Mustard

AstraZeneca's CEO, Pascal Soriot, said during their earnings call: "Artificial intelligence is changing the way we work and is already used throughout AstraZeneca, from discovery and development to commercial operations and healthcare delivery. This is not something that will happen in the future. Artificial intelligence is making a difference right now across all parts of our business." That is one of the important statements a big pharmaceutical company CEO has made publicly in 2026. Not because it is a surprise. But because it confirms the obvious. AstraZeneca is hoping their AI Development Agent will eventually cut the time needed for chemistry, manufacturing, and controls development in half. Merck signed a deal up to one billion dollars with Google Cloud for agentic AI across research and development, manufacturing, and commercial operations. GSK, BMS, and Novo Nordisk are all heading in that direction. The biggest pharmaceutical companies in the world are now openly committing to AI as part of their operations, not as a test project. That's a big step up. But here is the point I want to make: AI-driven drug discovery is the most crowded area, with hundreds of biotech companies working in this space. In manufacturing, the companies offering services are major industrial automation companies like Siemens, GE, and Honeywell, launching pharma AI platforms under Industry 4.0 brands, while software companies are adding machine learning modules for process control. The pharmaceutical companies’ response to AI in manufacturing is to add features to existing systems. Siemens adds an AI layer, Honeywell adds a maintenance module, and Werum adds machine learning to their MES. After all of it is added on, none of it still causes a major change. A 2025 MIT study found that 95% of enterprise AI pilots did not deliver real business outcomes, mostly because the systems stayed separate from real work processes, databases,s and company ownership. The failure rate exists because it's not possible to get the power of AI by adding intelligence to a broken system. Because the broken system stops the intelligence from working. At Mustard Biopharma, we did not add AI to a pharmaceutical factory. Instead, we built the factory for AI starting from the software layer. One operating system, one data foundation, and one autonomous control layer that does not need to connect six systems before making a decision. AstraZeneca's CEO is correct. AI is making a difference today. But the question remains, whether we are getting 10% of what's possible through add-on modules, or building for the full potential from the start? At Mustard Biopharma, we chose the full potential. #AIinPharma #AstraZeneca #Biopharma #LifeSciences

In 2023, something almost unbelievable happened. Cisplatin, an old, widely used cancer drug, went into shortage in the US. Cancer centers had to ration supplies. 93% of surveyed cancer centers reported carboplatin shortages. But guess what: this wasn't a drug discovery problem. Rather, it was a manufacturing problem. This is exactly the kind of infrastructure problem we are building Mustard Bio to solve.

TCS just launched an agentic AI platform for pharmaceutical drug development. TCS ADD AgentHub is a role-based, enterprise-ready AI platform that enables agentic AI in drug development at scale, transforming clinical trials and pharmacovigilance while maintaining regulatory and audit requirements. The results are compelling. Solutions powered by the platform have demonstrated up to 40% efficiency gains in clinical data management, up to 30% reduction in clinical study build effort, and up to 30% cost savings in end-to-end safety case processing. TCS, a $30 billion IT services company, just told the pharmaceutical industry that agentic AI is no longer experimental. It's enterprise-ready. It's audit-ready. And it works. I agree with all of that. But I want to name something this launch makes clear: TCS's platform embeds AI agents into clinical trials and pharmacovigilance workflows with humans retaining responsibility for governance and decision-making. That is agentic AI for drug development. For the R&D side of the value chain. The manufacturing side, where raw materials become finished medicine batches, where machines run bioreactors and filtration and formulation lines, is still waiting for its agentic AI moment. TCS's platform tells a pharma company's clinical team what's happening in their drug development process. Our Factory OS tells the machines what to do in the manufacturing process, autonomously, in real time, without waiting for human sign-off on every parameter adjustment. Two different layers. Both are necessary, and neither one substitutes for the other. What today's TCS announcement confirms is something we have believed since we started building Mustard Biopharma: agentic AI in a highly regulated, audit-required pharmaceutical environment is not just possible; it's the direction the entire industry is moving. We are building the manufacturing layer of that future. While the R&D side has gotten its platform, the factory floor ought to be the next. #AgenticAI #PharmaManufacturing #TCS #Deeptech

The FDA just lost its Commissioner. FDA Commissioner Marty Makary is being fired amid controversy over vaccine and drug policy decisions, leaving the agency without confirmed leadership. The FDA has already replaced acting heads of its drug and biologics centers, deepening leadership instability that raises serious concerns for pharmaceutical developers. This is significant. And not just for regulatory reasons. Every time the FDA leadership changes, the pharmaceutical industry faces a period of regulatory uncertainty, approval slowdowns, guidance documents stall, and companies building products that require FDA interaction have to recalibrate their timelines. But here is what that means for pharma manufacturing specifically: The companies most exposed to this instability are the ones whose manufacturing timelines are already stretched to 12–18 months by legacy CDMO infrastructure. Add regulatory uncertainty on top of an already slow manufacturing process, and a drug that should reach patients in 18 months could take 24 or 30. The companies least exposed are those whose manufacturing is faster and more automated, and that produce audit-ready compliance documentation as a natural output of the production process, rather than as a manual administrative task that requires regulatory handholding at every step. But we designed our "Factory OS" for exactly this kind of environment. Automated batch records, built-in cGMP compliance, and zero dependence on regulatory stability to run an efficient factory. Manufacturing infrastructure that is resilient to policy turbulence isn't a luxury, but it's a necessity. #FDA #Biopharma #PharmaPolicy #cGMP

Samsung Biologics just reported that its revenue for the quarter of 2026 is up by 30 percent from the same time last year. They are using all of their plants at the Bio Campus I in Songdo, which is the biggest biomanufacturing facility in the world, with a capacity of 605,000 liters. They also announced that they are buying the PolyPeptide Group for $1.8 billion. This will help them get into making peptide-based APIs. Add more manufacturing sites in places like Sweden, Belgium, France, the US, and India. This is an amazing story about how Samsung Biologics built a huge manufacturing business from nothing in less than 15 years. But there is something the numbers do not tell you: Samsung Biologics had a second quarter despite problems at its Songdo campus due to a union strike. A union strike occurred at a biomanufacturing facility worldwide. That one fact tells you a lot about how vulnerable the system's when it relies on a lot of people to make pharmaceuticals. The bigger the facility, the greater the risk that something will go wrong because of the people working there. This is not meant to criticize Samsung Biologics. It is an observation about how things are set up. When you need thousands of workers to do tasks by hand and work shifts, your ability to produce things is only as good as your relationship with the workers. If there is a strike, a pandemic, or some other problem with the workforce, even the biggest biomanufacturing facility in the world will slow down. The people who made Mustards Factory OS thought about this from the beginning. They designed it to be autonomous, controlled by software, and to need little help from people on the production floor. This is not because people are not important. Because the people waiting for medicine cannot afford to have the manufacturing process slowed down by labor disputes. The future of making pharmaceuticals is not about building factories. Instead, it is about building smarter ones. #PharmaManufacturing #CDMO #SamsungBiologics #AIinPharma #Deeptech

Samsung Biologics just reported that its revenue for the quarter of 2026 is up by 30 percent from the same time last year. They are using all of their plants at the Bio Campus I in Songdo, which is the biggest biomanufacturing facility in the world, with a capacity of 605,000 liters. They also announced that they are buying the PolyPeptide Group for $1.8 billion. This will help them get into making peptide-based APIs. Add more manufacturing sites in places like Sweden, Belgium, France, the US, and India. This is an amazing story about how Samsung Biologics built a huge manufacturing business from nothing in less than 15 years. But there is something the numbers do not tell you: Samsung Biologics had a second quarter despite problems at its Songdo campus due to a union strike. A union strike occurred at a biomanufacturing facility worldwide. That one fact tells you a lot about how vulnerable the system's when it relies on a lot of people to make pharmaceuticals. The bigger the facility, the greater the risk that something will go wrong because of the people working there. This is not meant to criticize Samsung Biologics. It is an observation about how things are set up. When you need thousands of workers to do tasks by hand and work shifts, your ability to produce things is only as good as your relationship with the workers. If there is a strike, a pandemic, or some other problem with the workforce, even the biggest biomanufacturing facility in the world will slow down. The people who made Mustards Factory OS thought about this from the beginning. They designed it to be autonomous, controlled by software, and to need little help from people on the production floor. This is not because people are not important. Because the people waiting for medicine cannot afford to have the manufacturing process slowed down by labor disputes. The future of making pharmaceuticals is not about building factories. Instead, it is about building smarter ones. #PharmaManufacturing #CDMO #SamsungBiologics #AIinPharma #Deeptech

LambdaVision Space Manufacturing A biotech startup is manufacturing artificial retinas on the International Space Station because gravity itself is the enemy of quality in their production process. On Earth, gravity-driven forces cause uneven protein layers, leading to significant material waste and limiting scalability. In microgravity, those layers form with unprecedented uniformity. Let this sink in: a company (LambdaVision) went to space (literally left the planet) because the manufacturing environment on Earth wasn't good enough for what the biology required. That's not just a bold step but a bullish one. Artificial retinas produced in microgravity show markedly improved uniformity, optical performance, and reproducibility compared to those made on Earth. They also show enhanced stability and biocompatibility while using less raw material. Sincerely, I find this story extraordinary, not just scientifically, but as a statement about manufacturing innovation. Note: LambdaVision didn't accept the constraints of existing manufacturing infrastructure. Instead, they went looking for a better environment for their specific biology, found it 250 miles above Earth, and built an automated production system to operate there. ISS testing allowed the team to refine automation, fault detection, and quality control, key steps toward meeting regulatory requirements. That's the same mindset behind Mustard Biopharma but applied to a different constraint. We are not going to space. But we are refusing to accept that the fragmented, manual, and legacy-software-dependent pharma factory floor is an unchangeable constraint. The biology of anti-cancer medicines deserves a manufacturing environment built specifically for what it requires: autonomous, precisely controlled, software-defined from the ground up. If LambdaVision went to the ISS because Earth's manufacturing environment wasn't good enough. Mustard Biopharma built Factory OS because the pharma factory's software environment isn't good enough. #PharmaManufacturing #Biotech #Deeptech #SpaceManufacturing #BioManufacturing

Just read that ADC landscape report. Surprisingly, there are over 2,000 programs, a lot of deals, and everyone is moving toward better payloads and linkers. Indeed, the science is really interesting, but what people don’t talk about enough is how difficult it still is to create these medicines. ADCs are not biologics. You have the antibody, a powerful payload, and the conjugation step. Still, getting clinical batches is already a challenge. And scaling it without making costs and timelines explode is more difficult. That is the bottleneck we are solving at MustardBio. We are creating modular factories that can make specialized medicines (including complicated ones like this) much faster and much cheaper than traditional big facilities. Imagine manufacturing that actually matches how biotech works now. Small batches, quick results, less money tied up. The medicines are becoming more advanced. Hence, the manufacturing side needs to catch up, or it will become the problem.

In 1990, the Human Genome Project started with one goal: to find every gene in human DNA. It took 13 years and $3 billion. That project started the biotech era. Companies like Amgen, Genentech, and Biogen used that knowledge to create the first generation of targeted biologics. Monoclonal antibodies, recombinant proteins, biosimilars. Medicines that had never been made before and could not be created by regular chemistry. The science was amazing. But making it was a different problem altogether. Producing medicines inside living cells. CHO cells, bioreactors, and complex fermentation processes needed a completely new kind of factory. Clean, fully controlled, and checked against rules. The industry made those factories the way it knew how at that time: by hand, costly machines, people running the tools, paper records, and systems that did not connect. That setup became the CDMO industry we know today. And it has not changed much since. Though the biology has improved through 40 years of science. Gene therapy, cell therapy, CRISPR, AI-designed antibodies. The factory that gets those molecules is still using the disconnected, manual old system that was made in the 1990s to handle the first group of biologics. The science moved fast, but the factory moved slowly. Thirty years later. That difference is what Mustard Biopharma is here to fix. We did not take the factory setup from the 1990s. Instead, we built another from scratch. Our “Factory OS” is what pharmaceutical manufacturing looks like when it's designed with the tools we have today. #PharmaManufacturing #Biopharma #Biotech #LifeSciences #Deeptech

NVIDIA GTC Agentic AI Jensen Huang stood on stage at GTC 2026 and declared the arrival of the agentic AI inflection point. Among the key announcements was NVIDIA's push toward agentic AI extending into healthcare, including Roche's AI factories, protein design reasoning models, and physical AI aimed at accelerating healthcare robotics. Huang described agentic AI as systems that can act autonomously to achieve goals, driving transformation across industries as the dominant paradigm shifts from conversational AI to autonomous action. The life sciences community is paying attention; agentic AI is gaining traction; 47% of healthcare organizations are now using or evaluating it for multi-step workflows, with pharma and biotech highlighting drug discovery as a top AI revenue driver. But here's the observation nobody at GTC made explicitly: Every agentic AI system in pharma is being built for discovery. Drug target identification, protein design, clinical trial optimization, and cell line development. But nobody is building agentic AI for the factory. The pharmaceutical manufacturing floor where raw materials become finished medicine batches is still running on fragmented legacy software stacks that require constant human coordination. And the agentic AI revolution that is reshaping drug discovery hasn't even touched the factory that produces the drug. That's the gap Mustard Biopharma is building into. Our “Factory OS” is the agentic layer for pharmaceutical manufacturing, an AI-native operating system that autonomously controls machines, orchestrates production end-to-end, and makes decisions in real time without waiting for human sign-off. For the last three years, all of the questions have been, “What is the ROI of AI," said one GTC panelist. I think now we are going to see an inflection of real business economics taking off. And in pharma manufacturing, that inflection hasn't happened yet. But Mustard Biopharma is building it. #AgenticAI #AIinPharma #NVIDIA #Biopharma #Deeptech