Genomics at Scale: What Falling Sequencing Costs Actually Unlock

A Five-Order-of-Magnitude Fall

The Human Genome Project took thirteen years and roughly three billion dollars to produce a single reference sequence of the human genome, completed in 2003. By 2014, Illumina’s HiSeq X Ten had pushed the reagent cost of sequencing an individual genome below $1,000. By 2024, the company was advertising whole-genome sequencing for as little as $200. And in early 2026, a cluster of competing platforms — Element Biosciences’ benchtop VITARI system, Ultima Genomics’ UG200 series, and MGI’s DNBSEQ line — have all made credible, independently verified claims to have crossed the long-mythologized $100 genome threshold.

Measured against Moore’s Law, the semiconductor industry’s famous doubling-every-two-years cost curve, genomic sequencing costs have fallen considerably faster over the past two decades. That comparison has appeared on enough conference slides to become a cliche, but the underlying fact remains genuinely remarkable: the cost of reading a complete human genome has dropped by roughly five orders of magnitude within a single career’s span, a rate of cost decline with few genuine parallels in the history of any technology, semiconductors included.

When the Cost Curve Stops Being the Story

What is actually interesting about 2026 is that the cost curve itself has stopped being the story. Raw sequencing capacity is well on its way to becoming a commodity — Ultima’s UG200 Ultra configuration alone can process sixty thousand genomes a year on a single instrument, and multiple vendors are now competing on that same axis, driving the kind of price competition that typically signals a technology has crossed from scarce, differentiated capability into interchangeable commodity infrastructure.

Commoditizing infrastructure is exactly the pattern that, in prior technology cycles, has shifted where the investable value actually sits: away from the raw capability and toward the interpretation, application, and workflow layers built on top of it. Cloud computing followed this arc — raw compute and storage became commoditized and cheap, while the genuine value migrated to the software and services layers built on top. Genomics in 2026 is following that same arc, several years behind where cloud computing followed it, but unmistakably on the same trajectory, and an investor who is still primarily evaluating sequencing-hardware companies in 2026 is, in a real sense, fighting the last war.

Where Cheap Sequencing Already Changes Clinical Practice

The most immediate application of cheap, fast sequencing is in oncology and rare disease diagnosis, where turnaround time genuinely changes clinical decisions rather than merely academic understanding. A tumor genomic profile that once took weeks to return, at a cost that limited its use to the most resourced hospital systems, can now inform first-line treatment selection within days at a fraction of the previous cost — directly enabling the kind of biomarker-driven, precision oncology approach that underlies much of the antibody-drug-conjugate and targeted-therapy pipeline discussed elsewhere in this issue, where treatment selection increasingly depends on knowing a specific tumor’s genomic profile before a physician can even choose the right drug class to try.

Newborn screening programs, population-scale genomic initiatives, and pharmacogenomic testing that flags which patients will respond poorly to a given drug before it is ever prescribed are all becoming economically viable at a scale that was simply not achievable even five years ago. Pharmacogenomic testing specifically deserves attention: a genomic panel that flags, before a first prescription is ever written, which patients carry variants associated with poor drug metabolism or a heightened adverse-reaction risk converts a costly trial-and-error prescribing process into something closer to a matched, first-attempt treatment decision — a genuinely underused capability given how inexpensive it has become to generate the underlying data.

The New Frontier: Single-Cell and Spatial Genomics

The frontier has consequently shifted from bulk whole-genome sequencing toward single-cell and spatial genomics — technologies that do not just read a genome but map gene expression across individual cells within their actual tissue context, preserving the spatial architecture that bulk sequencing necessarily destroys. Bulk sequencing takes a tissue sample, homogenizes it, and reads out an averaged genomic signal across every cell in the sample — useful, but it discards exactly the information that matters most in a tumor, where the cells at the invasive edge behave completely differently from the cells at the tumor’s core, a distinction bulk sequencing simply cannot see.

That is a materially harder and more expensive problem than sequencing a bulk genomic sample, and it is where the genuinely differentiated science, and the genuinely differentiated venture opportunity, has migrated as bulk sequencing itself has commoditized. Understanding not just what a cell’s genome says but where that cell sits relative to its neighbors inside a tumor, an organ, or a developing embryo is the next multi-year technical frontier, and it is being built by a different, smaller set of companies than the ones that won the bulk-sequencing cost race — a genuine changing of the guard within the genomics investment landscape that most generalist coverage of the sector has not yet caught up to.

An Equalizer for Underrepresented Populations

For emerging markets specifically, falling sequencing costs are a genuine equalizer in a way few other biomedical technologies have been. A national genomics initiative that would have been financially unthinkable a decade ago — building a population-representative genomic reference dataset for a country whose genetic diversity has historically been underrepresented in global research, most of which has been built on populations of European ancestry — becomes financially plausible at sub-$100-per-genome pricing.

That matters clinically as well as commercially: drugs and diagnostics validated primarily on European-ancestry genomic data do not always translate cleanly to other populations, since the frequency and clinical significance of specific genetic variants can differ meaningfully across populations with different ancestral histories. Building genomic reference data specific to India’s or East Africa’s population structure is a precondition for precision medicine actually working equitably across the geographies this fund operates in, rather than only in the markets where the original data was collected — a gap that falling sequencing costs finally make it economically realistic to close, rather than merely aspirational.

Where the Durable Value Now Sits

The interpretation layer is where we think the more durable venture theses now sit: companies building the clinical decision-support tools that translate a raw genomic variant call into an actionable treatment recommendation — a step that still requires substantial specialist expertise to do well, since a raw list of genetic variants is close to meaningless to a treating physician without software and expert curation that translates it into a specific, actionable recommendation. The data infrastructure that makes population-scale genomic datasets usable for drug-target discovery and clinical trial patient stratification is a second layer worth real attention, since a genomic dataset’s value scales enormously with how easily it can be queried, cross-referenced, and integrated with clinical outcomes data, a capability that requires genuine software and data-engineering investment well beyond the sequencing itself.

And the regulatory and reimbursement pathways that determine whether a genomic test actually changes a physician’s prescribing decision or simply sits in a patient’s file unread are a third, less technically glamorous but equally important layer — a genomic test with excellent analytical performance that no payer will reimburse and no clinical guideline recommends ordering has, in practice, close to zero real-world clinical impact, regardless of how sophisticated the underlying sequencing technology.

The Genome Was Never the Hard Part

Raw sequencing capacity was the right thing to fund fifteen years ago, when the constraint was genuinely the cost of reading a genome at all. It is close to the wrong thing to fund today, now that a handful of well-capitalized platform companies have driven that cost to the floor. The genome itself was never the hard part — turning it into a decision a doctor can act on is, and that is where we are spending our diligence time now, across the interpretation, infrastructure, and reimbursement layers rather than the sequencing hardware layer that first made all of this possible.