Solexa read billions of tiny DNA fragments in parallel, and Illumina bought the trick

By the time the Human Genome Project wrapped up, its raw sequence had cost roughly $300 million produced over years on several hundred capillary sequencers, and in 2007 an individual genome still ran about $10 million by the same method. The obvious next step was brute force: more capillaries, longer reads, bigger budgets.

Cambridge chemists Shankar Balasubramanian and David Klenerman founded Solexa in June 1998 on the opposite bet. Their reversible-terminator chemistry uses 3'-O-azidomethyl fluorescent nucleotides so each incorporation cycle goes essentially to completion without over-incorporation, and an engineered polymerase accepts the unnatural bases; after imaging, the block and dye are cleaved off and the next cycle starts. The missing piece — amplifying each surface-bound fragment into a bright clonal cluster by bridge amplification — came in from solid-phase colony work Solexa integrated before launch.

The 2008 proof landed in Nature: 135 Gb of sequence from a Yoruba male genome, produced in eight weeks on six Genome Analyzers at a consumables cost of about $250,000, at 40.6-fold average depth with more than 99.9% of the reference covered, plus four million SNPs and four hundred thousand structural variants. Consensus accuracy on a test BAC exceeded 99.999%.

Solexa had reverse-merged into Lynx Therapeutics in 2005 to stay funded; on January 26, 2007 Illumina completed a stock-for-stock merger that made it a wholly-owned subsidiary, a deal the Guardian later put at about $650 million. The Genome Analyzer shipped that year, and the cost curve never turned back: about $10,000 a genome in 2014, $600 by 2022, with Illumina holding around 80% of the sequencing market.

Millions of short reads in parallel beat one long read

Massive parallelism — tens of millions of colonies imaged per run — replaced fleets of capillary machines, so throughput scaled with optics and chemistry, not with floor space.

The human reference genome already existed, so short reads stopped being a defect: alignment replaced the assembly that long reads were for.

The reversible 3' block drove each cycle essentially to completion, keeping per-base accuracy high enough for consensus calls like 99.999%.

Illumina bought the technology rather than racing it, converting a Cambridge cash-strapped startup's chemistry into a market of one.

When throughput is the bottleneck, shrink the unit of work and run millions in parallel. A cheap, imperfect operation plus an external map can beat a perfect, expensive one.

Illumina rode the SBS curve to a near-monopoly: about 70% of the sequencer market by 2014, around 80% by 2022, when the NovaSeq X raised throughput to 20,000 genomes a year and pushed the price toward $200 per genome. The Broad Institute alone had sequenced more than 486,000 genomes by 2022. Cheap genomes underpin cancer blood tests, rare-disease diagnosis and the COVID-19 genomic surveillance that seeded vaccine design. Balasubramanian and Klenerman's chemistry is the direct ancestor of nearly all DNA data produced today.