CHARM Therapeutics Raises $80M to Take Its AI-Designed Leukaemia Drug to the Clinic

September 2, 2025

CHARM Therapeutics raised $80 million in an oversubscribed Series B financing in September 2025, co-led by New Enterprise Associates (NEA) and SR One, with participation from existing investors OrbiMed, F-Prime Capital, Khosla Ventures, and NVIDIA. The round brings CHARM's total capital raised to over $150 million. The funding will be used to advance CHARM's lead menin inhibitor candidate into clinical development, with a Phase 1 trial in acute myeloid leukaemia (AML) targeted for early 2026.

Acute myeloid leukaemia is an aggressive blood cancer with limited treatment options and poor long-term outcomes for many patients. The recent approval of first-generation menin inhibitors — drugs that work by blocking the interaction between the menin protein and oncogenic transcription factors — marked a genuine clinical advance. However, their efficacy is undermined by the rapid emergence of resistance mutations in the menin gene itself, which allow cancer cells to escape treatment. For many patients, the clinical benefit of these approved drugs proves short-lived. This resistance problem is precisely where CHARM has focused: rather than optimising existing scaffolds, it has used AI to design molecules that retain potency against all publicly described clinical resistance mutations from the outset.

CHARM's core technology is DragonFold, a proprietary protein-ligand co-folding platform that predicts how drug candidates interact with protein targets — including mutant variants — with a precision that traditional structure-based drug discovery cannot match at speed. DragonFold enabled CHARM to identify a development candidate that demonstrates robust tumour regression in preclinical models, is predicted to be efficacious at low human doses without causing QTc prolongation, and does not inhibit the CYP enzymes responsible for common drug-drug interactions. Co-founder David Baker — a 2024 Nobel laureate in chemistry recognised for his work on protein structure prediction — is one of the scientific architects of this approach, lending CHARM exceptional credibility in a competitive AI drug discovery landscape.

The proceeds will fund the Phase 1 clinical trial, which is expected to enrol patients in the first quarter of 2026, along with further development of CHARM's broader DragonFold platform. The company has simultaneously strengthened its oncology leadership: Briggs Morrison, former CEO of Syndax (which developed the first FDA-approved menin inhibitor, revumenib), and Kim Blackwell join the board as non-executive directors, bringing direct menin inhibitor development expertise at a pivotal juncture.

NEA and SR One's co-leadership of this round reflects the maturation of AI-driven drug discovery as an institutional asset class. NVIDIA's continued participation — through its NVentures arm — is also instructive: it reflects both the compute intensity of frontier protein-ligand co-folding models and the strategic interest GPU manufacturers have in proving out clinical outcomes in biotech. At over $150 million raised without yet entering clinical trials, CHARM is making a significant bet that the quality of its DragonFold-derived science will translate from preclinical models to patients — a transition that remains the central challenge for the entire AI drug discovery sector.

Sources