Touchstone backs Vietnam’s N2TP to build AI infrastructure for scientific discovery

Vietnam’s startup ecosystem has spent the past few years proving it can produce consumer apps, fintech platforms and edutech companies at regional scale. N2TP is attempting something less common, and arguably harder: building AI infrastructure for scientific research, where outputs are not measured in clicks or transactions, but in hypotheses, experiments, papers and patents.

The Hanoi-based company has raised seed funding from Touchstone Partners, the Vietnam-focused VC firm known for backing AI and deeptech companies such as Alpha Asimov, Eureka Robotics, Forte and Prep. The size of the round was not disclosed.

Also Read: AI infrastructure: The unsung hero of technological innovation

Founded in 2020, N2TP describes itself as an “AI Lab” focused on scientific research and intellectual property development in fields including AI infrastructure, biomedicine and biotechnology. Its core product is the N2TP AI4Science Platform, which aims to help research teams move beyond using AI merely to speed up isolated tasks and towards using it as part of the scientific process itself.

That distinction matters. In science, an AI-generated answer is not a discovery. A model may suggest a new molecule, pathway or biological relationship, but researchers still need to test whether the idea is logically sound, grounded in domain knowledge, experimentally feasible and reproducible. N2TP’s platform is designed to sit inside that loop: generating and assessing hypotheses, supporting simulation, helping design experiments, collecting new data and feeding results back into the system.

For Southeast Asia, where many research institutions and startups operate with tighter budgets than their counterparts in the US, Europe or China, this kind of infrastructure could be significant if it works at scale. The region has strong scientific talent but often lacks the same depth of capital, automated lab infrastructure and commercialisation pathways. AI-for-science tools could help narrow that gap, though they will not remove the need for serious laboratory validation.

From research bottleneck to repeatable loop

N2TP was founded by CEO Duong Thi Hong Nhung, a doctoral candidate at Hanoi University of Pharmacy who holds a master’s degree in pharmaceutical biochemistry; Chief Product Officer Do Ngoc Tuan, a computer science graduate of Goldsmiths, University of London; and Head of Research Operations Ho Hai Phong, who studied engineering at Kyushu University and finance at Waseda University in Japan.

The mix of pharmaceutical science, computer science and research operations reflects the problem N2TP is trying to solve. Scientific discovery is not slowed down only by a lack of ideas. It is slowed down by the work needed to turn an idea into something testable, then into evidence, then into a product, paper or patent.

N2TP says its AI4Science platform can help narrow the search space early by eliminating options that do not meet scientific or operational constraints. After a hypothesis has been validated through simulation, the system can support the next steps: experimental design, measurement and data collection. Those results can then be used to update the model and guide the next round of work.

In practical terms, this is less about replacing scientists than about building a tighter feedback loop between computation and experimentation. That is especially relevant in areas such as drug discovery and biotechnology, where teams may need to evaluate huge numbers of possible compounds, biological targets or experimental conditions before arriving at a viable path.

“We do not see AI as a tool to replace scientists. Scientists are still the ones who ask the questions, set the standards, oversee the process, interpret the results and bear responsibility for important decisions,” Nhung said. “What N2TP aims to build is infrastructure that more tightly connects hypothesis, reasoning, simulation and experimentation.”

Early output, but commercial questions remain

The company claims its platform has already improved research productivity. Over the past 12 months, N2TP says it has had 12 research papers accepted, presented or published at major global research conferences and forums, including ICML 2026, ACL 2026, UAI 2026, SIGMETRICS 2026, AAMAS 2026 and ISMB/ECCB 2025. It has also published three papers in Q1 journals: Scientific Reports, CPT: Pharmacometrics & Systems Pharmacology and Computers in Biology and Medicine.

Also Read: AI is eating the world and startups are riding the infrastructure wave

In addition, N2TP has filed seven patent applications in Vietnam and internationally across foundational AI, biomedicine and biotechnology. The company says this pace is at least four times faster than its own output under a traditional research model.

Those numbers are useful markers, but they are not the whole story. In deeptech, publications and patents show capability, but commercial value depends on whether the underlying technology can be turned into defensible products, licensing revenue, partnerships or internal drug and biotech pipelines. Many AI-for-science companies globally have found that strong models are only one part of the equation; access to high-quality data, wet-lab validation and regulatory pathways can be just as decisive.

N2TP plans to use the new funding for two main areas: developing its patent portfolio in strategic technologies and completing the research loop through deeper integration with automated laboratory processes and equipment. The latter will be important. AI systems become more useful in science when they can learn from experimental results quickly and repeatedly, rather than relying only on existing datasets.

A crowded global field, a quieter regional one

N2TP is entering a global market that has attracted serious capital and talent. Google DeepMind’s AlphaFold changed expectations for AI in biology by predicting protein structures at scale, while Isomorphic Labs is applying similar capabilities to drug discovery. US-listed Recursion uses machine learning and large-scale biological datasets to build drug pipelines, while Hong Kong-founded Insilico Medicine has become one of Asia’s most visible AI drug discovery companies.

Compared with these players, N2TP is at an earlier stage and is building from Vietnam, where deeptech capital is still developing. Its advantage, if it can sustain one, may come from a focused team, lower R&D costs and the ability to build intellectual property around specific scientific workflows rather than compete head-on with global giants across every part of the AI biology stack.

Within Southeast Asia, the field remains comparatively thin. The region has produced healthtech, diagnostics and biotech startups, but fewer companies are building foundational AI infrastructure for scientific discovery. That gives N2TP room to define a category locally, but also means it may need to look beyond Vietnam early for partners, customers and validation.

Why Touchstone is betting on harder tech

For Touchstone Partners, the investment fits a broader push into Vietnam’s deeptech sector. Since launching in 2021, the firm has backed companies across AI, robotics, education, agriculture, healthcare and climate technology, including through initiatives such as the Net Zero Challenge.

The N2TP deal also reflects a growing belief among some Vietnamese investors that the country should build more than application-layer startups. While software products can scale quickly, foundational intellectual property in AI, semiconductors, biotech and advanced manufacturing is increasingly seen as important for national competitiveness.

“N2TP shows that Vietnamese researchers are fully capable of building core technology that meets international standards, even in as demanding a field as biomedicine,” said Ngo Thuy Ngoc Tu, Director of Touchstone Partners. “We believe that AI infrastructure for scientific research will be a critical piece of Vietnam’s technological development in the years ahead.”

Also Read: Razer and NUS launch Singapore AI lab to rethink how games respond to players

The hard part starts now. N2TP has early research output, a technical thesis and new venture backing. To become more than a promising lab, it will need to show that its platform can produce repeatable scientific and commercial outcomes.

For Vietnam’s startup ecosystem, the company’s progress will be watched not only as a funding story, but as a test of whether the country can build deep technology companies whose value lies in original research and defensible IP. That is a slower path than most startup playbooks allow, but it may be the one that matters most.

The post Touchstone backs Vietnam’s N2TP to build AI infrastructure for scientific discovery appeared first on e27.