Pharma goes all-in on AI
What happened
- Big pharma is turning AI from experiments into enterprise-wide partnerships for discovery, manufacturing and commercial operations. - Novo Nordisk struck major ties with OpenAI while Merck signed a $1 billion deal with Google Cloud. - These deals show vendors must combine models, cloud and domain integration to win regulated industry business ( ).
Why it matters
Drugmakers are moving artificial intelligence from pilot projects into companywide contracts that cover research, factories, sales teams and back-office work. (merck.com) (novonordisk.com) On April 22, 2026, Merck said it signed a multi-year partnership with Google Cloud worth up to $1 billion to deploy an “agentic” platform across research and development, manufacturing, commercial and corporate functions. Merck said Google Cloud engineers will work alongside its teams and that the rollout will use Gemini Enterprise tools for its 75,000 employees. (merck.com) On April 14, 2026, Novo Nordisk announced a partnership with OpenAI to “transform how medicines are discovered and delivered,” adding one of the industry’s biggest model makers to its technology stack. Novo Nordisk listed the deal in its press materials two weeks before Merck’s announcement. (novonordisk.com) In pharma, “discovery” means finding biological targets and designing molecules; “manufacturing” means running tightly controlled plants that make drugs at the right dose and quality every time. The new deals cover both ends of that chain, not just lab research. (merck.com) (roche.com) Merck said its system will use predictive analytics and automation in manufacturing, data-driven personalization in commercial work and AI-powered tools in corporate operations. That scope is wider than the earlier wave of biotech AI deals, which were often limited to target discovery or molecule design. (merck.com) Other large drugmakers are building the same kind of full-stack setup. Roche said on March 15, 2026 that it had expanded to more than 3,500 Blackwell graphics processing units, or GPUs, across on-premise and cloud systems, and said the infrastructure is embedded across discovery, development, manufacturing and commercialisation. (roche.com) Sanofi has been making a similar case for scale. In May 2025, it said a new manufacturing-and-supply “Digital Accelerator” was part of a plan to become “the first biopharma company powered by AI at scale,” and it said AI could shorten time-to-market by up to one year. (sanofi.com) Sanofi also said its AI programs span research, manufacturing and patient engagement, and that one internal language model for messenger RNA design cut design time by 50%. That helps explain what big vendors now have to sell pharma companies: not just models, but cloud capacity, data integration and industry-specific workflows. (sanofi.com) OpenAI has also been pushing deeper into life sciences. On April 16, 2026, it announced GPT-Rosalind, a research model aimed at life sciences work, days after Novo Nordisk disclosed its partnership. (openai.com) The race is no longer about whether drugmakers will use artificial intelligence. The race is about which vendors can wire models into regulated companies’ data, factories and product launches fast enough to become part of the operating system. (merck.com) (roche.com) (sanofi.com))
Key numbers
- Novo Nordisk struck major ties with OpenAI while Merck signed a $1 billion deal with Google Cloud.
- (merck.com) (novonordisk.com) On April 22, 2026, Merck said it signed a multi-year partnership with Google Cloud worth up to $1 billion to deploy an “agentic” platform across research and development, manufacturing, commercial and corporate functions.
- Merck said Google Cloud engineers will work alongside its teams and that the rollout will use Gemini Enterprise tools for its 75,000 employees.
- (merck.com) On April 14, 2026, Novo Nordisk announced a partnership with OpenAI to “transform how medicines are discovered and delivered,” adding one of the industry’s biggest model makers to its technology stack.
What happens next
- Merck said Google Cloud engineers will work alongside its teams and that the rollout will use Gemini Enterprise tools for its 75,000 employees.
- (novonordisk.com) In pharma, “discovery” means finding biological targets and designing molecules; “manufacturing” means running tightly controlled plants that make drugs at the right dose and quality every time.
- (merck.com) (roche.com) Merck said its system will use predictive analytics and automation in manufacturing, data-driven personalization in commercial work and AI-powered tools in corporate operations.
Quick answers
What happened in Pharma goes all-in on AI?
Big pharma is turning AI from experiments into enterprise-wide partnerships for discovery, manufacturing and commercial operations. Novo Nordisk struck major ties with OpenAI while Merck signed a $1 billion deal with Google Cloud. These deals show vendors must combine models, cloud and domain integration to win regulated industry business ( ).
Why does Pharma goes all-in on AI matter?
Drugmakers are moving artificial intelligence from pilot projects into companywide contracts that cover research, factories, sales teams and back-office work. (merck.com) (novonordisk.com) On April 22, 2026, Merck said it signed a multi-year partnership with Google Cloud worth up to $1 billion to deploy an “agentic” platform across research and development, manufacturing, commercial and corporate functions. Merck said Google Cloud engineers will work alongside its teams and that the rollout will use Gemini Enterprise tools for its 75,000 employees. (merck.com) On April 14, 2026, Novo Nordisk announced a partnership with OpenAI to “transform how medicines are discovered and delivered,” adding one of the industry’s biggest model makers to its technology stack. Novo Nordisk listed the deal in its press materials two weeks before Merck’s announcement. (novonordisk.com) In pharma, “discovery” means finding biological targets and designing molecules; “manufacturing” means running tightly controlled plants that make drugs at the right dose and quality every time. The new deals cover both ends of that chain, not just lab research. (merck.com) (roche.com) Merck said its system will use predictive analytics and automation in manufacturing, data-driven personalization in commercial work and AI-powered tools in corporate operations. That scope is wider than the earlier wave of biotech AI deals, which were often limited to target discovery or molecule design. (merck.com) Other large drugmakers are building the same kind of full-stack setup. Roche said on March 15, 2026 that it had expanded to more than 3,500 Blackwell graphics processing units, or GPUs, across on-premise and cloud systems, and said the infrastructure is embedded across discovery, development, manufacturing and commercialisation. (roche.com) Sanofi has been making a similar case for scale. In May 2025, it said a new manufacturing-and-supply “Digital Accelerator” was part of a plan to become “the first biopharma company powered by AI at scale,” and it said AI could shorten time-to-market by up to one year. (sanofi.com) Sanofi also said its AI programs span research, manufacturing and patient engagement, and that one internal language model for messenger RNA design cut design time by 50%. That helps explain what big vendors now have to sell pharma companies: not just models, but cloud capacity, data integration and industry-specific workflows. (sanofi.com) OpenAI has also been pushing deeper into life sciences. On April 16, 2026, it announced GPT-Rosalind, a research model aimed at life sciences work, days after Novo Nordisk disclosed its partnership. (openai.com) The race is no longer about whether drugmakers will use artificial intelligence. The race is about which vendors can wire models into regulated companies’ data, factories and product launches fast enough to become part of the operating system. (merck.com) (roche.com) (sanofi.com))
