Valhalla Tech, a Full-Modality Molecular World Model Company, Raises Nearly $50M in Three Consecutive Months of Funding | 5Y News

Valhalla Tech, a full-modality molecular world model company, announced that it has completed multiple consecutive funding rounds within three months, raising nearly $50 million in total. Investors include 5Y Capital, ZhenFund, CASSTAR, Xiang He Capital, Guofang Venture Capital, Lushi Investment, MiraclePlus, L2F LightSource Capital Founders Fund, and prominent industry partners, with LightSource Capital serving as the exclusive financial advisor.

The proceeds will be used primarily for talent expansion, iterative upgrades to its self-developed full-modality molecular world model, and continued full-modality wet-lab validation — further strengthening the company's full-stack technical foundation toward Scientific AGI.

Valhalla Tech founder Odin was trained under Professor David Baker, recipient of the 2024 Nobel Prize in Chemistry, and emerged from the Baker Lab — widely regarded as the "West Point" of global AI protein design, where the Rosetta platform and the RFDiffusion generative model were born. Breaking from the AIDD industry's long-standing inertia of single-modality R&D, Odin proposed and built a new paradigm of "full-modality molecular design." The core idea starts from the fundamental laws of molecular physics, describing all intermolecular interactions within a unified physical framework to solve the technical challenge of characterizing cross-modal molecular commonalities and differences.

Based on this approach, the company is iteratively developing its next-generation multimodal generative molecular world model, the AlloDesign system, with the goal of enabling de novo design and structural optimization of arbitrary combinations across five molecular modalities — proteins, DNA, RNA, small molecules, and ions — covering the full spectrum of molecular R&D needs.

AlloDesign will embed many-body interaction physics assumptions directly into the training process, enhancing the model's capacity to model and design multimodal, multivariate interactions. This addresses the limitations of traditional single-modality models in accurately simulating ternary and higher-order molecular interactions, enabling the discovery of "undruggable" targets that conventional techniques cannot access. Complementing this, the company has built a self-developed high-throughput experimental platform that connects molecular design, automated synthesis, and affinity validation into a complete pipeline, enabling rapid construction, experimental characterization, and iterative optimization of candidate molecules. Experimental results will continuously feed back into the model and design strategies, driving deep coupling between computational prediction and experimental measurement to gradually form a data-driven, closed-loop R&D system.

The AlloDesign system has already made early progress in cyclic peptide design. Through multimodal hybrid training, stochastic atomization, and topological conditional control, the model has achieved leading performance on benchmark tasks involving linear peptides and head-to-tail amide bond cyclic peptides, and has further extended to de novo design of complex topologies including disulfide cyclization, isopeptide bond cyclization, and bridged rings. These advances demonstrate AlloDesign's emerging unified modeling capability across different cyclization modes and chemical connection types, laying groundwork for exploring complex cyclic peptide chemical spaces that traditional methods struggle to cover.

On the commercial front, Valhalla Tech will initially target drug discovery as its first application scenario, offering two services to global MNCs and biotechs: custom target molecule R&D and industrial-grade model licensing, with the aim of shortening early-stage drug development cycles. In the medium to long term, the company plans to build a self-iterating Scientific AGI infrastructure, extending its technical capabilities to additional scientific fields and deploying self-driving intelligent laboratories to construct virtual scientist systems capable of independent scientific reasoning and experimental iteration.

Odin, founder of Valhalla Tech, stated: "We've always believed that the overall technical roadmap for AGI in life science must evolve from single-modality to multimodal, from single-task to multi-task. For molecular design, a full-modality model isn't simply putting different molecule types into one model — it's about building from underlying physical laws to form a unified technical system that transcends molecular modalities and bridges computation with experiment. After this round, we'll continue investing in foundational models, high-throughput experimental platforms, and top-tier talent to accelerate AlloDesign's path from frontier research to real-world R&D scenarios."

5Y Capital commented: "Life science is undergoing a systemic paradigm shift spanning physical infrastructure, scientific data, and AI model layers, requiring a new generation of teams capable of end-to-end engagement with fundamental challenges. The Valhalla team is precisely such a group — entrepreneurial, highly execution-oriented, and exceptionally creative. 5Y believes they have the capacity to become leaders in this transformation."

About Valhalla Tech

Valhalla Tech is building the next-generation microscopic molecular world model, AlloDesign. The model aims to unify understanding and design of proteins, nucleic acids, small molecules, and complex molecular assemblies, continuously evolving through automated synthesis and wet-lab validation with real experimental feedback. The company targets AI-driven drug R&D as its first application, with the long-term goal of creating Scientific AGI infrastructure capable of autonomous reasoning, experimentation, and evolution to empower future scientific discovery.