Part 4 of 5 — How can a country build its own 1000 Genomes Project? Vietnam's journey from 2018 to 2026
In Part 3, we covered variant calling, cohort analytics, and the storage architecture. This part steps back from the technical stack to look at the bigger picture: Vietnam's journey from 2018, when VN1K was initiated, to 2026. Three things changed along the way — infrastructure, human resources, and government support — and together they determine whether a national 1000-genome project is feasible today.
Series Overview
- Part 1: Vision and end-to-end architecture
- Part 2: Design principles and implementation roadmap
- Part 3: Variant calling, cohort analytics, and data organization
- Part 4 (This Post): Infrastructure, talent, and government support: 2018 → 2026
- Part 5: Summary, extensions, and the questions ahead
1. Vietnam's journey: 2018 → 2026
VN1K in brief. Initiated in 2018 at the Bigdata Research Institute (now part of VinUniversity), VN1K is the first large-scale multi-omics and phenomics resource for the Vietnamese population: 1,011 individuals from 53 provinces and cities, more than 42 million genetic variants — about 8.5 million of them never recorded in major international databases — and the Vietnamese PanGenome Reference (VPR), the first population-specific reference genome for Vietnam. The work was published in Nature Communications in 2026 (VinUni announcement).
1.1. The starting point: VN1K in 2018
Eight years is a long time in genomics. When VN1K started in 2018, Vietnam had no population-scale genome reference of its own: genetic studies of Vietnamese people leaned on reference data built mostly from other ancestries, which is exactly the gap the project set out to close.
The environment around the project was just as challenging. Large-scale sequencing capacity was limited and often relied on partners abroad. Compute and storage for population-scale analysis were scarce — running joint genotyping across a thousand genomes, the way Part 1 describes, was out of reach for most teams. The number of people who could run a population-scale pipeline end to end could fit in a small room. And dedicated policy for genomic data and precision medicine was still at an early stage.
VN1K went ahead anyway. Between 2018 and publication, the team collected samples from 1,011 individuals across 53 provinces and cities and generated high-depth short-read whole-genome sequencing plus microarray, long-read whole-genome sequencing, RNA sequencing, and the first whole-genome methylation profile based on long reads. The VN1K data portal now exposes the resource to researchers.
1.2. Infrastructure: from scarce to available
In 2018, compute was the bottleneck. Most labs worked on small clusters or individual servers, data lived on local disks, and population-scale joint genotyping or cohort analytics required infrastructure that simply was not there.
The data center layer tells the same story in numbers:
| Data centers | 2018 | 2026 |
|---|---|---|
| Market | Grew ~12.7%/year from 2016, reaching ~US$858M in 2020 | Estimates vary by methodology: ~US$1.0–2.1B in 2025; capacity projected at ~950 MW by 2030 |
| Operators | Telco-grade facilities from Viettel, VNPT, FPT, and CMC | Viettel IDC, VNPT, FPT Telecom, and CMC Telecom lead by capacity; Hanoi Telecom and VNG operate as well; foreign players entering |
| Flagship projects | Viettel Hòa Lạc campus: ~800 racks when it opened (mid-2010s) | Viettel Hòa Lạc: 21,000 m², 30 MW, 2,400 racks; Viettel's 140 MW hyperscale campus under construction in Ho Chi Minh City; FPT AI Factory (NVIDIA GPU cloud) |
| National infrastructure | None | National Data Center No. 1 in Hòa Lạc; the Data Law (2024); data-localization rules |
| Pipeline | — | ~5.15M sq ft of upcoming white floor vs ~0.95M sq ft operational |
Today the market is anchored by four carrier-class operators — Viettel IDC, VNPT, FPT Telecom, and CMC Telecom — with Hanoi Telecom and VNG also operating facilities, and 33 commercial data centers in total. Foreign players such as Gaw Capital and Worldwide DC Solution (Singapore) have entered, and the announced pipeline (~5.15 million sq ft of white floor against ~0.95 million sq ft operational) is several times the current installed base. FPT's AI Factory adds the GPU layer: a US$200 million cloud with thousands of NVIDIA H100 GPUs, in service since January 2025 — capacity that matters for genomics, where tools like DeepVariant can run on GPUs.
For the genomics analysis stack, the picture is different too. High-throughput sequencing is accessible in-country, and the compute side can be built or rented locally: SLURM HPC clusters, GPU nodes for tools like DeepVariant, and S3-compatible object storage from domestic providers — the same architecture described in Part 2 and Part 3. The open-source stack this series proposes (Nextflow, nf-core, GLnexus/DPGT, Hail) runs on that infrastructure today; the proof-of-concept repositories referenced throughout the series are public and reproducible.
The point is not that infrastructure is free — it still takes capital and operational discipline. The point is that in 2018 the architecture was aspirational, and in 2026 it is deployable with local resources.
1.3. Human resources: from a few pioneers to a generation
In 2018, expertise was concentrated in a small number of people, many of them abroad, and every project effectively had to train its own team.
By 2026, the picture is different. University programs, dedicated training initiatives, and the return of experienced diaspora have widened the pool considerably, and multi-institution collaboration has become the norm rather than the exception. The gap that remains is not headcount but experience at population scale — the kind only long-running projects can build.
The bottleneck is shifting. Finding people is no longer the hardest part; keeping them, organizing them across institutions, and giving them long-term projects is. That is a much better problem for a country to have.
1.4. Government support: a growing role
In 2018, genomics in Vietnam was driven mainly by private-sector and university initiatives, and policy around genomic data was still taking shape.
By 2026, policy points in the same direction as the technology:
- Healthcare and population: the National Assembly approved the national target program on healthcare, population, and development for 2026–2035 (Resolution 262/2025/QH15).
- High technology: the Law on High Technology 133/2025/QH15 and Decree 260/2026 took effect on 1 July 2026, with incentives for priority high technologies.
- Budget: roughly US$3.8 billion of the 2026 state budget is directed to science, technology, innovation, digital transformation, and Project 06 (the national digital transformation and population data program).
- Biotech ambition: Vietnam is aiming to establish its first regional biohub, with AI-biotech convergence as a focus (VIR).
- Market momentum: the Vietnamese genomics market is estimated at US$180–220 million and projected to roughly double by 2030 (Global Health Week Vietnam).
Genomics is also serving national missions directly: VinGenChip — the first biochip designed from Vietnamese genomic data, developed by GeneStory based on VN1K findings — is being used in the program to identify martyrs' remains, where millions of DNA samples are analyzed to match relatives with unidentified remains.
1.5. What this means for a national 1000-genome project
Put the three together and the conclusion is straightforward. In 2018, VN1K had to build its infrastructure, its team, and its methods from scratch. In 2026, a national project can stand on what VN1K and others have already built: the reference exists, the talent pool is wider, the tools are open source, and policy is aligned.
What remains is execution and governance — the themes of this series: standardized variant calling, scalable joint genotyping, a research-ready analytics layer, sovereign storage, and clear data access rules. Those are engineering and organizational problems, not scientific unknowns.
| Dimension | 2018 | 2026 |
|---|---|---|
| Infrastructure | Small clusters and local disks; population-scale analysis out of reach for most teams | SLURM HPC, GPUs, and domestic S3-compatible object storage; the open-source stack runs in-country |
| Human resources | A handful of specialists, many abroad; every project trained its own | Wider pool from university programs, training initiatives, and returning diaspora; multi-institution collaboration |
| Government support | Early stage; genomics driven by private and university initiatives | National healthcare and population program 2026–2035; High Technology Law 2025; ~US$3.8B of the 2026 budget for science, innovation, and digital transformation |
| Data and reference | No Vietnamese reference genome; foreign references | VPR, the first Vietnamese reference genome; 42M+ variants (~8.5M novel); imputation panel |
References
Series repositories
- omicslab-hpc — Ansible automation to build the SLURM HPC cluster. https://github.com/vieomics/omicslab-hpc
- nf-germline-short-read-variant-calling — standardized short-read germline variant-calling pipeline. https://github.com/vieomics/nf-germline-short-read-variant-calling
- nf-modules — Nextflow modules and subworkflows shared across the series pipelines. https://github.com/vieomics/nf-modules
- omicslab-kit — gkit proof-of-concept implementations (GLnexus, DPGT, Spark-on-SLURM, Hail). https://github.com/vieomics/omicslab-kit
Sources
Key sources for this part — market figures vary between research firms because of methodology and scope:
- VN1K — Nature Communications paper · VinUni announcement · data portal
- Data center market 2016–2020 (~US$858M in 2020) — VietnamNet
- Viettel Hòa Lạc (21,000 m², 30 MW, 2,400 racks) — Viettel Family · earlier 800-rack campus
- First hyperscale data center (140 MW, 10,000 racks) — PS Market Research
- 2025 market and capacity estimates — Expert Market Research · Mordor Intelligence
- Operators, 33 data centers, and pipeline — Vietnam.vn · Arizton
- FPT AI Factory (NVIDIA H100 GPUs, ~US$200M) — FPT · Vietnam.vn
- Data Law and National Data Center — Tilleke & Gibbins · VOV
- Data localization (Decree 53/2022) — U.S. International Trade Administration
- 2026 state budget (~US$3.8B for science, innovation, digital transformation) — VietnamNet
- Healthcare and population program 2026–2035 — Resolution 262/2025/QH15
- High Technology Law 2025 — KPMG
- Biohub and genomics market — VIR · Global Health Week Vietnam
In Part 5, we recap how to build the program, the optional extensions and future roadmap, and the questions ahead.
This is Part 4 of the series. Continue to Part 5 for the recap, the future roadmap, and the questions ahead.
