Viva-Thera nominates HIV drug candidate after seven months using XtalPi's AI

Why it matters

Fu's published MPER research gives Viva-Thera a biological starting point for XtalPi's AI-assisted discovery. The seven-month candidate selection tests that combination as a development model, while efficacy and safety remain to be established.

Qingshan Fu founded Viva-Thera around a part of HIV he had studied for years. On September 22nd, Viva-Thera and XtalPi said that work had produced a preclinical small-molecule drug candidate in seven months, using XtalPi's AI design tools and automated laboratory. The milestone is a candidate selected for further development, not a medicine tested in people. It also gives XtalPi, which incubated Viva-Thera and holds an equity stake, a potential return tied to a drug program rather than platform work alone.

Fu was an author of a 2018 study of the same HIV membrane region while affiliated with Harvard Medical School. That history is central to the September 22nd result: Viva-Thera supplied structural biology and a strategy for targeting the virus; XtalPi supplied computational screening, molecule design and experimental automation. The seven-month clock describes the companies' discovery and optimization work. It does not include Fu's earlier research or any clinical development.

In a company announcement carried on the Financial Times market site, XtalPi said the partners searched a chemical space of approximately 10 million compounds and identified five validated hits before nominating the candidate. Those are company-reported figures. The announcement does not identify the candidate or provide the measurements needed to assess its potency, dosing prospects or safety. It establishes a direction for development, rather than evidence that the drug will work in patients.

The founder's research came first

The target is the membrane-proximal external region, or MPER, of HIV's gp41 protein. It sits close to the viral membrane and helps the virus fuse with a human cell. Its location has made it difficult to study in a form that reflects its surroundings. XtalPi says MPER and the adjacent transmembrane domain retain approximately 90% sequence identity across viral strains, a reason Viva-Thera is pursuing the region for treatments intended to work broadly. Sequence similarity, however, does not by itself establish that this candidate will work against a broad range of strains.

Viva-Thera says it determined an atomic-resolution structure spanning MPER, the transmembrane domain and the cytoplasmic tail, then recreated the membrane setting on nanoparticle surfaces for screening. The scope of that claim matters. Fu and his collaborators published a structure of MPER and the adjacent transmembrane domain in 2018. The newly announced work concerns a region that also includes the cytoplasmic tail; it should not be read as the first structural view of MPER itself. Viva-Thera's bet is that a more complete structural picture, paired with laboratory testing, can guide the design of a useful molecule.

HIV fusion is an established drug target. The FDA label for enfuvirtide, first approved in 2003, identifies it as a gp41 fusion inhibitor given by injection. Viva-Thera is pursuing a small molecule aimed at MPER, a different proposed approach within the fusion machinery. Whether its candidate can offer a practical advantage will depend on properties and results that candidate nomination alone cannot demonstrate.

One platform, several distinct tests

Fu's ambition extends beyond the small molecule. "Our vision is to bring together the interventions needed to prevent HIV and ultimately cure it," he said in XtalPi's announcement. Viva-Thera, established in 2025, also described vaccine, gene-therapy and cell-therapy work. XtalPi says its platform supports design and testing across those programs, giving the shareholder several potential routes for its technology and investment to bear fruit. Each route faces a different scientific test.

For the vaccine program, Viva-Thera reported that its MPER immunogens produced antibody levels in rhesus macaques about five times those of the study comparator. XtalPi also said antisera still inhibited HIV pseudovirus infection after a 1:1,000 dilution. The comparison measures an animal immune response, and the dilution describes a laboratory test of the resulting antisera. Neither result measures protection against HIV infection in people.

The gene-therapy program addresses a separate obstacle: HIV can persist in latent reservoirs even when antiretroviral treatment suppresses the virus in blood. Viva-Thera says it is developing a non-replicating HIV-derived vector to deliver CRISPR-Cas9 tools to CD4+ T cells. It reported vector construction, cell experiments and delivery work in humanized mice, with live-virus studies still ahead. The National Institutes of Health's explanation of latent reservoirs shows why this is a distinct undertaking from blocking viral entry.

XtalPi has described investment and incubation as part of its drug-development model, alongside joint research, asset co-development and licensing. Viva-Thera puts that model in focus: a scientist's long-running work on a difficult target now feeds a platform-backed portfolio in which XtalPi has a stake. For Fu, the immediate achievement is narrower and more concrete. Viva-Thera has selected a molecule to carry forward. The seven-month claim will matter most if the candidate's subsequent testing shows that speed preserved the qualities a viable HIV medicine needs.