phase1
Insilico Medicine. Positive Results of ISM001-055 for IPF. PR Newswire. 2024. https://www.prnewswire.com/news-releases/insilico-medicine-announces-positive-topline-results-of-ism001-055-for-the-treatment-of-idiopathic-pulmonary-fibrosis-ipf-developed-using-generative-ai-302302583.html
Hughes JP, et al. Principles of early drug discovery. British Journal of Pharmacology. 2011.
Wouters OJ, et al. Research Investment to Bring New Medicine to Market. JAMA. 2020. Referenced in: Cancer Discovery. 2015. https://aacrjournals.org/cancerdiscovery/article/5/2/OF2/4765/
Recursion-Exscientia merger. Recursion Pharmaceuticals. 2024. https://ir.recursion.com/news-releases/news-release-details/recursion-and-exscientia-two-leaders-ai-drug-discovery-space
Insilico Medicine-Sanofi milestone. EurekAlert! 2024. https://www.eurekalert.org/news-releases/1063134
Take your coffee—I've got something interesting to share with you today…
Where 90% Failure Becomes "Normal"?
I've spent years watching drugs fail spectacularly, but Pfizer's torcetrapib still haunts me. Nearly $800 million and over a decade of work [1]. The drug was supposed to prevent heart disease but didn't work...
Here's what keeps me up at night: we say "it's normal". We accept that 90% of drugs entering clinical trials will fail,[2] that only 10-14% make it from Phase I to FDA approval.[3,4] That's just how drug development works.
Well, I'm here to tell you that's changing now - dramatically!
The Numbers That Made Me Believe
When I first heard about 80-90% AI success rates in Phase I trials, I was really skeptical. I've seen too many hyped technologies fizzle. But then I dug into the 2024 Drug Discovery Today study analyzing AI-native biotech pipelines,[5] and the data speeks to me.
Let me share what really convinced me: Insilico Medicine's IPF drug, ISM001-055.
Using their Pharma.AI platform, they went from identifying a completely novel target (TNIK kinase) to Phase I trials in under 30 months.[6,7] They synthesized fewer than 80 compounds, not the hundreds of thousands you'd see in traditional screening. The drug sailed through Phase I with zero serious adverse events.[8]
But here's where it gets really exciting: their Phase IIa results from November 2024 showed the drug isn't just slowing IPF progression, it might actually stop or reverse it.[9] For a disease that's essentially a death sentence, that's transformative.
This isn't luck. This is what happens when AI can predict, with remarkable accuracy, which molecules will actually work in humans before you spend years and millions $ testing them.
How This Actually Works? Trust Me, It's Cooler Than It Sounds
I love explaining this because it's genuinely fascinating. Think of AI as having instantaneously reviewed every drug trial, every molecular interaction, every failure and success in pharmaceutical history, then using that knowledge to predict what will work.
Here's what I find most compelling:
Digital biology: We're creating virtual human tissues and disease pathways. You can test thousands of drug candidates in silico before touching a test tube. The predictive accuracy? It's legitimately impressive.
Smart screening: Instead of blindly testing compounds, AI narrows millions of possibilities down to the handful most likely to succeed. It's like having a molecular GPS that routes you around all the dead ends.
Pattern recognition: AI spots the subtle signals that separate drugs that work from those that fail. After tracking this space for years, I'm convinced this is where the real magic happens.
What This Means for All of Us?
For pharma leaders: If you're not integrating AI into your discovery pipeline, you're watching your competitors lap you. We're talking 60-70% timeline compression. Traditional discovery timelines of 10-15 years[10] are becoming 3-5 years. That competitive gap? It's existential.
For clinical and medical affairs teams: I know change is hard. You're suddenly working with data scientists, interpreting AI predictions, designing adaptive trials with AI-selected biomarkers. The results—higher success rates, faster patient recruitment, better outcomes—make it worth the learning curve.
For patients: This is personal for me. Rare diseases have been ignored because the economics didn't work. AI changes that equation completely. Lower costs and higher success rates mean conditions that affected too few people to matter financially now become viable targets. Kids diagnosed with rare genetic diseases in 2025 might see treatments by 2030 instead of never…
What I'm Watching in 2025-2030?
I'm closely tracking several companies that I believe are defining this space. The November 2024 Recursion-Exscientia merger created a massive AI drug discovery powerhouse with over 10 programs in development.[12] Insilico Medicine continues to impress me with their execution. BenevolentAI is doing fascinating work.
Meanwhile, every major pharma company—Boehringer-Ingelheim, Pfizer, Roche, Sanofi, Novartis—is either building AI capabilities or writing billion-dollar partnership checks.[13] Sanofi's $1.2 billion collaboration with Insilico? That's not experimental; that's strategic.
My prediction: by 2030, at least 40-50% of new clinical trials will involve AI-discovered molecules. The companies moving now will own this space. The ones waiting? They'll be acquisition targets.
Here's What You Need to Ask Yourself
I've been in enough boardrooms and conference calls to know the question on everyone's mind: How is your organization preparing for AI-first drug discovery?
Are you building internal capabilities? Partnering with AI platforms? Upskilling your teams? Or are you still in "wait and see" mode?
Because I'll tell you what I'm seeing: the 80% revolution isn't waiting. The gap between AI-enabled companies and traditional players is widening every quarter. The companies acting decisively right now - today - are the ones who'll define pharmaceutical innovation for the next decade.
Your coffee might be getting cold, but this opportunity? It's just heating up.
I'd love to hear your perspective on this—how is your organization approaching AI in drug discovery? Connect with me and let's talk about what you're seeing in your corner of the industry.
References
Jones PH. Torcetrapib development terminated. NCBI. 2006. https://pmc.ncbi.nlm.nih.gov/articles/PMC1702474/
Sun D, et al. Why 90% of clinical drug development fails and how to improve it? Acta Pharmaceutica Sinica B. 2022. https://pmc.ncbi.nlm.nih.gov/articles/PMC9293739/
Wong CH, et al. Estimation of clinical trial success rates. Biostatistics. 2019. https://pmc.ncbi.nlm.nih.gov/articles/PMC6409418/
Lowe D. Drug Failure and Approval Rates. Science | AAAS. 2019. https://www.science.org/content/blog-post/latest-drug-failure-and-approval-rates
Jayatunga MKP, et al. How successful are AI-discovered drugs in clinical trials? Drug Discovery Today. 2024. https://pubmed.ncbi.nlm.nih.gov/38692505/
Bio-IT World. Insilico Medicine Demonstrates Validity of AI Drug Discovery. 2024. https://www.bio-itworld.com/pressreleases/2024/03/15/new-paper-from-insilico-medicine-demonstrates-validity-of-ai-drug-discovery
Insilico Medicine. First AI drug enters Phase II trials. EurekAlert! 2023. https://www.eurekalert.org/news-releases/993844
Insilico Medicine. From Start to Phase 1 in 30 Months. 2022. https://insilico.com/phase1
Insilico Medicine. Positive Results of ISM001-055 for IPF. PR Newswire. 2024. https://www.prnewswire.com/news-releases/insilico-medicine-announces-positive-topline-results-of-ism001-055-for-the-treatment-of-idiopathic-pulmonary-fibrosis-ipf-developed-using-generative-ai-302302583.html
Hughes JP, et al. Principles of early drug discovery. British Journal of Pharmacology. 2011.
Wouters OJ, et al. Research Investment to Bring New Medicine to Market. JAMA. 2020. Referenced in: Cancer Discovery. 2015. https://aacrjournals.org/cancerdiscovery/article/5/2/OF2/4765/
Recursion-Exscientia merger. Recursion Pharmaceuticals. 2024. https://ir.recursion.com/news-releases/news-release-details/recursion-and-exscientia-two-leaders-ai-drug-discovery-space
Insilico Medicine-Sanofi milestone. EurekAlert! 2024. https://www.eurekalert.org/news-releases/1063134
Take your coffee—I've got something interesting to share with you today…
Where 90% Failure Becomes "Normal"?
I've spent years watching drugs fail spectacularly, but Pfizer's torcetrapib still haunts me. Nearly $800 million and over a decade of work [1]. The drug was supposed to prevent heart disease but didn't work...
Here's what keeps me up at night: we say "it's normal". We accept that 90% of drugs entering clinical trials will fail,[2] that only 10-14% make it from Phase I to FDA approval.[3,4] That's just how drug development works.
Well, I'm here to tell you that's changing now - dramatically!
The Numbers That Made Me Believe
When I first heard about 80-90% AI success rates in Phase I trials, I was really skeptical. I've seen too many hyped technologies fizzle. But then I dug into the 2024 Drug Discovery Today study analyzing AI-native biotech pipelines,[5] and the data speeks to me.
Let me share what really convinced me: Insilico Medicine's IPF drug, ISM001-055.
Using their Pharma.AI platform, they went from identifying a completely novel target (TNIK kinase) to Phase I trials in under 30 months.[6,7] They synthesized fewer than 80 compounds, not the hundreds of thousands you'd see in traditional screening. The drug sailed through Phase I with zero serious adverse events.[8]
But here's where it gets really exciting: their Phase IIa results from November 2024 showed the drug isn't just slowing IPF progression, it might actually stop or reverse it.[9] For a disease that's essentially a death sentence, that's transformative.
This isn't luck. This is what happens when AI can predict, with remarkable accuracy, which molecules will actually work in humans before you spend years and millions $ testing them.
How This Actually Works? Trust Me, It's Cooler Than It Sounds
I love explaining this because it's genuinely fascinating. Think of AI as having instantaneously reviewed every drug trial, every molecular interaction, every failure and success in pharmaceutical history, then using that knowledge to predict what will work.
Here's what I find most compelling:
Digital biology: We're creating virtual human tissues and disease pathways. You can test thousands of drug candidates in silico before touching a test tube. The predictive accuracy? It's legitimately impressive.
Smart screening: Instead of blindly testing compounds, AI narrows millions of possibilities down to the handful most likely to succeed. It's like having a molecular GPS that routes you around all the dead ends.
Pattern recognition: AI spots the subtle signals that separate drugs that work from those that fail. After tracking this space for years, I'm convinced this is where the real magic happens.
What This Means for All of Us?
For pharma leaders: If you're not integrating AI into your discovery pipeline, you're watching your competitors lap you. We're talking 60-70% timeline compression. Traditional discovery timelines of 10-15 years[10] are becoming 3-5 years. That competitive gap? It's existential.
For clinical and medical affairs teams: I know change is hard. You're suddenly working with data scientists, interpreting AI predictions, designing adaptive trials with AI-selected biomarkers. The results—higher success rates, faster patient recruitment, better outcomes—make it worth the learning curve.
For patients: This is personal for me. Rare diseases have been ignored because the economics didn't work. AI changes that equation completely. Lower costs and higher success rates mean conditions that affected too few people to matter financially now become viable targets. Kids diagnosed with rare genetic diseases in 2025 might see treatments by 2030 instead of never…
What I'm Watching in 2025-2030?
I'm closely tracking several companies that I believe are defining this space. The November 2024 Recursion-Exscientia merger created a massive AI drug discovery powerhouse with over 10 programs in development.[12] Insilico Medicine continues to impress me with their execution. BenevolentAI is doing fascinating work.
Meanwhile, every major pharma company—Boehringer-Ingelheim, Pfizer, Roche, Sanofi, Novartis—is either building AI capabilities or writing billion-dollar partnership checks.[13] Sanofi's $1.2 billion collaboration with Insilico? That's not experimental; that's strategic.
My prediction: by 2030, at least 40-50% of new clinical trials will involve AI-discovered molecules. The companies moving now will own this space. The ones waiting? They'll be acquisition targets.
Here's What You Need to Ask Yourself
I've been in enough boardrooms and conference calls to know the question on everyone's mind: How is your organization preparing for AI-first drug discovery?
Are you building internal capabilities? Partnering with AI platforms? Upskilling your teams? Or are you still in "wait and see" mode?
Because I'll tell you what I'm seeing: the 80% revolution isn't waiting. The gap between AI-enabled companies and traditional players is widening every quarter. The companies acting decisively right now - today - are the ones who'll define pharmaceutical innovation for the next decade.
Your coffee might be getting cold, but this opportunity? It's just heating up.
I'd love to hear your perspective on this—how is your organization approaching AI in drug discovery? Connect with me and let's talk about what you're seeing in your corner of the industry.
References
Jones PH. Torcetrapib development terminated. NCBI. 2006. https://pmc.ncbi.nlm.nih.gov/articles/PMC1702474/
Sun D, et al. Why 90% of clinical drug development fails and how to improve it? Acta Pharmaceutica Sinica B. 2022. https://pmc.ncbi.nlm.nih.gov/articles/PMC9293739/
Wong CH, et al. Estimation of clinical trial success rates. Biostatistics. 2019. https://pmc.ncbi.nlm.nih.gov/articles/PMC6409418/
Lowe D. Drug Failure and Approval Rates. Science | AAAS. 2019. https://www.science.org/content/blog-post/latest-drug-failure-and-approval-rates
Jayatunga MKP, et al. How successful are AI-discovered drugs in clinical trials? Drug Discovery Today. 2024. https://pubmed.ncbi.nlm.nih.gov/38692505/
Bio-IT World. Insilico Medicine Demonstrates Validity of AI Drug Discovery. 2024. https://www.bio-itworld.com/pressreleases/2024/03/15/new-paper-from-insilico-medicine-demonstrates-validity-of-ai-drug-discovery
Insilico Medicine. First AI drug enters Phase II trials. EurekAlert! 2023. https://www.eurekalert.org/news-releases/993844
Insilico Medicine. From Start to Phase 1 in 30 Months. 2022. https://insilico.com/phase1
Insilico Medicine. Positive Results of ISM001-055 for IPF. PR Newswire. 2024. https://www.prnewswire.com/news-releases/insilico-medicine-announces-positive-topline-results-of-ism001-055-for-the-treatment-of-idiopathic-pulmonary-fibrosis-ipf-developed-using-generative-ai-302302583.html
Hughes JP, et al. Principles of early drug discovery. British Journal of Pharmacology. 2011.
Wouters OJ, et al. Research Investment to Bring New Medicine to Market. JAMA. 2020. Referenced in: Cancer Discovery. 2015. https://aacrjournals.org/cancerdiscovery/article/5/2/OF2/4765/
Recursion-Exscientia merger. Recursion Pharmaceuticals. 2024. https://ir.recursion.com/news-releases/news-release-details/recursion-and-exscientia-two-leaders-ai-drug-discovery-space
Insilico Medicine-Sanofi milestone. EurekAlert! 2024. https://www.eurekalert.org/news-releases/1063134
Let's Decode the Future of Medicine with Technology
- Together
Let's Decode the Future of Medicine with Technology
- Together
No spam, unsubscribe anytime.
Let's Decode the Future of Medicine with Technology
- Together
No spam, unsubscribe anytime.