BioNTech SE · NASDAQ:BNTX

BioNTech — Investor Day (October 2025)

129 slides from BioNTech's investor day presentation, Oct 1, 2025

Slide 2: This slide presentation includes forward-looking statements
This slide presentation includes forward-looking statements This presentation contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995, as amended. In some cases, forward-looking statements can be identified by terminology such as “will,” “may,”
Slide 3: Agenda
BioNTech – Building a global immunotherapy powerhouse translating science into survival 14:00 Advancing a disruptive tech-bio company Prof. U. Sahin, M.D. 14:15 Developing the future of AI at BioNTech K. Beguir InstaDeep – Delivering across the full AI stack 14:25 Compute & model scaling A. Laterre
Slide 4: Advancing a disruptive tech-bio company
Advancing a disruptive tech-bio company Ugur Sahin Founder & CEO BioNTech AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 4
Slide 5: Margaret Keenan, Dec. 8, 2020, London Science Museum
Margaret Keenan Dec. 8, 2020 London Science Museum
Slide 6: BioNTech's AI capabilities with worldwide reach
BioNTech’s AI capabilities with worldwide reach BioNTech locations BioNTech AI centers Rwanda Kigali Austria Vienna Germany (HQ + 6 sites) Singapore U.S. Cambridge Gaithersburg China Shanghai Turkey Istanbul Australia Victoria United Kingdom London South Africa Cape Town UAE DubaiTunis Tunisia Franc
Slide 7: BioNTech – disruptive tech-bio company
Multiplatform oncology company Infectious diseases pipeline COVID-19 vaccine global impact Leader in integrated AI capabilities In-house manufacturing 16 >20Clinical programs Ongoing Phase 2 or 3 trials 7 Clinical programs in high unmet need indications 5 Billion doses distributed 4 Platforms includ
Slide 8: Vision: Building a global immunotherapy powerhouse
Vision Building a global immunotherapy powerhouse translating science into survival AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 8
Slide 9: Uniquely positioned to combine approaches to transform cancer care
We are uniquely positioned to combine approaches to transform cancer care ADC = antibody-drug conjugate. Immunomodulators • Focus on the critical IO pathways • Targeting different complementary pathways in cancer immunity cycle may promote a durable anti-tumor effect mRNA cancer immunotherapies • El
Slide 10: Potential for synergy: combining approaches
Potential for synergy mRNA cancer immuno- therapies Targeted therapies We are uniquely positioned to combine approaches to transform cancer care ADC = antibody-drug conjugate. Immunomodulators • Focus on the critical IO pathways • Targeting different complementary pathways in cancer immunity cycle m
Slide 11: Pumitamig's synergistic targeting of PD-L1 and VEGF
Local neutralization of angiogenic and immunosuppressive VEGF-A effects VEGF-A PD-L1 Targeting the TME and blockade of PD-1/PD-L1 signaling Anti-VEGF-A (Fab) Anti-PD-L1 (VHH) VEGF-A dimer in the TME PD-L1 ligand on tumor cells Pumitamig VEGF-A dimer in the TME Anti-VEGF-A (Fab) Anti-PD-L1 (VHH) PD-L
Slide 12: Next-generation bispecific can potentially expand the reach of IO therapy
Anti-PD-L1 Anti-VEGF >1.4 M estimated new cancer cases in the US and EU annually that cannot be addressed by current IO therapies Anti-PD-(L)1 approved Anti-PD-(L)1 not approved US and EU cancer incidence2 1. NCI SEER https://training.seer.cancer.gov/index.html. 2.US incidence source: NIH and Americ
Slide 13: Landmark strategic collaboration with BMS to advance pumitamig
Potential to transform standard of care and establish new IO backbone treatment option for patients with high unmet medical needs Maximizing potential of next-generation immunomodulator pumitamig1 with global co-development and co-commercialization partnership • Bispecific antibody targeting PD-L1 a
Slide 14: Root cause of cancer treatment failure
Interindividual variability & intratumoral heterogeneity Cancer evolution 5-20 years – up to 10.000 mutations Healthy cell Pre-cancer cell Mutations Mutations Mutations Individual patients DNA mutations Root cause of cancer treatment failure AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025
Slide 15: Potential for synergy: combining approaches
Targeted therapies Immunomodulators Potential for synergy We are uniquely positioned to combine approaches to transform cancer care ADC = antibody-drug conjugate. Immunomodulators • Focus on the critical IO pathways • Targeting different complementary pathways in cancer immunity cycle may promote a
Slide 16: Leveraging mRNA leadership with two approaches: iNeST and FixVac
1 6 Individual patient samples (blood and tissue) Artificial intelligence- driven neoantigen prediction On-demand tailored RNA manufacturing Individualized immuno- therapy Mapping of mutations Fixed combination of shared tumor antigens2 Multi-antigen approach tailored to each indication Neoantigens
Slide 17: iNeST: autogene cevumeran driving continuous innovation with data
Individual patient samples Mapping of mutations Neoantigen prediction On-demand tailored mRNA manufacturing Individualized immunotherapy NORMAL G G G AAACT T T T T CC TUMOR G G G AAACG T T T T CC 1 2 3 4 5 Just-in-time manufacturing Dedicated mRNA GMP production facilities Selection algorithms AI an
Slide 18: Neoantigen prediction
Neoantigen prediction: how do we identify, predict, and characterize neoantigens? Neoantigen rank Gene Mutation Length (aa) Transcript VAF MHC I score MHC II score Coverage in tumor VAF in tumor Coverage in normal tissue VAF in normal tissue 1 SNF8 V183M 27 16.05 0.1 2.16 155 0.33 119 0.00 2 SEMA7A
Slide 19: TCR-tumor antigen interaction is an unsolved computational problem
Defining the complex TCR-tumor antigen interaction is an unsolved computational problem Tumor Cell T Cells HLA Allele Diversity > 30,000 TCR Diversity > 5x108 Peptide Diversity > 100,000 AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 19
Slide 20: Dimensions of cancer heterogeneity
Sahin & Türeci, Science 2018 Tumor Heterogeneity Mutation and neoantigen profile, epigenetics, biology and evolution Immune system HLA restriction and immune SNPs Host and environment Factors include HLA haplotype, microbiome, epigenome, age, antigen exposure, drugs, and comorbidities Tumor Environm
Slide 21: AI-powered bio-engineering for personalized immunotherapies
Personalized immunotherapy AI-powered bio-engineering • iNeST1: Personalized immunotherapy platform levering AI to create therapies unique to each patient’s tumor • 4 ongoing trials • >450 patients treated2 • 18,000 neoantigens selected2 • Computational extension of immunotherapy target space3 • Sem
Slide 22: Complete AI integration and personalized medicine capabilities
Individualized treatment platforms to address inter-individual variability Deep genomics & immunology expertise to analyze patient data Automated in-house manufacturing to serve patients on time and globally AI-infused & digitally-integrated target & drug discovery and development ` Drug classes Int
Slide 23: Developing the future of AI at BioNTech
Karim Beguir Co-Founder & CEO InstaDeep Developing the future of AI at BioNTech AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 23
Slide 24: AI is not a single exponential but a triple exponential
AI is not a single exponential but a triple exponential DATA COMPUTE MODELS AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 24
Slide 25: Moore's Law: hardware compute efficiency doubles every two years
Moore’s Law: efficiency of hardware compute doubles every two years 1. Ray Kurzweil Q&A - The Singularity, Human-Machine Integration & AI | EP #83 - Peter Diamandis & Ray Kurzweil Singularity Q&A - (ref) - March 2024 AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 25
Slide 26: Compute efficiency of AI models doubling every 8 months
While the compute efficiency of AI models is also doubling every 8 months 1. Situational Awareness June The Decade Ahead Leopold Aschenbrenner - (ref) - June 2024 AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 26
Slide 27: Is AI likely to expand even further?
Is AI likely to expand even further? The AI Revolution: The Road to Superintelligence - By Tim Urban -January 22, 2015 AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 27
Slide 28: AI itself is now accelerating Machine Learning
AI itself is now accelerating Machine Learning, which creates a new supercycle AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 28
Slide 29: The Era of Experience and Reinforcement Learning
This supercycle, the Era of Experience, is driven by techniques such as Reinforcement Learning (RL) and Optimization that focus on "learning by doing" with AI agents. Welcome to the Era of Experience David Silver, Richard S. Sutton - (ref) - Apr 2025 Environment New State New Action AI Agent AI Day
Slide 30: InstaDeep is active in Reinforcement Learning research
InstaDeep is active in Reinforcement Learning research Oryx: a Performant and Scalable Algorithm for Many-Agent Coordination in Offline MARL NeurIPS Conference 2025 Breaking the Performance Ceiling in Complex Reinforcement Learning requires Inference Strategies NeurIPS Conference 2025 Oral Memory-En
Slide 31: 6 Nature journal publications in less than 12 months
Nucleotide Transformer Building and Evaluating Robust Foundation Models for Human Genomics Nature Methods 2024 ProtBFN & AbBFN Protein Sequence Modelling with Bayesian Flow Networks Nature Communications 2025 ChatNT A Multi-Modal Conversational Agent for Genomics Nature Machine Intelligence 2025 Mat
Slide 32: ChatNT made the cover of Nature Machine Intelligence
ChatNT, our conversational agent for Genomics, made the cover of Nature Machine Intelligence AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 32
Slide 33: Building across the full stack of AI
Compute & Model Scaling Applications Data Acquisition & Refinement AI Innovation InstaDeep and BioNTech are building across the full stack of AI: AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 33
Slide 34: Compute & Model Scaling
Alex Laterre Head of AI Research InstaDeep Compute & Model Scaling AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 34
Slide 35: Scaling laws drive AI innovation
Gold medal at IMO 2025 achieved1 1. DeepMind achieves gold medal-level performance on the 2025 International Mathematical Olympiad with a general-purpose reasoning LLM! (ref) – 21st of July 2025 2. OpenAI general-purpose reasoning models solved all 12 problems at the 2025 International Collegiate Pr
Slide 36: Compounding intelligence
Compounding intelligence Jensen Keynote that Nvidia GTC 2025 AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 36
Slide 37: A fully integrated AI ecosystem
A fully integrated AI ecosystem InstaDeep's AI supercomputer Purpose-built for AI, delivering full control, visibility, and performance. AIChor orchestration platform A Kubernetes-native AI training platform for seamless scaling and fast experimentation. ML software ecosystem in JAX Software for hig
Slide 38: Kyber, InstaDeep's AI supercomputer
~500 PetaFLOPS of Nvidia H100 GPUs 86,000 CPU Cores 1.2 Tons of Hardware per Rack ▪ Custom rack design engineered in-house ▪ Optimised for AI performance and cost efficiency ▪ Powered 100% by renewable energy ▪ Designed to scale seamlessly with next-generation hardware ▪ Tight hardware–software inte
Slide 39: AIchor orchestration platform
+15,000 experiments / month in 2025 +75% GPU usage Simple GitOps workflow: Commit → Build → Run → Monitor Scalable Kubernetes-native provisioning and auto-scaling across clusters Flexible Modular data plane for multi-cluster and multi-cloud compute AIchor, a complete AI training platform, ready for
Slide 40: ML software ecosystem in JAX
▪ Scale→ from rapid prototyping to large-scale training and deployment ▪ Efficiency→ “better, faster, cheaper" AI workloads that maximise hardware usage ▪ Modularity → interoperable, reliable, and optimised components working together ML software ecosystem in JAX Foundation models Decision-making &
Slide 41: Efficiently train 100B-parameter foundation models
1 | Efficiently train 100B-parameter foundation models Hierarchical model sharding Rack #7 Switch H100 DGX 8x H100 GPUs H100 DGX 8x H100 GPUs Switch Rack #11 Switch H100 DGX 8x H100 GPUs H100 DGX 8x H100 GPUs Switch Rack #14 Switch H100 DGX 8x H100 GPUs H100 DGX 8x H100 GPUs Switch Spine switch #1 S
Slide 42: Efficiently train 100B-parameter foundation models: optimizations
1 | Efficiently train 100B-parameter foundation models Hierarchical model sharding ✓ Intra-node: fully sharded data parallelism (NVLink) ✓ Inter-node: data parallelism (RoCE) ✓ Tensor and sequence parallelism available Code optimizations ✓ CuDNN kernels (e.g. Flash Attention) ✓ Mixed precision with
Slide 43: Scaling molecule screening with Machine Learning Interatomic Potential
Simulating molecular properties at scale is key to many industries, including drug discovery, materials, chemicals. Machine Learning Interatomic Potentials allow quantum accuracy orders of magnitude faster on molecular simulations 2 | Scaling molecule screening with Machine Learning Interatomic Pote
Slide 44: Scaling molecule screening with Machine Learning Interatomic Potential
Scientists Candidates Experiments Looking for any type of new molecule or material Success rate depends on selection quality Classical Force Fields Quantum Chemistry MLIP S pe ed Accuracy 2 | Scaling molecule screening with Machine Learning Interatomic Potential Simulating molecular properties at sc
Slide 45: Up to 5x speed-up in simulation speed
Up to 5x speed-up in simulation speed 160 atoms molecule for 1ns (runtime in min) Re-engineered OS code InstaDeep OS MLIP library InstaDeep backends & models Over 100,000 atoms on one GPU Imipenem binding to L,D-transpeptidase Better Cheaper Scalable Quantum Chemistry-level accuracy 2 | Scaling mole
Slide 46: A fully integrated AI ecosystem
A fully integrated AI ecosystem InstaDeep's AI supercomputer Purpose-built for AI, delivering full control, visibility, and performance. AIChor orchestration platform A Kubernetes-native AI training platform for seamless scaling and fast experimentation. ML software ecosystem in JAX Software for hig
Slide 48: Generative AI for genomics
AI innovation Generative AI for genomics Bernardo Almeida Senior Research Scientist InstaDeep AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 48
Slide 49: Nucleotide Transformer: one of the most popular genomics AI models
+1 Million Downloads Across model sizes1 +500 Citations Across model types2 1. Hugging Face Statistics. 2. Google Scholar. Nucleotide Transformer: one of the most popular genomics AI models on Hugging Face AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 49
Slide 50: Generative AI for genomics – publications in top-tier journals
Generative AI for genomics – publications in top-tier journals Nucleotide Transformer Building and Evaluating Robust Foundation Models for Human Genomics Nature Methods, 2024 SegmentNT Segmenting the Genome at Single- Nucleotide Resolution with DNA Foundational Models Nature Methods, 2025 Isoformer
Slide 51: Building a best-in-class model for genomics
Exploiting the data available with the aim of building a best-in-class model for genomics Learn from genomes OR Learn from functional data Nucleotide Transformer v3 NTv3 NT Evo Borzoi AlphaGenome Pre-training on genomes from >150,000 species Post-training on >17,000 functional tracks across 16 speci
Slide 52: Introducing NTv3
Introducing NTv3: a new, truly foundational, model for genomics with a million nucleotide context Multi-species more than 150,000 species genomes Multimodal genomes + functional tracks + genome annotation Multi-domains human genomics, plants genomics, metagenomics Long-range up to 1 million input nu
Slide 53: Pre-training: learning from 150,000+ species genomes
Pre-training | Learning from +150,000 species genomes AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 53
Slide 54: NTv3 learns through Masked Language Modelling
NTv3 learns through Masked Language Modelling Pre-training | Learning from +150,000 species genomes AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 54
Slide 55: Pre-training: Scaling Laws in Action
Pre-training | Scaling Laws in Action AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 55
Slide 56: NTv3, the fastest genomic foundation models
Fine-tuning | NTv3, The Fastest Genomic Foundation Models NTv3 scales up to 1 million nucleotides an order of magnitude more efficiently than competitive models *RC equivariant AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 56
Slide 57: NTv3, the fastest genomic foundation models
NTv3 scales up to 1 million nucleotides an order of magnitude more efficiently than competitive models *RC equivariant Fine-tuning | NTv3, The Fastest Genomic Foundation Models AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 57
Slide 58: NTv3 is amongst the best models for fine-tuning
Fine-tuning | NTv3 is amongst the best models for fine-tuning on downstream tasks Evaluation of different foundation models on 44 long-range downstream tasks, including gene expression, DNA accessibility and genome annotation across various human tissues. AI Day © 2025 BioNTech SE & InstaDeep Ltd. I
Slide 59: NTv3 fine-tuning performance on downstream tasks
Fine-tuning | NTv3 is amongst the best models for fine-tuning on downstream tasks Average performance across quantitative tasks Evaluation of different foundation models on 44 long-range downstream tasks, including gene expression, DNA accessibility and genome annotation across various human tissues
Slide 60: Post-training: learning from 17k+ genomic tracks
Post-training | Learning from +17k genomic tracks and genome annotation Predictions for 1mb input sequence at single-nucleotide resolution AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 60
Slide 61: NTv3 accurately predicts genomic tracks
Post-training | NTv3 accurately predicts genomic tracks at single-nucleotide resolution Predictions for 1 million nucleotide genomic region Example of NTv3 predictions for experiments in human K562 leukemia cells AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 61
Slide 62: NTv3 outperforms Borzoi at track prediction
Post-training | NTv3 accurately predicts genomic tracks at single-nucleotide resolution *Linder et al., Nature Genetics 2025 NTv3 outperforms state-of-the-art model (Borzoi*) at experimental track prediction. AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 62
Slide 63: NTv3 accurately annotates genomes
Post-training | NTv3 accurately annotates genomes at single-nucleotide resolution Predictions for 1 million nucleotide genomic region AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 63
Slide 64: NTv3 genome annotation, 200 thousand nucleotide region
Post-training | NTv3 accurately annotates genomes at single-nucleotide resolution Predictions for 200 thousand nucleotide gene region AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 64
Slide 65: NTv3 outperforms SegmentNT at gene finding
*de Almeida et al., Nature Methods 2025 Post-training | NTv3 accurately annotates genomes at single-nucleotide resolution NTv3 outperforms state-of-the-art model (SegmentNT*) at gene finding, regulatory element detection and splicing. AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 65
Slide 66: Predictive or generative: masked discrete diffusion framework
Exploiting the data available with the aim of building a best-in-class model for genomics Predictive OR Generative Nucleotide Transformer v3 NTv3 NT Enformer Evo Native predictions and can be finetuned De-novo and conditional sequence generation Thanks to the masked discrete diffusion framework, NTv
Slide 67: Designing regulatory enhancer sequences with NTv3
Generation | Designing regulatory enhancer sequences with NTv3 In collaboration with Alex Stark Experiment Design promoter-specific enhancers, at different levels of activity, in Drosophila cell line. Motivation Enhancers are sequence elements that modulate the expression of genes and can be used fo
Slide 68: NTv3 in-vitro performance for activity-specific design
Generation | NTv3 designs have in-vitro state-of-the-art performance for activity-specific design Experimental validation of enhancers with different strengths (RpS12 promoter) NTv3 successfully generated de novo enhancers matching the prompted activity levels. AI Day © 2025 BioNTech SE & InstaDeep
Slide 69: NTv3 in-vitro performance for promoter-specific design
Generation | NTv3 designs have in-vitro state-of-the-art performance for promoter-specific design Experimental validation of enhancers with promoter-specific activities *de Almeida et al., Nature Genetics 2023 NTv3 successfully generated de novo promoter-specific enhancers, achieving fold-change spe
Slide 70: NTv3: a new generation foundational model for genomics
NTv3: a new generation foundational model for genomics applications Genomic tracks (e.g. gene expression) Genome annotation (e.g. genes, splice sites finding) Variant prediction (e.g. eQTL, deleterious mutation) Sequence generation (e.g. enhancer design) NTv3 DNA sequence A C T A T C T A G A T A GT
Slide 71: Generative AI for protein and antibody engineering
AI Innovation Generative AI for protein and antibody engineering Bora Guloglu Senior Research Scientist AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 71
Slide 72: Our vision: one model, many tasks
AGL… AI (BFN-X) Our goal is to model as much of the data as possible ▪ Superior performance by learning a joint distribution across multiple data types and sources. ▪ Unparalleled flexibility in the hands of scientists with task-specific inference. Our vision: one model, many tasks AI AI Day © 2025
Slide 73: Laying the Groundwork
AGL… AI (BFN-X) Laying the Groundwork 1. Atkinson, T., Barrett, T.D., Cameron, S. et al. Protein sequence modelling with Bayesian flow networks. Nat Commun 16, 3197 (2025) AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 74
Slide 75: Why design Fvs?
Why design Fvs? Therapeutic antibodies have diversified in formats across the years. The Fv region is the common key recognition component in all modalities. AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 76
Slide 76: Antibody sequence design challenge
AbBFN2 CDR-H1 CDR-H2 CDR-H3 VH: EVQLLESGGGLVQPGGSLRLSCAASGFTFSSYAMSWVRQAPGKGLEWVSAISWNSGSIYADSVKGRFTISRDNSKNTLYLQMNSLRAEDTAVYYCARGWSQVDTAMDLDYGQGTLVTVSS D geneV gene J gene CDR-L1 CDR-L2 CDR-L3 VL: DIQMTQSPSSVSASVGDRVTITCRASQSVSSNLAWYQQKPGKAPKLLIYGASSLQSGVPSRFSGSGSGTDFTLTISSLQPEDFATYYCQQYNNWLTFGQGTR
Slide 77: AbBFN2: included data modalities
AbBFN2: included data modalities AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 78
Slide 78: AbBFN2 achieves SOTA results on sequence labelling tasks
AbBFN2 achieves SOTA results on 23/23 sequence labelling tasks1, demonstrating robust learning of the genetic and biophysical attributes of antibody sequences. Sequence annotation • Sequence labelling is a prerequisite for steered generation and design. • AbBFN2 is a one-stop labelling tool, • Simpl
Slide 79: Stabilisation of existing antibodies
Using an unstable starting candidate, AbBFN2 is able to refine the interfaces and the total antibody to increase stability1, which allows more stable pairing, better storage, and higher expression levels. Stabilisation of existing antibodies 1. Interface energies are calculated using the Rosetta Pro
Slide 80: Multi-objective design using AbBFN2
• AbBFN2 optimises sequences with multiple conditions1, using inference time compute scaling to generate diverse candidates for early discovery and optimisation. Multi-objective design using AbBFN2 >80%: Success rate (overall) >90%: Success rate (tractable candidates) 46.6: Number of mutations (1 ob
Slide 81: Experimental validation
Experimental validation Traditional humanization is a trial-and-error bottleneck: it often relies on arbitrary back-mutations, is time- and cost-intensive and risks disrupting binding through extensive changes. 1. Based on Marks et al, Bioinformatics, 2021; Tennenhouse et al, Nat. Biomed. Eng, 2024
Slide 83: Experimental validation: humanising antibodies
Experimental validation Objective Humanise antibodies to reduce side-effects, starting from the precursor sequence1 to generate designs Results Our sequences achieved similar expression with an average fewer edits compared to their manually designed clinical-stage counterparts. 1. For each precursor
Slide 84: AbBFN2: Cutting-edge in silico antibody design
AbBFN2: Cutting-edge in silico antibody design • Training on both sequence and associated metadata of interest produces a rich syntax for “prompt/task engineering”. • The “condition anywhere, generate anywhere” paradigm of AbBFN2 admits a wide variety of tasks that can be decided at inference time.
Slide 85: Data acquisition & refinement
Data acquisition & refinement Nicolas Lopez Carranza Head of BioNTechAI InstaDeep Youssef Ben Dhieb Senior ML Engineer InstaDeep AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 86
Slide 86: BioNTech AI strategy is driven by data
driven by data Potential for continued improvement as more data are generated and analysed BioNTech AI strategy is AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 87
Slide 87: We aim to learn as much as possible from the tumour
We aim to learn as much as possible from the tumour AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 88
Slide 88: InstaNovo: Sequence Space and Image Space
Sequence Space Image Space InternalInstaNovo AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 89
Slide 89: The Dark Proteome
Internal AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 90 The Dark Proteome encompasses uncharacterized proteins and hidden translation products beyond canonical proteins and known PTMs
Slide 90: InstaNovo de novo peptide sequencing to explore the dark proteome
InstaNovo technology enables de novo peptide sequencing to explore the 'dark proteome' and uncover unknown proteins in cancer. Internal AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 91
Slide 91: InstaNovo's library-free approach discovers dark proteome antigens
Database searchReference library JETs lncRNAs Known Dark Tumor Antigens (dTAs) TMDPKTGYQ TAYEPTTFW ATGPVMPTR MS2 spectra TMDPKTGYQ GARVEMEYR ATGPVMPTR All Possible Peptides Incl. any unanticipated targets TAYEPTTFW SWHADEQV IGEYKTSLS … InstaNovo’s library-free approach allows discovery of unanticipa
Slide 92: InstaNovo and InstaNovo+ outperform SOTA methods
InstaNovo (auto-regressive) and InstaNovo+ (diffusion) combine to outperform SOTA methods. Has already shown potential in detecting tumour specific epitopes from undocumented ORFs and aberrant splicing. Published in Nature Machine Intelligence Covered by Science Magazine Eloff, K., Kalogeropoulos, K
Slide 93: Introducing InstaNovo V2
Introducing InstaNovo V2 Larger Dataset 63 million labelled spectra Faster Prediction Up to 50x faster inference Higher Accuracy 10–15% increase in peptide recovery More Identifications Up to 2× the number of identifications "Introducing the next generation of InstaNovo models", https://instanovo.ai
Slide 94: AI-assisted tissue annotation tool
AI-Assisted tissue annotation tool (last year) Increased the efficiency of pathologists fivefold (5x) compared to manual annotation. AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 95
Slide 95: 5× faster pathologists — but still not enough
5× faster pathologists — but still not enough AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 96
Slide 96: Thousands of non-annotated whole slide images
97 I AI Day © 2025 InstaDeep Ltd. & BioNTech SE I October, 2025 Thousands of non-annotated whole slide images AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 97
Slide 97: How can we reduce pathologists' annotation efforts?
How can we reduce the pathologists’ annotation efforts while ensuring optimal model performance? AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 98
Slide 98: Random data points selection
Random data points selection AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 99
Slide 99: Random Data Points Selection
Random Data Points Selection AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 100
Slide 100: Use Foundation Models to Cluster the Data by Patterns
Use Foundation Models to Cluster the Data by Patterns AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 101
Slide 101: Use Foundation Models to Cluster the Data by Patterns
Use Foundation Models to Cluster the Data by Patterns AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 102
Slide 102: Use Foundation Models to Cluster the Data by Patterns
Use Foundation Models to Cluster the Data by Patterns AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 103
Slide 103: Use foundation models + smart data points selection
Use foundation models + smart data points selection AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 104
Slide 104: Tool to explore histology data at scale
We developed a tool to explore, understand, and work with our histology data at scale. AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 105
Slide 108: Nanoparticle design
Nanoparticle design Lexi Walls Senior Scientist II BioNTech Cheng Zhang Research Engineer InstaDeep AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 109
Slide 109: High valency nanoparticle vaccines yield strong antibody responses
High valency nanoparticle vaccines yield strong antibody responses towards tough infectious disease targets Hepatitis B vaccine1 Human papilloma virus vaccine2 Malaria vaccine3 Goal: Leverage AI to build nanoparticles suited to harness the power of mRNA vaccines Nanoparticle vaccines have a crown of
Slide 110: Goal: mRNA launched nanoparticle vaccines
111 I AI Day © 2025 InstaDeep Ltd. & BioNTech SE I October, 2025 Goal: mRNA launched nanoparticle vaccines mRNA delivery ER/Golgi Nanoparticle assembly Nanoparticle secretion Nanoparticle vaccine Generated by BioRender
Slide 111: Building a toolkit of diverse AI-designed de novo nanoparticles
Building a toolkit of diverse AI-designed de novo nanoparticles Internal AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 112
Slide 112: Building a nanoparticle piece by piece
Building a nanoparticle piece by piece Internal AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 113
Slide 113: Shape generation
Shape generation Generate thousands of de novo trimer shapes to enhance diversity of building blocks Utilizing AI protein design to build the nanoparticle components Internal AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 114
Slide 114: Symmetric assembling
Symmetric assembling Assembling the nanoparticle building blocks into desired shapes Assembled to thousands of symmetric shapes Internal AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 115
Slide 115: Sequence design
Sequence design Generate hundreds of thousands sequences to match the desired shapes and assemblies N L G V T F K W S … V D E V T A T Q T H Hundreds of sequences per particle S P R H T L A L R … A T M K E S V A E Designing amino acid sequences to form the protein shape Internal AI Day © 2025 BioNTec
Slide 116: DeepChain Folding Studio
DeepChain Folding Studio Filter and enrich to tens - hundreds of high-quality designs prior to laboratory testing Computationally rank and filter the nanoparticle models Internal AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 117
Slide 117: Confirming nanoparticle design and assembly in vitro
Negative stain electron micrographs confirming nanoparticle assembly Computationally designed nanoparticle model Candidate A Wet-lab experiments Candidate B In vitro: confirming nanoparticle design and assembly Internal AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 118
Slide 118: Nanoparticles can display vaccine antigens
Candidate + antigenCandidate + antigen Computationally designed nanoparticle + antigen model Negative stain electron micrographs confirming nanoparticle displays antigen In vitro: showcasing nanoparticles can display vaccine antigens Wet-lab experiments Internal AI Day © 2025 BioNTech SE & InstaDeep
Slide 119: TCR affinity enhancement
TCR affinity enhancement Antoine Delaunay Senior Research Engineer InstaDeep Michael Rooney Senior Director Comp Biology BioNTech Can you put this in as my head shot? Currently the slide shows John. MY title is senior director, computational biology AI Day © 2025 BioNTech SE & InstaDeep Ltd. I Octob
Slide 120: TCRs can access highly tumor-specific cancer targets
T cell receptors (TCRs) can access highly tumor-specific cancer targets Extracellular targets Modest tumor specificity Intracellular targets High tumor specificity HPV P53 RAS PRAME NY-ESO-1 TROP2CEACAM c-MET MHC TCRAntibody AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 121
Slide 121: T cells can achieve durable responses
T cells can achieve durable responses Checkpoint blockade in NSCLC Brahmer, JCO, 2023. PRAME-directed TCR-T Wermke, Nature Medicine, 2025. AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 122
Slide 122: Affinity enhancement is required to unlock TCR potential
Affinity enhancement is required to unlock the full potential of T cell receptors (TCRs) TCR-T solTCR 10-3 M 10-6 M 10-9 M 10-12 M “micromolar” “nanomolar” “picomolar” Weak binding TCRs Strong binding TCRs Natural TCRs Cell therapy Off-the-shelf biologic AI Day © 2025 BioNTech SE & InstaDeep Ltd. I
Slide 123: Conventional display-based affinity enhancement is labor-intensive
Conventional display-based affinity enhancement is labor- intensive but explores tiny sliver of TCR sequence space CDR1α library CDR2α library CDR3α library CDR1β library CDR2β library CDR3β library Displays Measurement Re pe at Natural TCR Enhanced TCR developability high potency high specificity ~
Slide 124: AI-guided exploration of TCR sequence space
AI-guided exploration of TCR sequence space enables efficient discovery of optimized variants AI-guided rational exploration of sequence space Measurement Re pe at Natural TCR Enhanced TCR ~1032 mutants with <15 mutations developability high potency high specificity AI Day © 2025 BioNTech SE & Insta
Slide 125: Residue environment determines optimal substitutions
Residue environment determines optimal substitutions but varies from TCR to TCR Overall TCR:pMHC docking is similar, but exact CDR loop positions are highly diverse Divergent germline CDR2β-MHC interactions in two structures that share both V-genes and MHC allele (PDB ID: 5nht, 6vm9) Twelve TCRpMHC
Slide 126: Our model outperforms state-of-the-art in TCR–pMHC structure prediction
0 10 20 30 40 50 Boltz-1¹ Chai-1² Our model For each model, fraction of targets where at least one of 5 generated structures achieved CDR3 RMSD < 2.5Å. Test set contains only unseen targets. Performance benchmark on test targets (CDR3 RMSD < 2.5Å ) 1. Wohlwend et al., Boltz-1: Democratizing Biomolec
Slide 127: TCR affinity improvement of more than 100,000-fold in 3 rounds
solTCR natural TCRs TCR-T measurement re pe at natural TCR enhanced TCR round K d, M variant sampler affinity predictor structure Source: Internal Example of a TCR affinity improvement of more than a 100,000-fold in 3 rounds Our AI pipeline achieves an average 50,000-fold TCR binding enhancement inc
Slide 128: Affinity-enhanced TCRs lead to durable in vivo tumor control
Affinity-enhanced TCRs lead to strong and durable in vivo tumor control in a pre-clinical model pHLA tumor target 1 pHLA tumor target 2 Source: Internal Source: Internal TCR dosing starts TCR dosing ends TCR dosing starts TCR dosing ends AI Day © 2025 BioNTech SE & InstaDeep Ltd. I October, 2025 I 1
Slide 129: Thank you for your time
Thank you for your time Next: Q&A AI Day © 2025 BioNTech SE & InstaDeep Ltd. October 2025

BioNTech — Investor Day (October 2025) on 6ix

BioNTech — Investor Day (October 2025)

Slide 1: AI Day October 1, 2025

Slide 2: This slide presentation includes forward-looking statements

Slide 3: Agenda

Slide 4: Advancing a disruptive tech-bio company

Slide 5: Margaret Keenan, Dec. 8, 2020, London Science Museum

Slide 6: BioNTech's AI capabilities with worldwide reach

Slide 7: BioNTech – disruptive tech-bio company

Slide 8: Vision: Building a global immunotherapy powerhouse

Slide 9: Uniquely positioned to combine approaches to transform cancer care

Slide 10: Potential for synergy: combining approaches

Slide 11: Pumitamig's synergistic targeting of PD-L1 and VEGF

Slide 12: Next-generation bispecific can potentially expand the reach of IO therapy

Slide 13: Landmark strategic collaboration with BMS to advance pumitamig

Slide 14: Root cause of cancer treatment failure

Slide 15: Potential for synergy: combining approaches

Slide 16: Leveraging mRNA leadership with two approaches: iNeST and FixVac

Slide 17: iNeST: autogene cevumeran driving continuous innovation with data

Slide 18: Neoantigen prediction

Slide 19: TCR-tumor antigen interaction is an unsolved computational problem

Slide 20: Dimensions of cancer heterogeneity

Slide 21: AI-powered bio-engineering for personalized immunotherapies

Slide 22: Complete AI integration and personalized medicine capabilities

Slide 23: Developing the future of AI at BioNTech

Slide 24: AI is not a single exponential but a triple exponential

Slide 25: Moore's Law: hardware compute efficiency doubles every two years

Slide 26: Compute efficiency of AI models doubling every 8 months

Slide 27: Is AI likely to expand even further?

Slide 28: AI itself is now accelerating Machine Learning

Slide 29: The Era of Experience and Reinforcement Learning

Slide 30: InstaDeep is active in Reinforcement Learning research

Slide 31: 6 Nature journal publications in less than 12 months

Slide 32: ChatNT made the cover of Nature Machine Intelligence

Slide 33: Building across the full stack of AI

Slide 34: Compute & Model Scaling

Slide 35: Scaling laws drive AI innovation

Slide 36: Compounding intelligence

Slide 37: A fully integrated AI ecosystem

Slide 38: Kyber, InstaDeep's AI supercomputer

Slide 39: AIchor orchestration platform

Slide 40: ML software ecosystem in JAX

Slide 41: Efficiently train 100B-parameter foundation models

Slide 42: Efficiently train 100B-parameter foundation models: optimizations

Slide 43: Scaling molecule screening with Machine Learning Interatomic Potential

Slide 44: Scaling molecule screening with Machine Learning Interatomic Potential

Slide 45: Up to 5x speed-up in simulation speed

Slide 46: A fully integrated AI ecosystem

Slide 47: AI Innovation

Slide 48: Generative AI for genomics

Slide 49: Nucleotide Transformer: one of the most popular genomics AI models

Slide 50: Generative AI for genomics – publications in top-tier journals

Slide 51: Building a best-in-class model for genomics

Slide 52: Introducing NTv3

Slide 53: Pre-training: learning from 150,000+ species genomes

Slide 54: NTv3 learns through Masked Language Modelling

Slide 55: Pre-training: Scaling Laws in Action

Slide 56: NTv3, the fastest genomic foundation models

Slide 57: NTv3, the fastest genomic foundation models

Slide 58: NTv3 is amongst the best models for fine-tuning

Slide 59: NTv3 fine-tuning performance on downstream tasks

Slide 60: Post-training: learning from 17k+ genomic tracks

Slide 61: NTv3 accurately predicts genomic tracks

Slide 62: NTv3 outperforms Borzoi at track prediction

Slide 63: NTv3 accurately annotates genomes

Slide 64: NTv3 genome annotation, 200 thousand nucleotide region

Slide 65: NTv3 outperforms SegmentNT at gene finding

Slide 66: Predictive or generative: masked discrete diffusion framework

Slide 67: Designing regulatory enhancer sequences with NTv3

Slide 68: NTv3 in-vitro performance for activity-specific design

Slide 69: NTv3 in-vitro performance for promoter-specific design

Slide 70: NTv3: a new generation foundational model for genomics

Slide 71: Generative AI for protein and antibody engineering

Slide 72: Our vision: one model, many tasks

Slide 73: Laying the Groundwork

Slide 74: AbBFN2

Slide 75: Why design Fvs?

Slide 76: Antibody sequence design challenge

Slide 77: AbBFN2: included data modalities

Slide 78: AbBFN2 achieves SOTA results on sequence labelling tasks

Slide 79: Stabilisation of existing antibodies