Five optical storage startups aim to provide long-term and energy-efficient archive technology: Cerabyte, Ewigbyte, HoloMem, Optera, and SPhotonix. They are developing optical technologies somewhat akin to Microsoft’s Project Silica research. All five are pursuing the dream of replacing tape, with its five to seven-year lifespan, with a 100-plus-year medium that is more resistant to chemical, shock, radiation, water, and heat dangers while retaining air-gapped immutability, low energy consumption, and high capacity, hopefully better than tape.
Cerabyte
Cerabyte stores its data using femtosecond laser etching of a ceramic recording layer on a glass tablet or slab substrate. Up to ten tablets are stored in LTO-size cartridges held in a tape library-type system. Robotic carriers move the cartridges between storage shelves and read and write stations. Tablets are longer-lived, faster, and cheaper than tape, lasting more than 100 years compared to tape’s seven to fifteen, shipping data at 1-2 GB/s versus tape’s 1 GB/s (compressed speed), and costing $1 per TB against tape’s $2 per TB.
It foresees 100-plus PB racks by 2030, and less than ten seconds to first byte. The German company, which has offices in Boulder and Silicon Valley, is developing a pilot production system and has partnered with and received investment from Pure Storage, Western Digital, In-Q-Tel, and the European Innovation Council (EIC) Accelerator fund.
We were told about particular advantageous aspects of Cerabyte’s technology by CMO and co-founder Martin Kunze. First of all, Cerabyte’s dark and absorbing coating decreases the energy to write a bit by a factor of 100 compared to uncoated glass, which is transparent for the laser. More laser energy is required to create a data storage structure in uncoated glass.
Kunze tells us that separation of uncoated glass sheets is almost impossible as the glass sheets adhere to each other. A so-called air-knife can be used to separate them but this is difficult to achieve. Cerabyte’s ceramic coating is actually an anti-adherent and used to lubricate surfaces in the metal industry. Because of this, the tablets can be readily stacked in cartridges and selected by a robot picker.
Interestingly, he says Cerabyte could patent its technology because it wrote and read its data from a ceramic coating whereas he understands that simply writing information onto any material, including glass, is not patentable, as it has been done for a very long time.
Cerabyte is thinking about a €50 million ($58 million) funding round, with potential hyperscaler involvement. A first-generation product could come next year and there is a 1,000-square-foot space in Vienna for first production.
Ewigbyte
Ewigbyte aims to develop optical tablet bit storage using femtosecond laser ablation of the uncoated glass surfaces, burning data blocks as it were. The tablets will be used for deep archive “write once, read never” (WORN) applications. Ewig is a German word meaning eternal.
It is a new German company started up in April by the ex-COO of Cerabyte, Dr. Steffen Klewitz, its CEO, and three co-founders: Head of Technology Phil Wittwer, Head of Operations Dr. Ina von Haeten, and lawyer Dr. Christian Marquand. Wittwer comes from being a Strategic Program Lead Automotive at AWS, and a head of SW standardization and Compliance at BMW before that. He is not an optical engineer. The company is developing what it calls a Minimum Viable Product (MVP), based on laser ablation machines. It says suppliers are ready to deliver components: lasers, optics, handler and robotics, and chassis. Ewigbyte will deliver, it claims, faster than tape or disk read/write speed and lower latency.
The WORN MVP is planned for 2026, with a data center WORM archive product demonstrated in 2028, and rack-scale warm data storage planned for 2029.
Wittwer has posted on LinkedIn: ”In the age of AI, density is no longer the decisive metric for cold data storage.” He predicts that storage prices will rise after decades of decline because enterprise HDDs, SSDs, and LTO tape all face rising prices and longer lead times. Data growth is heading to the zettabyte level and beyond, and energy costs are rising too, with supply increases limited. Cold data storage has to become less expensive and use less energy. It also has a role to play in digital sovereignty domains and Ewigbyte is aware of this.
Ewigbyte is looking for pre-seed investors.
Project Silica
Microsoft’s University of Cambridge-based Project Silica builds on University of Southampton research in nanoscale voxels, polarization-based patterns etched inside a silicon glass medium by femtosecond lasers. There could be up to 100 layers of such voxels in a glass tablet.
The research was carried out by a Professor Peter Kazansky-led team in the 2010-2019 period. Kazansky and his son started up SPhotonix – see below – last year, indicating that they are no longer involved with Project Silica.
The project involves redesigning tape library robotics as well as the laser, optical writing and reading stations, and library software.
Project Silica is managed by Richard Black, Microsoft Partner Research Manager. He started out there as a Senior Researcher 25 years ago. It has gone quiet in recent months. There were seven announcements in 2024, three in 2023, and four in 2022. So far there have been none in 2025.
Given the apparent slowdown in public activity and the lead researcher’s focus on his SPhotonix startup, there is a risk that Project Silica has stalled. We have asked the project’s team about its status and will add any further information here.
SPhotonix
We first wrote about SPhotonix’s 5D silica glass storage technology back in 2021. We said: “Scientists in the Optoelectronics Research Centre (ORC) at the University of Southampton, led by Professor Peter Kazansky, continued researching the optical properties of silica glass at nanoscale and have devised a dimensional recording method using two optical and three spatial dimensions of silica glass.” They create voxels – voids – in the glass, using femtosecond lasers, with customized edges – nanolamellae – to affect the way light can encode data, with 4 bits per voxel. The voxel-encoded glass crystals, disks, or tablets can last for thousands if not billions of years.
Peter Kazansky, yes, the Project Silica Kazansky, and his electrical engineer and entrepreneur son, Ilya, founded SPhotonix in April 2024 to productize their Femto-Etch technology. Ilya Kazansky is CEO and his father is Chief Science Officer. They raised $4.5 million in a pre-seed round in November to pay for the scaling of its Photonics and Engineering teams and commercializing its two core applications; advanced optics and data storage. At the moment it has a secure data write and data read service available at a facility in Switzerland, and hopes to release a $6,000 read device by July 2027. Its write device would cost around $30,000.
SPhotonix and its founders have participated in eye-catching and aspirational marketing events such as preserving the human genome (15 GB) in a 5D Memory Crystal, storing all of Wikipedia in a similar crystal, creating a Boucheron‘Quatre 5D Memory’ ring, “a groundbreaking fusion of deep tech and high jewelry,” and having the crystal feature in Tom Cruise’s Mission Impossible ‘The Final Reckoning’ movie.
Its technology can currently store 270 TB in a 25 mm diameter glass disk, and has a 4 MB/s write speed and 30 MB/s read speed. This is slow, and SPhotonix wants to raise it to the 500 MB/s read/write area. LTO-10 tape can read and write raw data at 400 MB/s and 2.5:1 compressed data at 1,000 MB/s. It wants to increase capacity to about 360 TB by using 3 mm thick, 5 x 5 inch fused quartz plates.
Its FAQ says a 12 cm-size, 1.2 mm thick CD disk could store several TB. It wants to up this to around 20 TB/disk by 2028 and sees an ultimate achievable capacity of 360 TB. This would need to be spun to be read, like any other optical disk.
SPhotonix does not want to build archival storage systems. Instead it wants builders of such systems to license its technology, and reckons its technology could be productizable in three to four years. The customers would be archive-burdened enterprise data centers and hyperscalers offering archival data services.
Our understanding is that the core science underlying SPhotonix’ technology is backed by years of validated research but the productizing of its technology so that read and write devices can be manufactured at scale, and perform quickly and reliably, is a significant challenge. It would need to be carried out, we think, with a manufacturing-capable partner. On its website, it says: “We invite partners to collaborate with us and accelerate our journey to market, bringing the future of data storage to life.”
Optera Data and HoloMem
Optera Data, according to its website, has a CEO, Singapore-based Geoff Macleod-Smith, who runs the Digistore Solutions Group, and a researcher, Dr Nicolas Riesen, based in the Adelaide area of Australia. It is developing data storage using fluorescent waveband changes at points in a recording medium. Projected costs are 10x lower than those from Cerabyte and 25x lower than tape.
We think this company is a long way from delivering any kind of product. There is no known funding or production technology partnerships and no recent announcements about its development efforts. We have asked and, if and when we hear, will add the information here.
UK-based HoloMem is developing a form of optical tape, with data recorded as multi-layer micro-holograms along the length of the tape. Little has been heard from the company for the past few months. We have asked for an update, and will place it here if and when we receive it.
Overall, our impression is that Cerabyte is further along the road than any other supplier. SPhotonix has funding and a solid research background, but no manufacturing partner. HoloMem has some funding and has an agile team of bright engineers. Ewigbyte has no demonstratable technology or funding. Optera seems to be a somewhat lightweight company, while Project Silica is possibly over-engineered and could now be a lapsed project.
Cerabyte’s Kunze said a working group has been established to develop a LAST (Long-term Accessible Storage Technology) framework that answers the question: “What does an IT professional in the future need to know to independently read the records on today’s long-term media?” It was initiated by Cerabyte, but led by people from Fedora, Arcitecta, IEEE, SNIA, DPC, OPF, Library of Congress, and Vint Cerf/Digital Vellum. Various long-term data storage startups such as SPhotonix, Optera, and DNA have been invited to join in as well.
