MEMO FROM THE FUTURE: ASTRAZENECA

CEO Edition

BOARD STRATEGY SESSION June 2030

TO: AstraZeneca Board of Directors

FROM: Pascal Soriot, CEO

DATE: June 2030

SUBJECT: AI-Driven Oncology Pipeline and the Future of Drug Discovery

OPENING

When I took over as CEO in 2012, AstraZeneca was a company rebuilding after losing patent cliffs. We've spent eighteen years re-establishing ourselves as an innovation-driven pharma company, investing heavily in R&D and building a portfolio of first-in-class medications across oncology, cardiovascular, and respiratory diseases.

Today, I'm presenting a memo about how AI is fundamentally reshaping drug discovery timelines and success rates in ways that could accelerate our most ambitious pipeline targets by 5-10 years.

This is not hyperbole. This is a structural shift in how molecular biology works.

THE REALITY

For decades, drug discovery has been a numbers game: screen millions of compounds, identify thousands with potential, run hundreds through clinical trials, and if you're lucky, one or two reach market.

The timeline: 10-15 years from initial concept to FDA approval. Cost: $2.5-3 billion per drug.

AI is compressing this dramatically.

Here's what we're seeing in our internal R&D:

• Molecular design: AI identifies promising drug candidates 40% faster than traditional screening. We're using AlphaFold to predict protein structures that would have taken us 6-12 months to determine experimentally, and now we get results in days.

• Oncology specificity: Our AI models can predict which compounds will be effective against specific tumor mutations with 78% accuracy on first try, versus 23% with traditional screening. This is reducing candidate failure rates in early stages.

• Clinical trial design: AI is helping us identify patient subpopulations more likely to respond to our drugs, shrinking trial sizes and timelines by 30-50%.

The competitive threat:

Google's DeepMind, Anthropic, and OpenAI are entering drug discovery. So are specialized biotech firms (Exscientia, Relay Therapeutics, Schrodinger). They're publishing results showing AI-designed molecules reaching clinical trials in 18-24 months instead of 5-7 years.

If we don't move fast, we risk being displaced by companies that can iterate drug candidates faster and cheaper than we can.

WHERE WE ARE

Current state: - Oncology pipeline: 47 compounds in development (20 in Phase II/III) - Cardiovascular pipeline: 28 compounds - Respiratory pipeline: 31 compounds - R&D spend: $8.2 billion annually - Time to first-in-human trial (average): 4.8 years - Time to Phase II proof of concept: 7.2 years

Financial performance: - Oncology revenue: $8.4 billion (growing 14% YoY) - Total pharma revenue: $28.6 billion - Gross margin: 74% - FCF: $7.2 billion

The company is healthy. But we're locked in an R&D race with competitors who are weaponizing AI in ways we haven't fully deployed yet.

THE AI OPPORTUNITY

I see three specific ways AI transforms our business model:

Opportunity 1: Accelerate Existing Pipeline

The play: Deploy AI across our existing 106 oncology/cardio/respiratory compounds to find development shortcuts, off-target uses, and patient subpopulations that could shorten timelines.

How: - Partner with or build in-house AI teams (we're already recruiting 60+ ML engineers) - Retrofit existing pipeline compounds through AI-driven molecular analysis - Use AI to identify biomarkers that predict patient response, enabling smaller trials - Target: Compress development timelines by 2-3 years on average

Estimated impact: - 6-8 pipeline compounds could reach market 2-3 years earlier = $3-5 billion in accelerated revenue by 2035 - Reduced trial costs: $150-200 million per compound savings

Timeline: 18-24 months to see meaningful results

Opportunity 2: Open New Oncology Indications

The play: Use AI to design entirely new molecules targeting specific oncology mutations and pathways that are currently "undruggable" or too expensive to develop.

How: - Deploy AI to scan cancer genome databases (TCGA, GDC, COSMIC) and identify mutation-drug matching opportunities - Use generative AI to design novel molecular structures that would be impossible through traditional chemistry - Focus on rare cancers (often 5-year development timelines become 2-3 years with AI guidance) - Build partnerships with academic oncology centers to validate targets

Estimated impact: - 8-12 new oncology programs launched (vs. 2-3 per year historically) - If 20% reach market, that's 1.6-2.4 new cancer drugs - Oncology market expansion: +$2-3 billion annual revenue by 2035

Timeline: 12-18 months to launch first AI-designed programs

Opportunity 3: Build an AI-First Biotech Subsidiary

The play: Spin out or create a subsidiary focused purely on rapid AI-designed drug discovery. Use it as an incubator for next-generation molecules, then integrate winners into core pipeline.

How: - Recruit specialized AI/biotech team (100-150 people) - Partner with AI labs (Anthropic, OpenAI, DeepMind) for exclusive access to latest models - Focus on rare diseases and orphan indications where development is faster and IPR protection is stronger - Operate on venture-capital-like model: fast iteration, high failure tolerance, big winners

Estimated impact: - 3-5 breakthrough compounds per year (vs. 1-2 today) - Creates a technology moat that protects core AstraZeneca - Potential for significant valuation premium (biotech investors pay 2-3x pharma multiples for "innovation engines")

Timeline: 6-12 months to establish; 2-3 years to show results

MY RECOMMENDATION

I'm recommending a combination of all three opportunities, with Opportunity 1 as the immediate focus.

Here's why: Opportunity 1 is low-risk, leverages our existing pipeline, and shows results within 18-24 months. Opportunity 2 expands our market addressable size in oncology. Opportunity 3 positions us for the long-term future where AI-designed drugs are the default.

Together, they allow AstraZeneca to:

• Remain the oncology leader by bringing drugs to market faster than competitors
• Expand into new therapeutic areas that were historically uneconomical
• Build a defensible technology moat through AI/biotech integration

EXECUTION PLAN

Phase 1: AI-Enabled Pipeline Acceleration (2030-2032)

Immediate actions: - Hire 80-100 ML scientists and computational biologists - Partner with 2-3 leading AI labs (Anthropic, DeepMind, academic partners) for exclusive research agreements - Deploy AI across top 30 pipeline compounds to identify acceleration opportunities - Establish "AI Clinical Trial Design" center to optimize trial patient recruitment and biomarker identification

Expected outcomes: - 6-8 compounds reach market 2-3 years earlier - Trial costs reduced 25-30% through smarter patient selection - Time to Phase II PoC reduced from 7.2 to 5.5 years

Phase 2: New Oncology Program Launches (2032-2034)

• Launch 10-12 new AI-designed oncology programs
• Focus on rare and resistant cancers (where market dynamics favor rapid approval)
• Partner with academic oncology centers for validation and patient recruitment

Expected outcomes: - Oncology pipeline expanded from 47 to 60+ programs - Oncology revenue growth accelerates from 14% to 18-20%

Phase 3: Biotech Subsidiary Launch (2031-2033)

• Establish AI-First subsidiary with 150+ person team
• Focus on rapid iteration, high-risk/high-reward drug design
• Build partnerships with academic labs and hospital systems

Expected outcomes: - 3-5 breakthrough programs per year entering preclinical stage - Subsidiary achieves unicorn valuation (potential IPO or acquisition opportunity)

FINANCIAL IMPLICATIONS

By 2035:

• Oncology revenue: $18-22 billion (from $8.4B today, 20%+ CAGR)
• Total pharma revenue: $45-52 billion (from $28.6B today, 12-15% CAGR)
• R&D spend: $11-12 billion annually (10-15% of revenue)
• Gross margin: 72-75% (slight margin compression due to lower-cost AI-designed drugs)
• Annual FCF: $12-15 billion

Stock returns: Valuation multiple expands from 3.5x revenue (current) to 4.5x+ due to accelerated innovation. Stock price target: $240-280 by 2035 (from $128 today).

CLOSING THOUGHT

The pharmaceutical industry is entering a new era. AI is not just a tool for optimizing our existing processes; it's fundamentally changing what's possible in drug discovery.

Companies that embrace this transition—that invest heavily in AI infrastructure and talent, that rethink how they design clinical trials, that build partnerships with AI labs—will be the innovation leaders of the 2030s.

AstraZeneca has the capital, the pipeline, and the scientific foundation to lead this transition. But we need to move decisively and quickly.

Let's execute on this path.

Pascal

Confidential — Board of Directors Only