Every few months a headline promises to store humanity’s entire digital life on a coaster-sized piece of glass for as long as the universe has existed. The latest version comes from SPhotonix, a UK startup whose “5D memory crystal” can supposedly hold 360 terabytes per disc and survive for 13.8 billion years. The coverage writes itself: Superman’s crystal, data that outlives the human race, a Tom Cruise movie cameo.
Here is the question almost nobody asks: does any of this matter to a business that is not a national archive? The short answer is not yet, and the more interesting answer is why the gap between the press release and the product is the actual story.
5D optical data storage is a write-once archival method that etches nanoscale structures, called voxels, inside fused silica glass using ultrafast femtosecond laser pulses. It is called “five-dimensional” because each voxel encodes data across five parameters at once: three spatial coordinates (X, Y, Z) plus the orientation and intensity of the voxel’s birefringence, the way it bends polarized light. The glass needs no power to retain data and is read back optically. It is built for cold storage, meaning data you write once and almost never touch.
What is actually true, separated from the marketing
The physics is real and the people are serious. The technique traces back to 2013 research by Professor Peter Kazansky at the University of Southampton, and SPhotonix spun out of that lab in 2024. The company recently raised its first outside funding and said it is moving the technology from the bench toward data center pilots over the next two years.
The problem is that nearly every headline number is a projection, not a shipping spec. Here is the honest scorecard.
| Claim in the headlines | What SPhotonix actually says | The catch |
|---|---|---|
| 360 TB per disc | Current density is “several TB per disc,” with roughly 20 TB hoped for within a couple of years | 360 TB is the theoretical ceiling, not today’s product |
| Lasts 13.8 billion years | True at 190 degrees Celsius; at room temperature the figure is closer to 10 to the 20th power years | The “age of the universe” line attaches to the hot-oven test, not your shelf |
| Survives extreme heat | Withstands fire up to 1,000 degrees Celsius | Accurate, and genuinely useful for disaster resilience |
| Ready for the data center | Prototypes write at 4 MB/s and read at 30 MB/s | That is slower than a 1990s hard drive |
| Cost-effective at scale | Early writer around $30,000, reader around $6,000 | Hardware economics are nowhere near commodity |
None of this means SPhotonix is dishonest. It means the press did what the press does with a 13.8-billion-year number: ran the ceiling as if it were the floor.
The 360TB asterisk
The single most important fact about this technology is the distance between what it can store today and what the marketing implies. SPhotonix’s own FAQ puts current achievable density at a few terabytes per disc, on the same order as a regular hard drive, with a near-term hope of 20 TB and an “ultimate” estimate of 360 TB.
That is roughly a 70x gap between the headline and the product. For a founder, this is a familiar shape. It is the same chart you would draw for almost any deep-tech startup raising on a future capability rather than a current one. The number is not a lie. It is a roadmap wearing a press release’s clothing, and the job of a skeptical reader is to keep asking which one they are looking at.
The speed picture tells the same story. A write speed of 4 MB/s means filling a single disc to even a few terabytes takes days of continuous lasering. The roadmap targets 500 MB/s within three to four years, which would land it in the same ballpark as magnetic tape, but those speeds have not been shown outside the lab.
This is not a flaw the company is hiding. It is the entire point of the category. Cold storage is allowed to be slow because the data is rarely read. The mistake is reading “stores data forever” as “fast, cheap, and ready.” It is none of those three yet.
Why this market exists at all
Here is the part the gadget blogs skip, and it is where the real business sits. The world is drowning in data that nobody looks at but nobody is allowed to delete.
Between 60 and 80 percent of all stored data is archival or “cold,” meaning it is accessed once a month or less, and most of it currently sits on power-hungry hard drives. The volume of cold data alone is expected to pass 6 zettabytes, and global data creation is forecast to more than triple from roughly 218 zettabytes in 2025 to over 718 zettabytes by 2030. Regulations in finance, healthcare, and government force companies to retain records for years or decades, whether or not anyone ever reads them again.
That creates a genuine and growing pain: storing enormous volumes of rarely-touched data without paying to keep spinning drives powered, cooled, and replaced every few years. A glass disc that needs zero electricity to sit in a vault for a century is a real answer to a real cost problem. The market for archival storage is measured in the billions of dollars and growing steadily, which is why this is not a science fair project. It is a land grab for the bottom tier of the storage pyramid.
The business model is the actual story
Strip away the durability theater and SPhotonix is making a specific and revealing strategic bet. Rather than building and selling storage as a service, it plans to license the media and the optical platform to partners and integrate into existing data centers, a model it compares to Arm and NVIDIA.
This is a smart, capital-light move for a company with $4.5 million in the bank, and it sidesteps the brutal economics of competing head-on with hyperscale cloud archives. License the physics, let someone else carry the data center capex. The risk is the one every IP-licensing play carries: you only get paid if a partner ships volume, and you do not control whether they do.
It also puts SPhotonix in a crowded race. Microsoft has been running Project Silica, its own glass-storage program, for years, with roots in the same Southampton lab. Cerabyte is pushing ceramic-coated media for robotic libraries. A whole separate camp is betting on DNA storage. The contenders are not really fighting each other yet. They are all fighting tape and hard drives for the privilege of being the default cold tier a decade from now.
| Contender | Medium | Pitch | Status |
|---|---|---|---|
| SPhotonix 5D crystal | Fused silica glass | Highest volumetric density, billions of years | Early pilots, licensing model |
| Microsoft Project Silica | Glass platters | Cloud-scale, backed by a hyperscaler | Research, internal Azure focus |
| Cerabyte | Ceramic on glass | Low-cost, fast robotic access | Prototype |
| DNA storage | Synthetic DNA | Insane density, biological reads | Lab-stage, slow and costly |
What it means for founders and operators
Be honest with yourself about whether you are the customer here. If you run a content site, a SaaS app, or a normal business, you will not buy a 5D crystal writer this decade, and you do not need to track this weekly. Your “archive” is an S3 Glacier bucket, and it will stay that way.
Two things are worth taking from this, though. First, it is a clean case study in reading deep-tech claims. When you see a number that sounds impossible, find the conditions attached to it. The 13.8 billion years was hiding a 190-degree oven, and the 360 TB was hiding the word “ultimate.” Second, the underlying market signal is real even if this specific product is early: the cost and energy burden of storing cold data is becoming a genuine line item, and whoever solves it cheaply wins a very large, very boring prize.
Frequently asked questions
What is a 5D memory crystal?
It is a small disc of fused silica glass with data etched inside it by a femtosecond laser. Each microscopic mark, called a voxel, stores information across five parameters: its three spatial coordinates plus the orientation and intensity of how it bends polarized light. It is a write-once medium designed for very long-term archival storage.
Can a 5D crystal really store 360 TB?
Not today. SPhotonix describes 360 TB as the projected ultimate capacity. Its own materials put current density at a few terabytes per disc, with around 20 TB targeted within a couple of years.
Does the data really last 13.8 billion years?
That figure is an extrapolated decay time at 190 degrees Celsius. At room temperature the projected stability is far longer, on the order of 10 to the 20th power years. Both numbers are extrapolations, not measured outcomes, and the discs can also survive fire up to 1,000 degrees Celsius.
Why is it so slow?
Current prototypes write at about 4 MB/s and read at about 30 MB/s. That is fine for cold archival storage, where data is written once and rarely retrieved, and where waiting ten seconds or more for access is acceptable.
Who are the main competitors?
Microsoft’s Project Silica is the closest rival in glass storage, Cerabyte is developing ceramic-based media, and several companies are pursuing DNA-based storage as an alternative long-term archive.
Should my business use this?
Almost certainly not yet. It is aimed at deep-archive and compliance-driven cold storage at the data center level, not at everyday business backups, which are well served by existing cloud archive tiers.
The Business Model Analyst Take
The 5D memory crystal is real science wrapped in unreal marketing, and both halves deserve respect for what they are. The durability is the bait. The business is the cold-storage land grab underneath it, and SPhotonix’s licensing bet is the part actually worth watching, because it tells you the company knows it cannot win on capex alone.
For everyone outside a hyperscaler or a national archive, the practical takeaway is smaller and sharper than the headlines. This is a masterclass in reading a spec sheet. Find the conditions hiding behind the impossible number, and you will spot the difference between a breakthrough and a roadmap every time. The glass might last until the end of the universe. The press cycle will not survive the next funding round.
