All of life encodes information with DNA. While tools for sequencing, synthesis, and editing of genomic code have transformed biological research, intelligently composing new biological systems would also require a deep understanding of the immense complexity encoded by genomes. We introduce Evo 2, a biological foundation model trained on 9.3 trillion DNA base pairs from a highly curated genomic atlas spanning all domains of life. We train Evo 2 with 7B and 40B parameters to have an unprecedented 1 million token context window with single-nucleotide resolution. Evo 2 learns from DNA sequence alone to accurately predict the functional impacts of genetic variation—from noncoding pathogenic mutations to clinically significant BRCA1 variants—without task-specific finetuning. Applying mechanistic interpretability analyses, we reveal that Evo 2 autonomously learns a breadth of biological features, including exon–intron boundaries, transcription factor binding sites, protein structural elements, and prophage genomic regions. Beyond its predictive capabilities, Evo 2 generates mitochondrial, prokaryotic, and eukaryotic sequences at genome scale with greater naturalness and coherence than previous methods. Guiding Evo 2 via inference-time search enables controllable generation of epigenomic structure, for which we demonstrate the first inference-time scaling results in biology. We make Evo 2 fully open, including model parameters, training code, inference code, and the OpenGenome2 dataset, to accelerate the exploration and design of biological complexity.

The hype around Artificial Intelligence (AI) in drug discovery and synthetic biology has never been higher. With over $5 billion invested in AI-driven biotech startups in 2024 alone, the industry is racing to redefine biological design. But are we on the cusp of true transformation, or are there fundamental limits to what AI can achieve? In this panel, leading AI/ML pioneers and biotech innovators will explore the evolving landscape of AI in biological design. From accelerating drug discovery pipelines to engineering novel biomolecules, we’ll discuss the breakthroughs, bottlenecks, and the shifting role of high-throughput experimentation in an AI-first world. What will it take to move from promise to impact? And how can material providers and technology enablers help push the boundaries? Join us for a dynamic conversation on the future of AI in biotech—where it’s headed, and what it will take to get there.

In AI development, there’s a trade-off between scale and sophistication. This session asks whether biological AI should follow the “bigger is better” mantra or focus on smarter, domain-specific architectures. One camp argues that a simple model fed with colossal datasets will outperform a clever model with limited data – echoing the view that more data beats complex algorithms. Others point out that biology’s complexity (from multi-step pathways to 3D genome organization) demands AI with built-in knowledge or special architectures to learn effectively from smaller, high-quality datasets. Through case studies in drug discovery and genomics, we will discuss if success lies in scaling up simple neural networks on big data or in engineering biologically informed AI models that excel with less data. What are the ROI trade-offs for researchers and investors in each approach?