Verified
Novartis
Experimenting- Pharma
- 75,267 employees (Dec 2025)
- No AI chief on Executive Committee
- Copilot and Claude approved
- AI R&D partners
Novartis deploys AI mainly in research and development, building tools on its data42 platform and working with AI discovery partners such as Isomorphic Labs, Generate:Biomedicines, Relation Therapeutics and Chai Discovery. Novartis gives staff generative AI assistants, and 2026 job postings list Microsoft Copilot, Azure AI Foundry, Anthropic Claude and Amazon Quick Suite as approved agent platforms. CEO Vas Narasimhan said in Oct 2026 that AI had begun to improve the P&L, without giving a figure, and Novartis names no single AI owner.
novartis.comTheir approach to AI
- The platform
- Novartis runs no single named AI platform: R&D builds tools such as a Target Discovery Engine on data42, a data platform Novartis began building with Palantir in 2016 or 2017, and 2026 job postings describe an enterprise AI and MLOps platform on AWS and Databricks, an Agentic Factory and a US agentic AI operating system.
- The models
- Novartis lists Microsoft Copilot Studio, Azure AI Foundry, Anthropic Claude and Amazon Quick Suite as approved agent platforms (Sep 2026 job posting); discovery partners such as Isomorphic Labs, Generate:Biomedicines and Chai Discovery apply their own AI models.
- The reach
- Novartis gives employees generative AI assistants, and Technology Magazine reported in Aug 2026 that tens of thousands of Novartis employees use AI tools in their daily work; Novartis has published no exact count and runs an enterprise-wide Ready for AI program (May 2026).
- The catch
- Novartis has published no figure for any AI return: CEO Vas Narasimhan said on Oct 27, 2026 that AI had started to improve the P&L, citing marketing and sales productivity, without quantifying it, and Novartis reports no count of employees using AI tools, names no single owner of AI and discloses no AI spending.
Snapshot
- Employees
- 75,267
- Full-time equivalents at Dec 31, 2025, down from 75,883 a year earlier (2025 Form 20-F)
- Technology budget
- Not disclosed
- R&D expense was $11.2B in 2025; Novartis gives no IT or AI spending figure
- AI/ML headcount
- Not disclosed
- Biomedical Research has about 5,500 scientists and associates; Operations, which includes IT, data and digital, has more than 25,000
- Who owns AI
- No single owner
- CIO Bernd Bucher runs technology in Steffen Lang's Operations unit; Fiona Marshall and Shreeram Aradhye set the R&D AI strategy (Jul 2026)
- Platform
- Hybrid
- data42 built with Palantir and R&D tools built in-house; Copilot Studio, Azure AI Foundry, Claude, Amazon Quick Suite, Databricks bought (2026 job postings)
- Models
- Microsoft, Anthropic, partnersest.
- No public statement on which large language models sit behind internal tools
- Reach
- Tens of thousandsest.
- Employees using AI tools in their daily work (Technology Magazine, Aug 2026); Ready for AI program enterprise-wide
- Use cases
- Research to commercial
- Examples shown to investors across target discovery, chemistry, trials, access and marketing (Nov 2025)
Key findings
Novartis showed investors on Nov 20, 2025 AI examples across its value chain, from a Target Discovery Engine, generative chemistry and predictive safety to clinical document generation, predictive site selection and marketing content, but has not said how many AI use cases run in production or what any of them returns.
Novartis has no AI chief on its Executive Committee: Chief Information Officer Bernd Bucher runs technology inside President of Operations Steffen Lang's unit of more than 25,000 associates, Fiona Marshall and Shreeram Aradhye defined the R&D AI strategy Novartis described in Jul 2026, and Katie Mazuk's US Strategy, Platforms & Transformation organization builds its own agentic AI foundations.
Novartis paid $37.5 million upfront to Isomorphic Labs in Jan 2024 and $65 million to Generate:Biomedicines in Sep 2024 including $15 million of equity, agreed $55 million of upfront, equity and research funding for Relation Therapeutics in Dec 2025, and signed an antibody discovery collaboration with Chai Discovery on undisclosed terms in Jul 2026.
Novartis job postings from Sep and Oct 2026 list Copilot Studio, Azure AI Foundry, Anthropic Claude and Amazon Quick Suite as approved platforms for employee agents, Cognizant signed a five-year AI-enabled IT operations engagement with Novartis in Jul 2026, and Anthropic's Long-Term Benefit Trust appointed CEO Vas Narasimhan to Anthropic's board on Apr 14, 2026.
Novartis CEO Vas Narasimhan said on Oct 27, 2026 that the third quarter was the first in which he saw AI tangibly improve the P&L, and the CFO said AI had significantly affected parts of more than $11 billion a year in third-party spend, but Novartis gave no figure for AI's share; the same day Novartis nominated former Microsoft executive Rajesh Jha to its Board for his AI and digital platform experience.
Where Novartis sits
- Stage
- Experimenting
Novartis has reported no business return it credits to AI, for any use case or for the company as a whole: CEO Vas Narasimhan told CNBC on Oct 27, 2026 that Q3 2026 was the first quarter he saw AI tangibly improve the P&L, citing marketing and sales productivity, and CFO Mukul Mehta said the same day that AI had significantly affected some categories of Novartis's more than $11 billion a year in third-party spend, but Novartis gave no figure for AI's share of either, or of its 7% fall in SG&A. Novartis has also not said how many employees use its AI tools: the tens of thousands using AI daily comes from a Technology Magazine report (Aug 2026), the about 6,000 Novartis US employees in a Data and AI Foundations course (Sep 2026) are trainees in one unit, and AI ownership is split between Chief Information Officer Bernd Bucher in Operations and the R&D and US units. Novartis therefore stays at Experimenting, short of Piloting, which needs one AI-credited return or an officially reported rollout with a user count and a named owner.
How the stages are defined - Ownership
- Split: CIO in Operations runs IT and data; R&D and US units run their own AI
- Platform
- Hybrid: R&D tools on data42, built with Palantir; Microsoft, Anthropic and AWS tools bought
- Models
- Microsoft, Anthropic Claude, Amazon Quick Suite; partners' discovery AI
- Hosting
- Microsoft Azure and AWS cloud platforms, with Databricks for machine learning; Claude hosting and any company AI data centers not disclosed
- Reach
- Tens of thousands use AI tools daily (Aug 2026 report); no exact count
How they deploy
AI aimed at R&D productivity
Novartis lists scaling data science and technology among the foundations of its strategy (Q3 2026 results) and told investors in Nov 2025 that it is scaling AI across the value chain to raise productivity and performance. Novartis said in Jul 2026 that Fiona Marshall, President of Biomedical Research, and Shreeram Aradhye, President of Development, defined an end-to-end strategy for AI across R&D. CEO Vas Narasimhan said on Jul 21, 2026 that Novartis is scaling AI across every phase of R&D, that it will take 8 to 10 years to know the scale of AI's impact, and that he hopes AI can cut discovery-to-patient time from 12 to 14 years to about eight and lift clinical success rates from 8% to 15% or higher.
Partner for platforms, build where products fall short
Novartis says its R&D forms strategic collaborations with AI companies whose platforms complement its own expertise, and develops fit-for-purpose tools internally when off-the-shelf products do not meet the needs of regulated science (Jul 2026). Novartis's 2025 Form 20-F names AI-enabled drug discovery partnerships with Isomorphic Labs, Generate:Biomedicines, Profound Therapeutics and Relation Therapeutics, and Novartis added Chai Discovery's antibody design models in Jul 2026. Novartis also signed a separate computational discovery deal with Schrödinger in Nov 2024 that Schrödinger's filing does not describe as AI.
Data foundations first
Novartis began building data42, its R&D data platform, with Palantir in 2016 or 2017, according to CEO Vas Narasimhan (Jul 2026); data42 held about 2 million patient-years of data from more than 2,000 clinical studies in Mar 2020. Christian Diehl, Chief Data & Digital Officer of Novartis Biomedical Research, wrote in Sep 2026 that data42 began as foundation work with no immediate impact and now underpins Novartis's AI-driven research. Novartis's 2025 Form 20-F describes companywide IT programs to replace outdated systems and create a unified data marketplace, and the Lean Digital Core program launched in 2025.
Agents through low-code platforms
Novartis job postings from Jul to Sep 2026 describe an Agentic Fastrack program in Operations' information technology unit that builds Copilot and Claude agents, retrieval tools and prompt workflows on approved low-code and AI productivity platforms, an Agentic Factory that industrializes validated agent proofs of concept, and an agentic AI operating system for the US business. Novartis has not described these programs in its press releases or investor materials.
IT operations run with an AI-enabled outsourcer
Cognizant announced on Jul 27, 2026 a five-year engagement to run Novartis's applications, infrastructure, security, digital workplace and platform services as one AI-enabled operating model, using Cognizant's Neuro AI platform for predictive operations, self-service and AI-led software engineering. Cognizant did not disclose the value of the Novartis contract.
Augment roles, train everyone
Novartis told investors in May 2026 that it views AI primarily as a tool to augment work, supporting decisions, productivity and the reduction of repetitive tasks, rather than to replace roles, and that it monitors how roles and skills change. Novartis backs that with an enterprise-wide Ready for AI program, role-specific courses such as the Development AI Academy, and a Novartis US Data and AI Foundations course for about 6,000 employees that ends with each participant building an AI agent (Sep 2026).
- Deployment model
- Novartis spreads AI ownership across its organizational units. Operations, led by Steffen Lang, includes information technology, data and digital alongside manufacturing, quality, supply, procurement and global business services; Technology Magazine named Bernd Bucher as Novartis's Chief Information Officer in Aug 2026, and Novartis's 2025 Form 20-F says its CIO has 25 years in IT, including 15 at Novartis, and is responsible for technology strategy, delivery and operations globally. The R&D units run their own data and AI teams: Christian Diehl is Chief Data & Digital Officer of Biomedical Research, and Max Lawson leads the AI in Development team. Novartis US has its own Strategy, Platforms & Transformation organization under Katie Mazuk, which is building an agentic AI operating system (Jul 2026 job posting), and Novartis's Innovation Lab opened the GARAGE by iLab at its US campus in East Hanover, New Jersey in Apr 2025.
- Model providers
- Novartis has not published which large language models sit behind its internal tools. Technology Magazine reported in Aug 2026 that Novartis gives employees access to the latest generative AI assistants and sources high-performance compute through partnerships with global technology providers. A Novartis job posting from Sep 28, 2026 lists Copilot Studio, Azure AI Foundry, Anthropic Claude, Amazon Quick Suite and Power Platform among approved technologies for building agents, retrieval tools and prompt workflows, and prefers candidates with Copilot or Claude certification. Novartis named Microsoft its strategic AI and data science partner when the two set up an AI innovation lab in Oct 2019. Anthropic's Long-Term Benefit Trust appointed CEO Vas Narasimhan to Anthropic's board on Apr 14, 2026; Anthropic's announcement does not mention Novartis as a customer. In research, Isomorphic Labs, Generate:Biomedicines and Chai Discovery, whose Chai-3 model Novartis can use, apply their own AI platforms to Novartis targets.
- Data and infrastructure
- Novartis's data42 platform, which Novartis began building with Palantir in 2016 or 2017, held about 20 petabytes of R&D data in Mar 2020. Novartis and AWS announced a multiyear collaboration in Dec 2019 to build an enterprise-wide data and analytics platform, starting in manufacturing and supply, and an Oct 13, 2026 job posting for a director of AWS engineering describes Novartis's AWS-based cloud platforms and asks for experience with AWS AI services such as Bedrock and AgentCore. A Sep 2026 posting describes enterprise AI, generative AI and MLOps platform services on AWS with Databricks-based machine learning and generative AI workflows, and the Agentic Fastrack posting names Azure AI Foundry. Cognizant announced in Jul 2026 a five-year engagement to run Novartis's IT operations under an AI-enabled model. Novartis has not disclosed where Claude runs or whether it hosts any AI models in its own data centers. Steffen Lang said in Jul 2026 that Novartis seeks AI technology partners with credible climate commitments, given the energy data centers use.
Platforms and internal tools
data42Begun 2016 or 2017; live by Mar 2020
data42 is Novartis's R&D data platform, built with Palantir and combining clinical, molecular, preclinical and biology data; it held about 2 million patient-years of data from more than 2,000 clinical studies in Mar 2020, and Christian Diehl wrote in Sep 2026 that it underpins Novartis's AI-driven research.
Target Discovery EngineBy Nov 2025
The Target Discovery Engine is a Novartis research tool that Novartis listed among its AI examples in Nov 2025; Novartis says its AI models integrate human genetic and multi-omic data, imaging, clinical trial results and published research to surface disease drivers.
Intelligent Decision System (IDS)By Nov 2025
The Intelligent Decision System is a Novartis tool that Novartis placed between trial design and trial execution in its Nov 2025 AI examples; Novartis has not published what it does.
AI Content EngineBy Oct 2026
The AI Content Engine is a Novartis product in the International unit that speeds the creation, personalization and localization of commercial and medical content for healthcare professionals; an Oct 2026 job posting seeks an analyst to roll it out across countries.
Agentic Fastrack and Agentic FactoryBy Aug 2026
Agentic Fastrack is a Novartis IT program that builds Copilot and Claude agents on approved low-code and AI productivity platforms, and the Agentic Factory industrializes agent proofs of concept validated by a US AI architecture and lab team, according to Novartis job postings from Aug and Sep 2026.
- Models
- Copilot Studio, Azure AI Foundry, Anthropic Claude, Amazon Quick Suite
Enterprise AI and MLOps platformBy Sep 2026
Novartis's enterprise AI, generative AI and MLOps platform provides model serving, inference services, orchestration and self-service provisioning on AWS, with Databricks-based machine learning and generative AI workflows, according to a Sep 2026 job posting.
Vendors and partners
MicrosoftOct 2019
Strategic AI and data science partner; Copilot, Copilot Studio, Azure AI Foundry and Power Platform · Novartis and Microsoft set up an AI innovation lab in Oct 2019 with a goal of bringing AI to every Novartis associate's desktop; Microsoft tools are approved for Novartis agents in 2026.
Amazon Web ServicesDec 2019
Cloud platforms and data and analytics; Amazon Quick Suite approved for agents · AWS and Novartis announced a multiyear collaboration in Dec 2019 to build an enterprise-wide data and analytics platform, starting in Novartis Technical Operations.
Palantir2016 or 2017
Built the data42 R&D data platform with Novartis · CEO Vas Narasimhan said on Jul 21, 2026 that Novartis started working with Palantir in 2016 or 2017 to create data42.
AnthropicBy Sep 2026
Claude approved for building employee agents · Novartis CEO Vas Narasimhan has sat on Anthropic's board since Apr 14, 2026, appointed by Anthropic's Long-Term Benefit Trust.
DatabricksBy Sep 2026
Machine learning and generative AI workflows on Novartis's AI platform
CognizantExpanded Jul 2026
AI-enabled integrated IT operations across applications, infrastructure, security and digital workplace · Cognizant announced a five-year engagement on Jul 27, 2026, run on its Neuro AI platform; the value was not disclosed.
Isomorphic LabsJan 2024
AI drug discovery for small molecules · Novartis paid $37.5 million upfront in Jan 2024; in Feb 2025 the collaboration added up to three research programs on the same financial terms.
Generate:BiomedicinesSep 2024
Generative AI design of protein therapeutics · Novartis paid $65 million upfront, including a $15 million equity purchase, with more than $1 billion in potential milestones.
SchrödingerNov 2024
Computational discovery of small molecules · Novartis agreed to pay $150 million upfront and up to $2.272 billion in milestones across the initial projects (Schrödinger 8-K, Nov 2024).
Relation TherapeuticsDec 2025
Machine learning target discovery for atopic diseases · Relation receives $55 million in upfront payment, equity investment and research funding, with about $1.7 billion in potential milestones (Dec 9, 2025).
Chai DiscoveryJul 2026
AI antibody design models, including Chai-3 · Chai Discovery announced the multi-target antibody discovery collaboration on Jul 13, 2026 without financial terms.
Profound TherapeuticsBy 2025
AI-enabled drug discovery partnership · Novartis's 2025 Form 20-F names Profound Therapeutics among its AI-enabled R&D partnerships; Novartis has not published the terms.
Leadership
AI, data and platform leaders
Executives
Board
AI org map
No central AI office; IT, data and digital in Operations, with R&D, US and commercial units running their own AI teams
Novartis created a Chief Digital Officer role reporting to the CEO in Jan 2018, combined that role with leadership of a new Customer & Technology Solutions unit in Feb 2021, and its Chief Data and AI Officer announced in Feb 2023 that he was stepping down; its Executive Committee in Nov 2026 has no digital, data or AI chief. Information technology, data and digital now sit in Operations under Steffen Lang, where Chief Information Officer Bernd Bucher runs technology globally and the Chief Security Officer reports to the CIO. Biomedical Research and Development each have data and AI leaders, and Novartis US builds its own agentic AI foundations under Katie Mazuk.
Novartis has not disclosed how many people work on AI or data science. Novartis's AI work sits in four places: Operations, whose more than 25,000 associates include the IT, data and digital organization under Chief Information Officer Bernd Bucher; the R&D units, with about 5,500 people in Biomedical Research and 13,530 full-time employees in Development, each with its own data and AI leaders; Novartis US, whose Strategy, Platforms & Transformation organization and US Chief Data Officer run data and AI for the US business; and commercial teams such as the Innovation Lab. Novartis job postings from Jul to Oct 2026 seek AI foundations engineers and agentic AI proof-of-concept leads in the US, Claude and Copilot specialists and AWS engineering leaders in Hyderabad, and a site AI champion at the Puurs plant in Belgium.
Reporting lines
Chief Executive Officer
Operations
Steffen Lang, President
More than 25,000 associates in manufacturing, quality, supply, IT, data and digital, procurement and global business services
Information technology, data and digital
Bernd Bucher, Chief Information Officer
Technology strategy, delivery and operations globally; Agentic Fastrack, AWS engineering and enterprise AI platforms sit here according to 2026 job postings
Information Security & Compliance
Chief Security Officer (not named)
Biomedical Research
Fiona Marshall, President
About 5,500 scientists and associates across nine sites
Biomedical Research Data & Digital
Christian Diehl, Chief Data & Digital Officer
Runs data42
Development
Shreeram Aradhye, President and Chief Medical Officer
13,530 full-time employees (Dec 2025)
AI in Development
Max Lawson, Head of Process Innovation
Reporting line inferred from Lawson's role in Development; not published
Novartis US
Victor Bulto, President
US Strategy, Platforms & Transformation
Katie Mazuk, Chief Strategy, Platforms & Transformation Officer
Builds an agentic AI operating system (Jul 2026 job posting); reporting line inferred from the Novartis US leadership team page
People & Organization
Rob Kowalski
Innovation Lab
Sterlena Taylor, Head
Runs the GARAGE by iLab at East Hanover, New Jersey; reporting line not published
Board Science & Technology Committee
John D. Young, Chair (Dec 2025)
Reviews AI use cases and the science, data and technology strategy
Novartis's Executive Committee page, checked Nov 1, 2026, lists ten members and no digital, data or AI officer. Novartis's 2025 Form 20-F describes the Chief Information Officer without a name; Technology Magazine named Bernd Bucher in Aug 2026. The CIO's reporting line to Steffen Lang is inferred from Lang's biography, which places information technology, data and digital in Operations, and from 2026 job postings that place IT roles in the Operations division. The Board dissolved its Risk Committee at the 2026 AGM and now reviews strategic risks, including AI, as a full Board.
Where the AI workforce sits
IT, data and digital in Operations
Runs enterprise platforms, AWS engineering, the Agentic Fastrack program and AI platform services according to 2026 job postings.
Not disclosedR&D data science and AI teams
Biomedical Research and Development each have data and AI leaders; Novartis gives no headcount.
Not disclosedNovartis US and commercial AI teams
Novartis US builds agentic AI foundations and has a US Chief Data Officer; the International unit runs the AI Content Engine.
Not disclosed
Named people
Use cases
Novartis showed investors AI examples across its value chain on Nov 20, 2025, from a Target Discovery Engine, generative chemistry and predictive safety to protocol design, clinical document generation, site selection, market access, marketing content and field enablement. Novartis has not published a business return for any named AI use case or for AI as a whole: CEO Vas Narasimhan said on Oct 27, 2026 that AI had started to improve the P&L on a tangible basis, citing marketing and sales productivity, without a figure, and the one quantified operations example, about 6% lower electricity use at a Singapore plant (Jul 2026), is credited to process intelligence and real-time energy monitoring rather than to AI.
| Use case | Area | Status | ROI |
|---|---|---|---|
| Target Discovery Engine AI models combine human genetic and multi-omic data, imaging, clinical trial results and published research to surface disease drivers and strengthen confidence in the targets Novartis chooses. | Biomedical Research | In production | |
| Generative chemistry and molecule optimization Novartis teams apply generative approaches to explore large molecular design spaces digitally and move from hypothesis to drug candidate with fewer iterations. | Biomedical Research | In production | |
| Biologics optimization Novartis applies AI to design and optimize biologics, including a protein therapeutics collaboration that uses Generate:Biomedicines' generative AI platform. | Biomedical Research | In production | |
| Predictive and preclinical safety Models that learn from Novartis preclinical and clinical data flag potential safety concerns earlier, informing which candidates to advance, refine or stop. | Biomedical Research | In production | |
| Partner-led small-molecule discovery Isomorphic Labs and Schrödinger run multi-target discovery collaborations on Novartis targets with their AI and computational platforms; Novartis has not disclosed candidates from either. | Biomedical Research | Pilot | |
| Protocol design and trial simulation Generative AI paired with knowledge graphs summarizes prior study results to help draft protocols, and trial computational twins predict the operational effect of design choices. | Development | In production | |
| Clinical document generation Generative AI drafts clinical and regulatory documents from approved sources for Novartis experts to review, a skill taught in the Development AI Academy. | Development | In production | |
| Predictive site selection and patient finding Novartis listed predictive site selection and patient finding among its study start-up AI examples in Nov 2025, without further detail. | Development | In production | |
| ClinOps next-best-action Novartis trial teams use AI-enhanced analytics to spot early risks in supply, monitoring and recruitment and to recommend next best actions that keep studies on schedule. | Development | In production | |
| AI Content Engine for marketing and medical content Novartis's AI Content Engine speeds the creation, personalization and localization of commercial and medical content for healthcare professionals and is being rolled out across countries. | International | In production | |
| Field enablement and market access Novartis listed AI for field teams, channel optimization and contract operations, and insights and knowledge navigation among its commercial examples in Nov 2025, without further detail. | Commercial | In production | |
| Energy and process optimization at plants Steffen Lang said in Jul 2026 that digital tools, advanced analytics and AI give Novartis sites real-time insight to optimize energy use, strengthen predictive maintenance and reduce waste. | Operations | In production | |
| Budgeting and forecasting Novartis Finance increasingly uses AI for budgeting, forecasting and basic analytical tasks (Nov 2025). | Finance | In production | |
| Copilot and Claude agents The Agentic Fastrack program builds Copilot and Claude agents, retrieval tools and prompt workflows for business teams on approved low-code platforms, with guardrails for GxP, GDPR and access control. | Firmwide | Pilot | |
| Agentic AI operating system Novartis US is building reusable platform services, tools and components for agentic AI products and workflows, and US proof-of-concept leads hand validated agents to an Agentic Factory to industrialize. | Novartis US | Pilot |
Adoption
Novartis has published no exact count of employees who use AI tools; Technology Magazine reported in Aug 2026 that Novartis gives employees the latest generative AI assistants and that tens of thousands of them use AI tools in their daily work. Novartis said in May 2026 that it is rolling out a Ready for AI program across the company, with an AI Foundations course, an AI Immersion Month on specific tools, a learning agent and a champions community, alongside role-specific programs such as the Development AI Academy. Novartis US said in Sep 2026 that about 6,000 US employees are taking a nine-week Data and AI Foundations course launched in May 2026, ending with each participant building an AI agent. Novartis's mandatory compliance training in 2025 covered data privacy and AI, and the Puurs plant in Belgium is hiring an AI champion to drive adoption of AI and agent-based solutions on Novartis's globally available platforms (Sep 2026 job posting).
By the numbers
Rollout, first pilot to today
2016
Novartis began working with Palantir in 2016 or 2017 to create the data platform it calls data42, CEO Vas Narasimhan said in Jul 2026.
Feb 2017
Vas Narasimhan, then Novartis's head of drug development, said Novartis processes about 1 million safety cases a year and was looking at natural language processing for a first pass, and that it was analyzing 10 years of trial history to choose trial sites.
Jan 2018
Novartis's first Chief Digital Officer started in a new role reporting to the CEO, charged with data use in R&D, engagement with patients and doctors, and automation.
Oct 2019
Novartis and Microsoft set up an AI innovation lab, with Microsoft as strategic AI and data science partner and a goal of bringing AI to every Novartis associate's desktop.
Dec 2019
AWS and Novartis announced a multiyear collaboration to build an enterprise-wide data and analytics platform, starting in Novartis Technical Operations.
Mar 2020
Novartis's data42 platform held about 2 million patient-years of data from more than 2,000 clinical studies.
2021
Novartis published its first commitment to the ethical and responsible use of AI.
Feb 2021
Novartis combined its Chief Digital Officer role with the leadership of a new Customer & Technology Solutions unit after the first Chief Digital Officer left.
Feb 2023
Novartis's Chief Data and AI Officer announced he was stepping down.
Jan 2024
Isomorphic Labs signed a multi-target small-molecule discovery collaboration with Novartis for $37.5 million upfront.
2024
Novartis adopted its AI Risk & Compliance Management Framework and an Ethical Use of Data & Technology Policy with an AI Handbook.
Sep 2024
Generate:Biomedicines signed a multi-target protein therapeutics collaboration with Novartis, with $65 million upfront including $15 million of equity.
Feb 2025
Isomorphic Labs and Novartis expanded their collaboration by up to three research programs.
Apr 2025
Novartis's Innovation Lab opened the GARAGE by iLab at its East Hanover, New Jersey campus to build AI simulations and extended reality tools.
Nov 2025
Novartis showed investors AI examples from target discovery to patient support under the heading of scaling AI across the value chain.
May 2026
Novartis said it was rolling out the Ready for AI fluency program across the company, with a Development AI Academy for clinical teams.
May 2026
Novartis US launched a nine-week Data and AI Foundations course for about 6,000 employees.
Jul 2026
Cognizant announced a five-year, AI-enabled engagement to run Novartis's global IT operations.
Sep 2026
Novartis job postings described the Agentic Fastrack and Agentic Factory programs building agents on Copilot Studio, Azure AI Foundry, Claude and Amazon Quick Suite.
Economics & controls
Spend and ROI
Technology budget
Novartis does not disclose a technology, IT or AI budget. Novartis's research and development expense was $11.2 billion in 2025, and Novartis has not said how much of it goes to AI. Novartis launched the Lean Digital Core program in 2025 to rebuild core processes and technologies in Germany, the UK, the Netherlands and the US, and Board committees monitored it alongside an Enterprise Data Governance and Management program. Novartis's CFO said on Oct 27, 2026 that Novartis spends upwards of $11 billion a year with third parties, on items such as agencies, travel and consulting.
What AI itself costs
Novartis's disclosed AI spending is its discovery deals: $37.5 million upfront to Isomorphic Labs (Jan 2024), $65 million to Generate:Biomedicines including $15 million of equity with more than $1 billion in potential milestones (Sep 2024), and $55 million of upfront payment, equity and research funding for Relation Therapeutics with about $1.7 billion in potential milestones (Dec 2025); the Chai Discovery (Jul 2026) and Cognizant (Jul 2026) terms are undisclosed. Isomorphic Labs valued its Novartis and Eli Lilly deals together at nearly $3 billion excluding royalties. Novartis separately agreed to pay Schrödinger $150 million upfront, with up to $2.272 billion in milestones, for computational discovery (Nov 2024). An Oct 2026 Novartis job posting describes an AI sourcing and savings roadmap with value-based and consumption-based contracts for AI software and services.
Reported ROI
Novartis has reported no financial figure for the return on AI. CEO Vas Narasimhan told CNBC on Oct 27, 2026 that Q3 2026 was the first quarter he had seen AI improve Novartis's P&L on a tangible basis, pointing to productivity in marketing and sales, and the CFO said the same day that technology and AI have significantly affected some third-party spend categories; Novartis's SG&A was 7% lower in Q3 2026 than a year earlier, but Novartis did not say how much of that came from AI. The Board's 2025 assessment of Narasimhan lists AI-enabled R&D and commercial programs without a figure, and Steffen Lang's example of about 6% lower electricity use at the Singapore plant (Jul 2026) is credited to process intelligence and real-time energy monitoring, not to AI.
Productivity and headcount
Novartis told investors in May 2026 that it views AI mainly as a tool to augment work rather than replace roles, and that it monitors how roles and skills change. Novartis had 75,267 full-time employees at Dec 31, 2025, down from 75,883 a year earlier, and has not tied that change to AI.
Governance and risk
AI risk framework
Novartis adopted an AI Risk & Compliance Management Framework in 2024, supported by a dedicated team, that covers internally developed, co-developed and purchased AI, classifies each use case as low, mid, high or forbidden risk, records the EU AI Act classification, and addresses transparency, explainability and human oversight. Novartis said in Jul 2026 that it ran a gap assessment against the EU AI Act and other emerging regulations in 2025 and 2026. Novartis's Audit and Compliance Committee received a presentation on AI risk governance and the risk classification of AI applications in 2025.
Principles and policy
Novartis published its first commitment to the ethical and responsible use of AI in 2021 and consolidated its eight principles into four in 2024 (respect humanity, be transparent and collect fairly, use responsibly, protect data and technology), set out in an Ethical Use of Data & Technology Policy and an AI Handbook.
Board oversight
Novartis's Science & Technology Committee, chaired at the end of 2025 by John D. Young, reviewed AI use cases in 2025, the governance committee discussed AI's effect on jobs and upskilling, and the Risk Committee examined the strategic risk of an AI-powered research, development and commercial continuum before it was dissolved at the 2026 AGM. Novartis made data, digital and AI a standing Board topic from 2026, counted four of twelve directors with data and digital expertise in Feb 2026, and had each Board member complete a Data and Technology Curriculum covering data privacy and AI in 2025.
Controls on agents
Novartis's 2026 job postings require agent builders to meet GxP, GDPR, identity and access management and responsible AI requirements and to set guardrails and monitoring for each solution. Participants at Novartis's 2026 Risk Intelligence Forum, which brought together senior leaders and external experts, said that as AI becomes more autonomous, governance must supervise AI-driven actions, not only AI-generated outputs.
Disclosed risks
Novartis's 2025 Form 20-F lists the failure to implement AI solutions and capture AI's opportunities as a risk, warning of competitors with more sophisticated models or partnerships, flawed or biased algorithms, hallucinated outputs, data rights and diverging regulation across jurisdictions.
Cybersecurity
Novartis's Information Security & Compliance team is headed by a Chief Security Officer who reports to the Chief Information Officer and follows the NIST Cybersecurity Framework and ISO 27001; the Executive Committee oversees cybersecurity strategy, according to the 2025 Form 20-F.
Recent signals
Oct 27, 2026
Novartis CEO Vas Narasimhan told CNBC that Q3 2026 was the first quarter he had seen AI tangibly improve the P&L, citing marketing and sales productivity, without giving a figure.
Oct 27, 2026
Novartis nominated Rajesh Jha, a former Microsoft executive vice president who helped integrate AI across Microsoft's products, for election to its Board, citing his AI and digital platform experience.
Sep 28, 2026
A Novartis job posting for its Agentic Fastrack program listed Copilot Studio, Azure AI Foundry, Anthropic Claude and Amazon Quick Suite as approved platforms for employee agents.
Sep 16, 2026
Novartis US said about 6,000 employees are taking a Data and AI Foundations course built with MIT xPRO content, in which each participant designs and builds an AI agent.
Aug 2026
Technology Magazine profiled Novartis Chief Information Officer Bernd Bucher and reported that tens of thousands of Novartis employees use AI tools in their daily work.
Jul 28, 2026
Novartis set out the end-to-end R&D AI strategy defined by Fiona Marshall and Shreeram Aradhye, from target identification to clinical documents.
Jul 27, 2026
Cognizant announced a five-year, AI-enabled engagement to run Novartis's applications, infrastructure, security, digital workplace and platform services on its Neuro AI platform.
Jul 27, 2026
Steffen Lang said digital tools, advanced analytics and AI help Novartis sites optimize energy use, and Novartis, saying AI was among the topics investors raised most in 2026, planned an AI-focused panel for its Nov 19, 2026 investor event.
Jul 21, 2026
Vas Narasimhan said Novartis is scaling AI across every phase of R&D, that data42 began with Palantir in 2016 or 2017, and that the scale of AI's impact will take 8 to 10 years to show.
Jul 13, 2026
Chai Discovery announced a multi-target antibody discovery collaboration giving Novartis access to its AI design models, including Chai-3, without financial terms.
May 2026
Novartis said it views AI as a tool to augment rather than replace roles and is rolling out its Ready for AI program across the company.
Apr 14, 2026
Anthropic's Long-Term Benefit Trust appointed Novartis CEO Vas Narasimhan to Anthropic's board, giving Trust-appointed directors a board majority.
Feb 2026
Novartis told investors that data, digital and AI become a standing Board topic from 2026 and that it is assessing whether to add specific AI skills to the Board.
Dec 9, 2025
Relation Therapeutics signed a target discovery collaboration with Novartis in atopic diseases worth $55 million in upfront payment, equity and research funding and about $1.7 billion in potential milestones.
Nov 20, 2025
Novartis presented investors with AI examples across its value chain, from a Target Discovery Engine to field enablement and patient support.
Quick answers
- How does Novartis deploy AI?
- Novartis deploys AI mainly in research and development, building tools on its data42 platform and working with AI discovery partners such as Isomorphic Labs, Generate:Biomedicines, Relation Therapeutics and Chai Discovery. Novartis gives staff generative AI assistants, and 2026 job postings list Microsoft Copilot, Azure AI Foundry, Anthropic Claude and Amazon Quick Suite as approved agent platforms. CEO Vas Narasimhan said in Oct 2026 that AI had begun to improve the P&L, without giving a figure, and Novartis names no single AI owner.
- What does Novartis use AI for?
- Novartis showed investors AI examples across its value chain on Nov 20, 2025, from a Target Discovery Engine, generative chemistry and predictive safety to protocol design, clinical document generation, site selection, market access, marketing content and field enablement. Novartis has not published a business return for any named AI use case or for AI as a whole: CEO Vas Narasimhan said on Oct 27, 2026 that AI had started to improve the P&L on a tangible basis, citing marketing and sales productivity, without a figure, and the one quantified operations example, about 6% lower electricity use at a Singapore plant (Jul 2026), is credited to process intelligence and real-time energy monitoring rather than to AI.
- Who leads AI at Novartis?
- AI at Novartis is led by Bernd Bucher (Chief Information Officer); Christian Diehl (Chief Data & Digital Officer, Novartis Biomedical Research); Katie Mazuk (Chief Strategy, Platforms & Transformation Officer, Novartis US). Ownership of AI on the org chart: Split: CIO in Operations runs IT and data; R&D and US units run their own AI.
- What AI platform does Novartis use?
- Novartis runs no single named AI platform: R&D builds tools such as a Target Discovery Engine on data42, a data platform Novartis began building with Palantir in 2016 or 2017, and 2026 job postings describe an enterprise AI and MLOps platform on AWS and Databricks, an Agentic Factory and a US agentic AI operating system.
- Which AI models does Novartis use?
- Novartis lists Microsoft Copilot Studio, Azure AI Foundry, Anthropic Claude and Amazon Quick Suite as approved agent platforms (Sep 2026 job posting); discovery partners such as Isomorphic Labs, Generate:Biomedicines and Chai Discovery apply their own AI models.
- How many Novartis employees use AI?
- Novartis gives employees generative AI assistants, and Technology Magazine reported in Aug 2026 that tens of thousands of Novartis employees use AI tools in their daily work; Novartis has published no exact count and runs an enterprise-wide Ready for AI program (May 2026).
- How far has AI spread at Novartis?
- Novartis has reported no business return it credits to AI, for any use case or for the company as a whole: CEO Vas Narasimhan told CNBC on Oct 27, 2026 that Q3 2026 was the first quarter he saw AI tangibly improve the P&L, citing marketing and sales productivity, and CFO Mukul Mehta said the same day that AI had significantly affected some categories of Novartis's more than $11 billion a year in third-party spend, but Novartis gave no figure for AI's share of either, or of its 7% fall in SG&A. Novartis has also not said how many employees use its AI tools: the tens of thousands using AI daily comes from a Technology Magazine report (Aug 2026), the about 6,000 Novartis US employees in a Data and AI Foundations course (Sep 2026) are trainees in one unit, and AI ownership is split between Chief Information Officer Bernd Bucher in Operations and the R&D and US units. Novartis therefore stays at Experimenting, short of Piloting, which needs one AI-credited return or an officially reported rollout with a user count and a named owner.
Open questions
- How many Novartis employees use Microsoft Copilot, Claude or other AI tools each week, and how far has the Ready for AI program reached?
- Does any single Novartis executive own enterprise AI and report its results, and how does CIO Bernd Bucher's remit relate to the R&D and US AI teams?
- How much of the P&L improvement Vas Narasimhan attributed to AI in Q3 2026, and of the 7% fall in SG&A, came from AI?
- Which large language models sit behind Novartis's internal tools, and is Claude reached directly or through a cloud provider?
- What business return have Novartis's deployed AI use cases delivered, such as the Target Discovery Engine, clinical document generation, ClinOps next-best-action and the AI Content Engine?
- Have the Isomorphic Labs, Generate:Biomedicines, Relation Therapeutics and Chai Discovery collaborations produced development candidates, and what milestones has Novartis paid?
- How much does Novartis spend on technology and on AI each year, including the Cognizant IT operations contract?
- What does Novartis's Intelligent Decision System do, and where is it used?
- What arrangements govern Novartis's purchases from Anthropic now that Vas Narasimhan sits on Anthropic's board?
- What will Novartis disclose about AI in R&D at its Nov 19, 2026 investor panel?
Sources
Novartis
- Executive Committee
- Vas Narasimhan, CEO, biography
- Steffen Lang, President, Operations, biography
- Fiona Marshall, President, Biomedical Research, biography
- Shreeram Aradhye, President, Development, biography
- Victor Bulto, President, US, biography
- Rob Kowalski, Chief People & Organization Officer, biography
- Mukul Mehta, CFO, biography
- Katie Mazuk, Novartis US leadership team
- John D. Young, Board member, biography
- Smarter decisions at scale: How Novartis is using AI to advance R&D (Jul 28, 2026)
- Christian Diehl: The day AI made me rethink what science could do (Sep 23, 2026)
- Novartis US invests in data and AI capabilities for 6,000 employees (Sep 16, 2026)
- Max Lawson: How AI helps us bring medicines to patients faster (Mar 10, 2026)
- Building the future at the Novartis Innovation Lab (Mar 18, 2026)
- Rob Kowalski: Leadership choices in the age of AI (Jul 27, 2026)
- data42 program story (Mar 11, 2020)
- Commitment to ethical and responsible use of AI
- Novartis and Microsoft AI collaboration (Oct 1, 2019)
- AWS strategic collaboration with Novartis (Dec 4, 2019)
- Novartis appoints a Chief Digital Officer (Aug 24, 2017)
Novartis job postings
- Agentic Fastrack: Claude and Copilot specialist (Sep 28, 2026)
- Director, AI Foundations Engineer, US business (Jul 7, 2026)
- Associate Director, Agentic AI PoC Lead (Aug 21, 2026)
- Business Analyst, AI Content Engine (Oct 13, 2026)
- Director, AWS Engineering (Oct 13, 2026)
- Senior Specialist, Platform Analytics AI on AWS (Sep 8, 2026)
- Site IT/OT AI Champion, Puurs (Sep 1, 2026)
- Global IT Category AI Transformation Manager (Oct 7, 2026)
Filings and investor materials
- Novartis 2025 Form 20-F (filed Feb 4, 2026)
- Novartis Annual Report 2025
- Report on Nonfinancial Matters 2025
- Q3 2026 results media release (Oct 27, 2026)
- Q3 2026 investor presentation (Oct 27, 2026)
- Q3 2026 earnings call transcript, Investing.com (Oct 27, 2026)
- Meet Novartis Management 2025 presentation (Nov 20, 2025)
- Meet Novartis Management 2025 speaker biographies
- Q4 2025 impact and sustainability update (Feb 2026)
- Q1 2026 impact and sustainability update (May 2026)
- Q2 2026 impact and sustainability update (Jul 2026)
- Schrödinger 8-K on the Novartis agreement (Nov 12, 2024)
Partners
- Isomorphic Labs: two pharmaceutical collaborations (Jan 7, 2024)
- Isomorphic Labs: Novartis collaboration expansion (Feb 18, 2025)
- Generate:Biomedicines: multi-target collaboration with Novartis (Sep 24, 2024)
- Chai Discovery: collaboration with Novartis on antibody discovery (Jul 13, 2026)
- Cognizant: expanded partnership to run Novartis IT operations with agentic AI (Jul 27, 2026)
- Anthropic: Vas Narasimhan appointed to the board (Apr 14, 2026)
News
- CNBC: Novartis CEO says AI has tangibly improved P&L for the first time (Oct 27, 2026)
- CNBC: Companies promised AI gains, and how the payoff showed up in earnings (Nov 1, 2026)
- Technology Magazine: Novartis, reimagining medicine with AI at scale (Aug 2026)
- Technology Magazine: Bernd Bucher, CIO of Novartis (Aug 2026)
- Yahoo Finance: AI is helping speed up drug discovery timelines, Novartis CEO says (Jul 21, 2026)
- Pharmaceutical Executive: Relation Therapeutics and Novartis collaboration (Dec 9, 2025)
- CDO Magazine: Miruna Sasu becomes US Chief Data Officer at Novartis (Nov 2025)
- BioCentury: Novartis Chief Data and AI Officer steps down (Feb 8, 2023)
- Netzwoche: Novartis gets a new digital chief (Jan 27, 2021)
- pharmaphorum: AI helping us make smarter, faster decisions, says Novartis R&D chief (Feb 2017)
- Business Today: Vas Narasimhan joins Anthropic's board (Apr 15, 2026)
- IMD: Novartis AI maturity profile (2025)
Data confidence
- Confirmed
- Stated in a primary or reputable public source.
- Estimated
- Derived or approximated from public reporting.
- Unconfirmed
- Reported, but not yet corroborated.
Novartis profile · Enterprise AI diffusion report by Imagine · facts checked against their sources on Nov 1, 2026
