Introduction: The Big Picture

$855 billion. That is the toll antimicrobial resistance could take on the world annually according to the World Health Organization. And one of its quiet accelerants is a problem the healthcare industry treats as routine: drug shortages. When the right antibiotic is not on the shelf, clinicians fall back on broad-spectrum alternatives – feeding the very resistance behind that number. The shortages are neither rare nor fleeting. The U.S. Pharmacopeia reports that more than 40 life-saving drugs have been in shortage for over 3 years. Behind those figures is not a market inefficiency. It is a surgery postponed, a prescription unfilled, a therapy that exists but doesn’t arrive – in every region of the world.

The evidence stacks up market by market. In the United States, simply managing drug shortages cost hospitals $894 million according to a Vizient report. A figure that excludes the added cost of higher-priced substitute medications. In Europe, the European Association of Hospital Pharmacists' 2025 survey found that every single hospital pharmacist who responded had faced a critical medicine shortage in 2024. And the burden is not confined to wealthy systems: the World Health Organization found that essential-medicine shortages strike high-, middle-, and low-income countries alike, with the harshest consequences falling on the regions least equipped to absorb them.

The life sciences industry meant to prevent this is, by its own numbers, still bleeding the value it needs to solve it: the cost of poor quality can reach 25 to 40% of sales, an average of 330 drug recalls are initiated every year, the average new drug now costs $2.23 billion to bring to the market, and EU device certifications now stretching 13 to 18 months, all while a US tariff, an EU stockpiling race, and renewed Middle East conflict expose a supply chain with no room left to absorb shock. This is the paradox of life sciences: one of the most consequential industries on the planet, still run on fragmented data and dashboards that report the delay instead of preventing it, while patients pay for the distance.

A global challenge of this scale demands progress on many fronts – in research, in policy, in access, and in the way medicines and medical devices are made and supplied. No single company solves it alone. But much of what turns a scientific breakthrough into a shortage is not science – it is the ability to manufacture and supply at speed, at scale, and within an unforgiving regulatory frame.  This is where SAP can make a difference, and it is what SAP was built for:  to help the world run better and improve people's lives. In this article, I will share how we are answering that call: SAP’s vision and strategy for the life sciences industry, Autonomous Regulated Manufacturing, built to bring artificial intelligence into the center of the design-to-operate value chain for biopharma and medical device companies everywhere. The goal is not incremental. It is to close, for good, the gap between what medicine can do and what reaches the patient who needs it.

The Problem: Why Life Sciences Struggles

The forces converging

Three forces are converging that no operating model from the past can absorb. The first is economic: biopharma faces a steep patent cliff, with BCG estimating that some $150 billion in revenue will be affected through 2027 alone, and revenues from drugs still under patent falling to just 10% of 2021 levels by 2034, even as newer modalities like cell and gene therapy carry far greater manufacturing complexity and cost. The second is regulatory: enforcement is tightening on every front at once: DSCSA serialization in pharma — EU MDR/IVDR and the FDA’s QMSR harmonization (effective 2026) in medical devices, and emerging frameworks for AI and software in medical devices under IEC 62304. The third is structural: supply chain fragility, geopolitical realignment, shortages of regulated talent, and sustainability mandates are straining operations across both industries simultaneously. These pressures are not cyclical – they are structural, and they compound.

Why traditional approaches fall short

Talk to any C-suite executive in pharma and medical devices, and you hear the same story: every step demands exhaustive documentation, and every decision must be backed by reproducible data. Roughly 30% of staff time is spent on documentation-related activities, including product dossiers, machine logs, batch records, and more. A biotech batch record can comprise 5,000 to 45,000 manual entries according to McKinsey.

New product introduction stretches 10 to 15 years in pharma; even incremental device updates can take 13 to 18 months under EU MDR. Overall Equipment Effectiveness averages at 35% according to McKinsey against more than 80% in high-throughput industries, meaning plants sit idle or stopped nearly two-thirds of the time.

The root cause is twofold: fragmented data and processes never built to be lean. Data exists in abundance — across ERP, MES, LIMS, and quality systems — but sits in silos, while the workflows on top of it grew up function by function, layered with handoffs, redundant checks, and controls added over decades of regulation. The result is a value chain where no one can see how a decision in one part will ripple through the rest. Gartner finds fewer than half of life-sciences supply chains can even model how a decision will affect their outcomes. Unable to act early or precisely, companies fall back on the only lever left: buffering against uncertainty with safety stock, over-ordering, and yet more checks — each a hedge that freezes cash and inflates cost, some of it expiring unsold. The industry keeps supplying and stays compliant, but only by absorbing enormous cost to compensate for what it cannot predict. Every decision creates a tradeoff elsewhere, and the web of interdependencies now running through the chain has outgrown what any team could reasonably hold together in real time.

Decades of Lean, Six Sigma, and operational excellence optimizations delivered real value, but mostly in isolation, leaving the core challenge unsolved: resilient, scalable processes across the full value chain. This is not a technology gap. It is a process architecture gap, and closing it means reimagining the operating model, not bolting AI onto existing workflows. Digital applications aggregated data but scaled the silos rather than breaking them.

The opportunity the industry cannot afford to miss

This is the part of the story that gets lost in the noise of shortages and delays: life sciences is entering its most transformative growth cycle in decades, and the companies that cannot manufacture at the speed and scale this growth demands will watch the opportunity pass to someone who can. For example, the global cell and gene therapy market alone was valued at roughly $21 to 25 billion in 2024 and 2025 and is projected to exceed $100 billion by 2034, a growth rate few industries ever see. Personalized medicine, biologics, and precision therapies are moving from experimental to mainstream faster than the manufacturing base built to support them. At the same time, as per the United Nations, the world's population aged 65 and over is projected to double over the next three decades, reaching 1.6 billion people by 2050, more than 16% of humanity, driving sustained, compounding demand for exactly the therapies this industry is racing to bring to market.

The question every life sciences leader must now answer is not whether demand is coming. It is whether their manufacturing and supply chain can meet it when it arrives, or whether the same fragility straining the system today will turn tomorrow's growth into tomorrow's shortage.

Industries that resist orchestrated automation don't decline gradually; they collapse suddenly. The companies that build AI-orchestrated design-to-operate capabilities in 2026 and 2027 will define the industry through 2035, the same window in which cell and gene therapy is projected to grow fivefold and the patients this industry serves will only grow older and more numerous.

The Solution: Introducing Autonomous Regulated Manufacturing

A vision built for the whole, not the parts

At SAP, this is the future we are building toward: Autonomous Regulated Manufacturing for Life Sciences - an AI-powered central nervous system of agents that makes your design-to-operate operations predictive rather than reactive, which enables the value chain to continuously sense risks, anticipate disruptions before they surface, and propose mitigation scenarios that keep it uninterrupted. It is a vision we are realizing, alongside our customers. Complementing SAP’s line of business domain assistants, these industry-specific agents operate across the full value chain, orchestrating processes specific to the life sciences industry, with humans in control and compliance built in by design. The destination is a value chain resilient enough that therapies and medical devices reach the patients who depend on them, reliably and without delay.

Autonomous Regulated Manufacturing is not full self-driving automation, and it does not replace your ERP, MES, LIMS, or any other application, SAP or non-SAP. It is a layer that runs on top of the systems you already have, connecting them rather than replacing them, so adopting it is not a transformation project. Nor is it a fix for one siloed problem; faster documentation or faster production alone still leaves the rest of the chain blind. The real problem runs deeper: a value chain where decisions are made locally but consequences travel across the entire business. Solving that means reimagining the process end to end, as one connected system layered over what you already run, where a change anywhere, like R&D adjusting a molecule specification, immediately reveals its impact on suppliers, production, quality, and delivery, before the decision is made, not months after. That is the vision: AI that helps your people find the right balance across the whole chain, built on your existing landscape, so innovation in one domain never becomes disruption in another.

A vision built with human authority at the center

This is intelligent orchestration with human authority at the center, augmenting your people, not replacing them. Agents surface the right information, propose decisions, and document rationale; you retain oversight while they execute faster. Regulators expect human judgment on critical decisions that safeguard product safety and efficacy. Autonomous Regulated Manufacturing is engineered for exactly that, AI agents reason and recommend; qualified people decide and sign. Compliance is engineered in by design, so every recommendation is traceable, every action auditable, and every decision remains in human hands.

How it Works: The Technology Behind it

Autonomous Regulated Manufacturing rests on three reinforcing layers: Joule, which lets people engage the system in natural language with regulatory context built in; the SAP Autonomous suite and Industry AI, which pair enterprise-wide autonomy with life sciences specific industry context that makes it safe to act; and the SAP Business AI Platform, which governs every agent and — through the SAP Knowledge Graph and Business Data Cloud — connects all your data into one context-aware semantic layer. Together: how people interact with it, what gives it the capacity to act, and what makes it trustworthy enough for a regulated business.

That third layer is what separates SAP from everyone else building AI for life sciences, and the reason comes down to one fact: your ERP is already the brain of your company, and SAP AI is the only AI that thinks with it, not beside it. Every other AI provider builds an agent on top of a single application and calls that intelligence. But an agent that only sees a QMS, or only sees a document repository, is only ever as smart as that one system, blind to the batch, the supplier, the equipment, and the person connected to the event it's looking at. SAP AI starts from the ERP backbone that already connects finance, supply chain, manufacturing, and quality into one operational truth, so it doesn't guess at what's happening in your business. It already knows.

Take, for example, SAP's vision for the CAPA Assistant. A third-party tool finds a quality event inside a QMS, one record, on its own. SAP's vision connects the event to the batch it came from, the equipment that ran it, the operator who executed it, their training record, and the supplier behind the material, all at once, because they were never separate to begin with. Or take SAP's vision for the Document Control Assistant: a third-party system tells you a file is current, and nothing more. SAP's vision tells you what changed in your business today, and exactly which documents, processes, and people that change now affects, before it ever becomes a deviation.

That's the difference between an AI that optimizes a step and one that optimizes the outcome. Developing on anything other than the system that already holds the truth about your business means rebuilding, by hand, the connections SAP already has.

The difference isn't the AI. It's what the AI can access.

The Value: What This is Worth

So what does this predictive, connected value chain deliver in practice? A chain that corrects course early—with people guiding every decision—so disruptions are resolved before they ever reach the patient depending on the therapy or device at the end of the chain. That advantage shows up in hard numbers, across three levers: cost, growth, and risk.

Take a representative $50 billion life sciences company with roughly $17.5 billion in COGS. Based on SAP's industry expertise, at full deployment Autonomous Regulated Manufacturing could unlock on the order of $212 million in annual recurring benefits—about 1.2% of COGS, equivalent to a 0.42 percentage-point EBIT improvement. The potential value spans three levers, and scales with a company's footprint, product mix, and digital maturity.

• On cost: the largest lever at roughly $188 million a year: connected agents cut manufacturing labor (~$50M), overheads (~$37M), scrap (~$35M), and transportation (~$21M), plus lower planning, inventory, and compliance costs across the chain—with end-to-end orchestration adding ~$19 million by eliminating the manual firefighting between steps.
• On growth: higher on-time-in-full (OTIF) performance recovers ~$21 million a year in recurring revenue. Beyond that sits a far larger, episodic prize—because in pharma, a single day of exclusivity for a $3 billion product is worth roughly $8 million, launching a major product 12 to 24 months earlier can translate into $50 to 200 million or more in cumulative lifetime revenue.
• On risk: avoiding penalties and rework from missed regulatory changes protects an estimated ~$3.5 million a year in this model—modest against the real exposure, since FDA warning letters alone correlate with a 7–12% drop in market value within a month. The resilience to absorb a major quality finding is worth far more: serious remediation runs $5–20 million, and full consent decrees have historically cost $100 million to $1 billion. And on supply disruption, McKinsey estimates the risk represents a potential loss of around 25% of EBITA for pharma company—so protecting even $50–200 million in a single crisis turns risk mitigation from insurance into competitive advantage.

That is the collective power of predictive operations: value that does not come from one domain performing better, but from every domain correcting, informing, and strengthening every other, at once.

This is what predictive operations make possible: faster new product introduction, a supply chain that anticipates disruption instead of reacting to it, and manufacturing that scales without scaling risk. But behind every one of these numbers is a moment that either happens on time or doesn't. A surgery that proceeds as scheduled because the device arrived. A clinical trial that stays on protocol because the investigational drug shipped on time. A patient who picks up their prescription because the pharmacy shelf was never empty. Autonomous Regulated Manufacturing exists to bring out the best in every life sciences company, so that when a patient needs care, the chain behind it never breaks. This is what it means to save lives.

How to Get There: Your Path to Adoption

So where do you start, and what should the first move look like? SAP is already walking this path with many of its life science customers. Autonomous Regulated Manufacturing is not a moonshot, and it does not demand a leap of faith. The companies capturing value start with the highest-value use case, execute and deploy it together with SAP, prove the outcome, then expand into the next domain, building credibility and confidence with every step.

You do not need a multi-year transformation program as a prerequisite, and you do not need to build it alone. The approach is modular, outcome-focused, and built one deliberate step at a time, piece by piece, domain by domain, until what emerges is the central nervous system of your enterprise. That is how the companies who lead this industry will get there: not with a single leap, but with a clear strategy, a trusted partner, and the discipline to expand only once value is proven. This is how you build the enterprise that does not just keep pace with the future of life sciences. It leads it.

Conclusion: Lead the Change

Biopharma and medical devices face the same truth from different angles: today's operating model can't deliver the growth or compliance this moment demands.

Autonomous Regulated Manufacturing is a vision being built patiently, not chased quickly, piece by piece, right now, in the same facilities making the medicines and devices patients already depend on. The road is long, and it will test an industry pulled toward short-term urgency at every turn, but the results that change how care reaches people have never come from what's fastest. They come from committing to what's right and staying the course long enough to see it through, and that is the work worth doing now.

Everything good and lasting in life came from a long-term vision, and took the time it needed to grow strong. Choosing Autonomous Regulated Manufacturing today isn't a balance-sheet decision, it's the reason a patient gets the therapy, device, or medication they need, exactly when they need it. This is not abstract industry progress. It's a life changed by a decision made now.

The industry's next chapter is being written now, by the leaders willing to build it, not alone, but together with SAP. The question is not whether AI will transform life sciences. It already is. The only question left is whether you will be one of the leaders who had the conviction to build it, or one who waited to see how it turned out.