ROCHE HOLDING AG — EXECUTIVE MEMO

"Leading Through Transformation: AI Reshapes Pharma's Future"

To: Global Management Committee From: Office of the Chief Executive Date: June 2030

SITUATION ASSESSMENT

We have successfully executed a strategic transformation that shifts Roche from traditional pharmaceutical company facing patent cliff to AI-enabled precision medicine platform with accelerating pipeline. The transformation is real but fragile. This memo addresses the decisions required to sustain momentum.

WHERE WE STAND

Pipeline Strength: 23 programs in Phase 1-2 (vs. 11-13 historical average). Expected launch cadence reaches 2-3 annually by 2033.

R&D Productivity: Cost per approved drug declining toward CHF 1.2-1.5B (vs. CHF 3B+ in 2025). Development timelines compressed by 40-60%.

Market Reception: Investors have rewarded execution; stock appreciated 68% since 2025 despite broader pharma sector compression. Diagnostics integration increasingly valued by market.

Competitive Position: 2-3 year lead on major competitors in AI drug discovery integration. Patent portfolio strengthening with new AI-enabled indications.

Organizational Adaptation: Successfully hired 2,000+ computational scientists, data engineers, and bioinformaticians. Integration of AI capabilities into traditional R&D culture remains challenging but progressing.

DECISION 1: PIPELINE ACCELERATION VS. QUALITY ASSURANCE

We face a tension: the faster we advance the pipeline, the more we expose ourselves to clinical failures. AI has improved target selection and lead optimization, but has not eliminated pharmaceutical risk.

Current Approach: 23 Phase 1-2 programs advancing at accelerated pace (18-month preclinical timelines vs. 48-month historical average).

Early-Stage Data: Favorable safety and efficacy signals in Phase 1-2 data across program classes. This suggests AI predictions are translating to clinical reality.

The Risk: If we advance too aggressively and encounter Phase 3 failures, we risk: - Investor confidence erosion ("AI promised productivity, delivered failure rate") - Regulatory credibility damage (FDA/EMA skepticism about AI methodologies) - Internal team demoralization (failed launches create organizational exhaustion)

The Opportunity: If we advance confidently and deliver approvals on schedule, we establish Roche as the company that solved pharma's innovation crisis. This justifies premium valuation indefinitely.

Recommendation: Maintain current acceleration pace while implementing enhanced safety monitoring: - Independent safety review boards for all AI-optimized programs - Phase 2b interim analyses to catch efficacy/safety issues before Phase 3 - Contingency planning for rapid program termination if signals emerge - Conservative external communication (underpromise on timelines)

This balances confidence with prudence. We are advancing faster than competitors but not recklessly.

DECISION 2: COMPETITIVE MOAT & OPEN-SOURCE RISK

ASML has a technological monopoly. We do not. Our AI drug discovery advantage rests on: 1. Proprietary datasets (patient outcomes from diagnostics) 2. Internal computational talent 3. Partnership ecosystem access 4. Integration with existing pharma operations

But we operate in a world where: - Large language models are increasingly open-source - Drug discovery AI algorithms are published in peer-reviewed literature - Competitors (Merck, Pfizer) are hiring top AI talent aggressively - Biotech startups are exploring similar approaches

Our 2-3 year lead may not survive until 2035 if competitors execute effectively.

Strategic Choice: How much should we: - Invest in proprietary algorithms vs. leveraging open-source approaches? - Keep data proprietary vs. contribute to open science (regulatory goodwill)? - Hire additional AI talent vs. focus on retaining current team? - Build internal vs. partnering with biotech ecosystem?

Options:

Option A (Closed Ecosystem): Maximize proprietary advantage; restrict data sharing; build all capabilities internally. This protects moat but requires sustained 25%+ R&D growth, limits partnership flexibility, and risks talent attraction as competitors offer more freedom.

Option B (Open Partnership Model): Contribute to open-source drug discovery efforts; share datasets with academic partners; focus internal work on proprietary integration and clinical translation. This expands talent pool, improves regulatory relationships, but reduces proprietary differentiation.

Option C (Hybrid Approach): Maintain proprietary advantage in highest-value areas (oncology, precision diagnostics integration) while contributing to open science in adjacent areas. Focused talent acquisition in core competencies while leveraging external ecosystem for broader capability.

Recommendation: Option C, with emphasis on proprietary advantage in precision medicine and oncology. These are our highest-margin therapeutic areas. We should: - Contribute diagnostic data to open-science initiatives (EMA/FDA collaboration) - Share non-core drug discovery algorithms with academic partners - Maintain proprietary focus on AI+diagnostics integration and precision patient stratification - Target hiring in precision medicine and biomarker-driven drug development - Establish Roche-led consortium for open drug discovery standards (competitive positioning as thought leader)

This preserves moat in high-value areas while building goodwill and talent attraction through open science engagement.

DECISION 3: GEOGRAPHIC EXPANSION & REGULATORY STRATEGY

Our current R&D footprint is predominantly Swiss/European with growing U.S. presence (Genentech, acquired biotech). This creates geographic constraints:

Regulatory Complexity: Different regulatory frameworks (FDA, EMA, NMPA) require different submission strategies. Current approach optimizes for FDA-first then EMA-follow approach.

Talent Access: Top computational biology talent increasingly concentrates in U.S. (Bay Area), China (Beijing, Shanghai), and increasingly Taiwan.

Geopolitical Risk: U.S.-China technology competition could impact our R&D strategy if we maintain presence in both geographies.

Innovation Ecosystem: Silicon Valley, Boston, and increasingly Toronto have deep AI/biotech ecosystem density. Roche maintains presence in these hubs but could expand.

Options:

Option A (Europe-Centric): Double down on Swiss headquarters with selective U.S. and Asian presence. Advantages: cultural coherence, tax efficiency, European regulatory integration. Disadvantages: talent access constraints, innovation ecosystem distance.

Option B (Distributed Model): Establish three-hub approach (Europe for precision diagnostics, U.S. for computational biology and oncology, Asia for later-stage manufacturing and market entry). Advantages: talent access, regulatory flexibility. Disadvantages: organizational complexity.

Option C (Selective Expansion): Maintain Swiss headquarters as innovation hub while expanding U.S. presence (San Francisco, Cambridge) for computational science hiring. Minimize Asia presence due to geopolitical complexity.

Recommendation: Option C. Roche is a Swiss institution with global operations. Our competitive advantage in precision medicine is rooted in diagnostic integration—a capability we've built over decades in Switzerland. Rather than diffusing innovation across geographies, we should: - Expand San Francisco and Cambridge operations to 1,200+ scientists (from current 650+) for computational biology hiring - Maintain Swiss headquarters as precision diagnostics and oncology hub - Establish Asia partnerships rather than owned operations (reduces geopolitical exposure) - Focus on FDA/EMA regulatory pathway; NMPA pathway is lower priority

This preserves organizational coherence while accessing global talent.

DECISION 4: PRICING STRATEGY & PAYER RELATIONSHIPS

AI-optimized drugs have a value proposition we haven't fully capitalized on: superior efficacy, better safety profiles, and precision targeting reduce overall healthcare system costs.

But we operate in an environment of intense pricing pressure: - U.S. government implementing price negotiation authority - European payers demanding cost-effectiveness data - Emerging market pricing constraints - Competition intensifying as older patents expire

The Question: Should we: - Price AI-optimized drugs at premium to existing therapies (justified by superior efficacy)? - Price competitively with existing therapies (build market share through access)? - Develop value-based pricing models (payment linked to patient outcomes)?

Financial Impact:

Scenario A (Premium Pricing - 20% premium on base drugs): - Higher per-unit margins (50%+ gross margins) - Lower market penetration due to payer resistance - Estimated 2035 revenue from new launches: CHF 12-14B

Scenario B (Competitive Pricing - 5% premium): - Higher market penetration (70%+ uptake) - Moderate per-unit margins (40-42%) - Estimated 2035 revenue from new launches: CHF 18-22B

Scenario C (Value-Based Pricing): - Patient outcome and healthcare system cost savings drives pricing - Requires comprehensive health economics evidence - Potentially highest long-term value but complex execution - Estimated 2035 revenue from new launches: CHF 14-17B

Recommendation: Scenario B with strategic premium in oncology. Here's why: - AI-optimized drugs offer genuine value proposition to payers (superior efficacy, safety) - Aggressive premium pricing invites political backlash and regulatory pressure - Competitive pricing builds market share, which generates revenue scale to fund next pipeline - Oncology can command premium due to life-extension value - Non-oncology (autoimmune, metabolic) should price competitively

We should implement value-based pilots in 2-3 indications to build health economics evidence. This positions us favorably with payers long-term.

DECISION 5: M&A STRATEGY & PORTFOLIO RATIONALIZATION

ASML has simplified portfolio focus. Roche has more than 50 therapeutic areas with varying productivity levels.

Current State: - Oncology: 40% of revenue, highest growth, strongest pipeline - Virology (HIV, HCV, Influenza): 18% of revenue, mature products, limited pipeline - Autoimmune/Inflammatory: 15% of revenue, moderate pipeline - Metabolic/Ophthalmology/Other: 27% of revenue, fragmented

The Question: Should we rationalize the portfolio toward highest-growth areas?

Scenarios:

Option A (Full Focus): Divest or partner everything except oncology and precision diagnostics. Concentrates resources on highest-opportunity areas. Risk: loses profitable mature revenue; attracts antitrust scrutiny.

Option B (Selective Rationalization): Divest underperforming assets (certain virology products); maintain core therapeutic areas; invest in AI across all areas. Balanced approach. Moderate resource concentration.

Option C (Portfolio Maintenance): Maintain current breadth; apply AI acceleration across all therapeutic areas. Maximizes revenue; dilutes investment focus.

Recommendation: Option B with focus on maintaining profitable maturity while concentrating growth investment: - Virology: selective partnerships for mature products; maintain HIV/HCV franchises but reduce portfolio breadth - Autoimmune/Inflammatory: integrate with diagnostics for precision patient stratification - Oncology: maximum AI investment concentration - Diagnostics: position as platform enabling precision across all therapeutic areas - Ophthalmology: explore partnership/divestiture for underperforming assets

This generates immediate cash from portfolio rationalization (CHF 3-5B potential divestiture proceeds) while focusing capital on highest-growth opportunities.

DECISION 6: ORGANIZATIONAL STRUCTURE & TALENT STRATEGY

We have successfully hired 2,000+ computational scientists in 2028-2029. Retention and integration are critical.

Challenges: - AI talent compensation competition with tech companies - Cultural integration (AI scientists vs. traditional pharma scientists) - Career path definition (how do computational scientists advance in pharma?) - Knowledge transfer (how do AI predictions get translated by traditional teams?)

Actions Required:

1. Compensation Competitiveness: Roche salaries for senior computational scientists (PhD-level) trail tech companies by 20-30%. We should implement:
2. 10-12% increases for AI/computational talent
3. Equity grant expansion (larger grants, faster vesting for top performers)
4. Signing bonuses for external hires ($200-500K range for senior computational scientists)

5. Expected cost: CHF 400-500M annually in incremental compensation

6. Career Path Definition: Create explicit career tracks:

7. Individual contributor path (computational scientist → principal scientist → distinguished scientist)
8. Management path (lead scientist → senior research director → site director)

9. Business development path (computational scientist → therapeutic area lead → portfolio manager)

10. Organizational Integration: Establish joint teams pairing computational scientists with traditional medicinal chemists and biologists. This accelerates knowledge translation and breaks down silos.

11. Retention Programs:

12. Sign 3-year retention agreements for top 200 computational scientists
13. Establish internal mobility program (enable transfers across therapeutic areas)
14. Create innovation fellowship program (2-year rotations through different therapeutic areas)

DECISION 7: EXTERNAL COMMUNICATION & EXPECTATIONS MANAGEMENT

We have built positive investor sentiment through clear, consistent execution. As the pipeline advances and launches occur, we must manage external expectations carefully.

Risk: Over-promising on AI benefits, then delivering solid-but-not-miraculous results, erodes investor confidence.

Mitigation Strategy: - Conservative external guidance on timelines and efficacy - Early Phase 2 data transparency (build track record) - Clear communication on failure rates (AI improves success but doesn't eliminate pharma risk) - Emphasis on long-term structural advantage rather than near-term sales projections

We should position Roche as "the company that made pharmaceutical R&D efficient" rather than "the company whose AI drugs are revolutionary." The former is defensible; the latter invites disappointment.

STRATEGIC PRIORITIES FOR NEXT 18 MONTHS

1. Advance Phase 2-3 pipeline: Execute on development timelines; publish Phase 2 data to establish credibility
2. Expand computational capacity: Add 500-800 new AI scientists; solidify retention of current team
3. Integrate diagnostics: Link precision diagnostics to new pipeline launches
4. Portfolio optimization: Evaluate virology rationalization; pursue 1-2 strategic partnerships
5. Payer engagement: Develop health economics evidence for premium pricing justification
6. Regulatory relationships: Strengthen FDA/EMA alignment on AI drug development

BOARD DECISIONS REQUIRED

1. Approve CHF 500M incremental compensation budget for AI talent retention and market competitiveness
2. Authorize portfolio rationalization strategy (virology partnerships/divestitures)
3. Approve value-based pricing pilot programs in 2-3 therapeutic areas
4. Commit to Phase 2-3 pipeline advancement timelines
5. Authorize organizational restructuring to integrate computational and traditional pharma scientists

The window for pharmaceutical R&D transformation is narrow. Competitors are moving. If we execute decisively over the next 18-24 months, Roche will cement position as the premier AI-enabled pharma company. If we hesitate, we risk losing first-mover advantage.

This memo is confidential and intended for board discussion only.