Novartis AI Strategy: Productivity Now, R&D Later

Novartis treats AI as a productivity tool now and an R&D bet for later. Its CEO calls himself a heavy user of AI and openly criticizes AI hype. He expects AI to transform drug discovery over seven to ten years, and operations and customer work sooner. In October 2026, Novartis began crediting AI with part of its cost savings, but it has published no figure for what AI delivered.

The ambition runs on two clocks. CEO Vas Narasimhan said in a 2026 Fortune interview in Davos that AI “has the opportunity in a 7 to 10 year horizon to transform R&D in our sector,” but that “more tangibly in the near term it’s clear that AI will transform how we approach customers, how we run our operations.”

The CEO is a heavy user and a skeptic. In November 2026, Narasimhan said: “I have a complex relationship with AI.” He called claims that AI will solve drug discovery within a decade “such an oversell,” and warned that “it’s much harder just to take ChatGPT and throw it into Novartis and think magic’s going to happen” (One Young World). He also said: “I am a heavy user.”

AI sits inside the corporate strategy, not beside it. Novartis lists “scaling data science and technology” among the foundations of its strategy (third-quarter 2026 results). Its 2025 annual report names AI as a strategic risk and warns of the consequences “if we fail to drive an effective enterprise AI strategy.” That enterprise AI strategy has not been published.

The focus is R&D first, then operations. Novartis uses AI across:

• research: finding disease targets in genetic and other biological data, and generative chemistry to design molecules (Novartis, July 2026)
• clinical development: trial design, site selection, simulated “computational twins” of trials, and first drafts of clinical and regulatory documents
• manufacturing and quality: production yield, AI visual inspection and predicting quality deviations (Technology Magazine, August 2026)
• enterprise functions: AI tools in finance, HR and marketing, and AI-led IT operations

Novartis has not ranked these areas or said what it will not prioritize.

R&D runs on partnerships.

• Isomorphic Labs, Alphabet’s AI drug discovery company: three targets, with $37.5 million upfront and up to $1.2 billion in milestones (January 2024, BiopharmaTrend), expanded by up to three more programs in February 2025 (Isomorphic Labs)
• Schrödinger: $150 million upfront and up to $2.3 billion in a computational drug discovery deal (November 2024, Business Wire)
• Chai Discovery: AI antibody discovery using its newest models, terms undisclosed (July 2026, The Pharma Letter)

Operations runs on vendors. In July 2026, Cognizant won a five-year contract to run Novartis’s global IT operations with agentic AI (Cognizant). A September 2026 job posting describes an internal “Agentic Fastrack” program that builds agents on Microsoft Copilot Studio, Azure AI Foundry and Anthropic’s Claude.

Fluency is role-based. Chief Information Officer Bernd Bucher described an AI fluency framework in August 2026. Employees assess their own understanding, then follow learning paths by role, from AI features in Outlook and PowerPoint to advanced training on AI platforms. He said tens of thousands of employees use AI tools in their daily work, and that “confidence comes with actually using it” (Technology Magazine). It is a role-based take on AI fluency.

The US goes further. Since May 2026, about 6,000 Novartis US employees have been expected to complete a nine-week data and AI course built with MIT xPRO by the end of the year. In the final project, each employee builds an AI agent for their own role (Novartis, September 2026). It is a demanding form of company-wide AI training: the end point is building, not just using. Novartis has not reported completion rates.

Ownership is spread. Bucher runs technology delivery, and the heads of research and development, Fiona Marshall and Shreeram Aradhye, front AI in R&D. No Chief AI Officer or AI budget has been disclosed. In April 2026, Narasimhan joined the board of Anthropic, the AI company (Reuters, via Yahoo Finance).

The workforce message is about plants. Speaking about US manufacturing, Narasimhan said that with automation, AI and technology “we need a fraction of the people” (Fortune). Novartis has cut jobs in 2026, including 572 at its East Hanover, New Jersey, site over four rounds (BioSpace, July 2026), but has not attributed the cuts to AI.

Governance is published and has reached the board.

• Novartis first published an AI ethics commitment in 2021 and added an AI Risk & Compliance Management Framework in 2024 (Novartis).
• Every AI use, built or bought, is rated low, mid or high risk, or forbidden, in line with the EU AI Act.
• From 2026, data, digital and AI are a standing board topic (Novartis, February 2026).
• The board has nominated Rajesh Jha, a former Microsoft executive who helped integrate AI across Microsoft’s products, for election in March 2027.

AI fluency, coded against the Lead with AI Fluency Matrix (September 2026 edition):

• Fluency floor: Not published company-wide. A fluency framework sets role-based paths. In the US, about 6,000 employees are expected to build an AI agent for their role in 2026, a Level 3 (Integrating) expectation for one market.
• Highest level evidenced: Level 3 (Integrating), company-claimed. AI is built into manufacturing inspection, trial planning and document drafting, but Novartis has published no figures for these. Level 4 is not evidenced.
• Portfolio design: Multi-level. Role-based learning for everyday users and technical staff, with specialist AI partnerships in R&D.

Novartis has started tying AI to costs, without putting a number on it. On the third-quarter 2026 results call on 27 October, CFO Mukul Mehta said selling, general and administrative costs were 7% lower than a year earlier. He linked the savings to tighter control of more than $11 billion a year in third-party spending, in categories such as agencies, travel and consulting that he said technology and AI have affected (call transcript). The same day, CNBC headlined its interview with Narasimhan: “Novartis CEO says AI has tangibly improved P&L for the first time” (CNBC). The results release itself credits “productivity efforts” and does not name AI.

The R&D claims are anecdotes. Narasimhan said some work that took chemists or biologists six months “can now be done in a day,” and trial analyses that took four to six months now take a week (Fortune). Novartis says AI visual inspection has “significantly reduced” false rejections in manufacturing, but gives no figure. All of these are company claims.

Three things are missing: an AI value figure, AI targets and an employee usage figure. Moving from AI adoption to AI value means putting numbers on the cost savings Novartis now credits partly to AI.

The next marker is close. Novartis’s investor day on 19 November 2026 includes an AI panel (Novartis).

Change log

• 2021: Novartis publishes its first commitment to ethical and responsible use of AI.
• January 2024: The Isomorphic Labs collaboration starts with three targets.
• 2024: The AI Risk & Compliance Management Framework is introduced.
• November 2024: Schrödinger deal, with $150 million upfront.
• February 2025: The Isomorphic Labs collaboration is expanded.
• February 2026: The 2025 annual report names AI a strategic risk, and AI becomes a standing board topic.
• April 2026: Narasimhan joins Anthropic’s board.
• May 2026: Novartis US launches a data and AI course for about 6,000 employees.
• July 2026: The Chai Discovery work expands, Cognizant wins AI-led IT operations, and R&D leaders set out how AI is used across research and development.
• August 2026: The CIO describes the AI fluency framework.
• October 2026: The CFO links lower third-party spending to technology and AI, and the board nominates Rajesh Jha.
• November 2026: Narasimhan calls AI drug discovery claims “such an oversell.”

• Disclosure: 2 of 4. Novartis answers where it will focus and what it will take. Where it is going is set out mainly in CEO interviews, and how it will know is a qualitative link to cost savings.
• It runs AI on two clocks: productivity now, R&D transformation in seven to ten years.
• Its CEO is a heavy user and a skeptic of AI hype, and since April 2026 he has sat on Anthropic’s board.
• Governance is published and has reached the board, with AI a standing board topic from 2026.
• The numbers: since the Sandoz spin-off, net sales rose 20% while headcount fell 1%.
• The say-do gap is evidence. AI now gets credit for cost savings, but Novartis has published no figure for what AI saved or produced.
• Watch the 19 November 2026 investor day. The test is whether the AI panel puts numbers on AI’s contribution.

What the numbers show: Since the Sandoz spin-off, net sales rose 20% while headcount fell 1%, lifting revenue per employee 21%. Novartis is growing without adding people, and its CFO credits part of its 2026 cost savings to technology and AI. These figures do not show how much of the change AI caused.

Revenue, headcount and revenue per employee, FY2022 to FY2025

• FY2022: $50.55bn revenue, 102,000 employees, $496k per employee
• FY2023: $45.44bn revenue, 76,057 employees, $597k per employee
• FY2024: $50.32bn revenue, 75,883 employees, $663k per employee
• FY2025: $54.53bn revenue, 75,267 employees, $725k per employee

Change from FY2023, the first comparable year, to FY2025: revenue up 20%, headcount down 1%, revenue per employee up 21%.

Notes: Revenue is net sales, as reported. Headcount is full-time equivalent employees at year-end. Novartis spun off Sandoz, its generics business, on 4 October 2023. 2022 is as originally reported and includes Sandoz (headcount about 102,000). 2023 to 2025 cover continuing operations only, so the comparable trend starts in FY2023.

These figures give context on scale and productivity. They do not show that AI caused any change. Sources: Novartis 20-F filings.

What is Novartis’s AI strategy?

Novartis uses AI to raise productivity now, in operations, manufacturing and commercial work, and expects AI to transform R&D over seven to ten years. AI sits within the “scaling data science and technology” foundation of its corporate strategy. Novartis has not published a standalone AI strategy.

‍

How does Novartis use AI in drug discovery?

Novartis uses AI to find disease targets in genetic and other biological data and to design molecules with generative chemistry. It works with AI drug discovery companies including Isomorphic Labs, Schrödinger and Chai Discovery.

‍

Who leads AI at Novartis?

Novartis has no Chief AI Officer. Chief Information Officer Bernd Bucher runs technology delivery, including AI. Fiona Marshall and Shreeram Aradhye lead AI use in research and development. CEO Vas Narasimhan sets the tone and joined Anthropic’s board in April 2026.

‍

Is Novartis cutting jobs because of AI?

Novartis has not attributed its 2026 job cuts to AI. Its CEO has said automation and AI mean its US plants need “a fraction of the people,” and its CFO credits technology and AI for part of a fall in overhead costs.

‍

How is this profile put together?

Lead with AI profiles use public sources only, such as annual reports, earnings materials, filings and on-record interviews, dated at the time of writing. AI fluency is coded against the Lead with AI Fluency Matrix. Company figures are marked as company-claimed unless they have been independently verified.