DNA for data storage?

Executive Summary 

• OpenAI’s GPT-4o advances text-to-image generation with improved text rendering and character consistency.

• Terrestrial Energy enters the nuclear reactor market via a $280M SPAC merger, competing with Oklo.

• Meta AI expands to Europe, bringing multilingual chat to WhatsApp and other apps.

• DNA emerges as a promising, durable data storage solution, though cost and error challenges remain.

News

🏙️ OpenAI launched the most advanced image generator. Link

• With this launch, OpenAI has turned the tables on the best text-to-image tool in the market. AI image tools commonly struggle with drawing text and maintaining consistency across multiple stages (e.g., when prompted to generate a character first and then keep this character consistent for further images). This model executes these tasks very well.

• While it is still not perfect and won’t fully replace your in-house design team, GPT-4o is a significant step forward.

• This tool is available for Plus plans and is expected to be launched for Enterprise and Free users soon.

☢️ New player in a nuclear reactor market. Link

• Terrestrial Energy, a North Carolina-based nuclear startup, merged with an acquisition company, expecting to net $280 million and list on Nasdaq as IMSR. Prior to the SPAC merger, it had raised $94 million.

• Small nuclear reactors are very attractive to current AI players, promising to close the energy gap, required for the development of advanced models.

• Terrestrial is an emerging competitor to Oklo, which went public in 2024, and secured funding by Sam Altman. Meta, Amazon and Google are also funding separate nuclear reactor developments.

 💬 WhatsApp received AI in Europe. Link

• Meta AI, launched in the US in 2023, is now being rolled out in Europe, starting with its chat function in six languages across Facebook, Instagram, WhatsApp, and Messenger.

• Meta AI will be accessible through a new blue circle icon and can be used in direct messages and group chats by typing "@MetaAI" followed by a prompt, allowing users to easily find information, generate content, and get assistance within their social apps.

• On a competitive note, Grok, developed by xAI, is available on Telegram starting from the 27th of March.

AI tool of the week

Character AI

It is a platform for creating and interacting with AI-powered characters, each with unique personalities, backstories, and conversational styles. Simply choose a character or create your own, and engage in dynamic, text-based conversations that can range from casual chats to collaborative storytelling.

This platform gained significant popularity in its early stages, becoming a viral sensation and attracting millions of users. The company has received substantial funding from investors like Andreessen Horowitz, valuing the company at over $1 billion in 2023. It is rumoured to be valued at around $2.5 billion now.

However, Character.AI has its limitations: interactions can sometimes feel repetitive or nonsensical, particularly with complex or nuanced topics. The platform also faces challenges with content moderation, struggling to prevent characters from generating inappropriate or harmful responses.

Discussion

DNA for data storage?

Every second, humanity generates staggering amounts of data—but where will it all go? Traditional storage systems are running out of space and require significant energy. The answer may lie in DNA, nature’s storage system. With its incredible density and durability, DNA has the potential to become the medium for preserving humanity’s information legacy.

How does it work?

DNA is composed of four chemical building blocks—adenine (A), cytosine (C), guanine (G), and thymine (T)—whose sequences encode biological messages. These same sequences could be encoded with digital information by translating binary code (1s and 0s) into combinations of A, C, G, and T.

Main benefits

One gram of dry DNA can hold such a vast amount of data that all the information on the internet could fit into a volume smaller than a sugar cube.

In addition, DNA is incredibly durable. When stored under optimal conditions, it can last for thousands or even millions of years without degrading. Unlike traditional data centers that require constant cooling and energy input, DNA can be stored at room temperature with minimal energy costs, making it an environmentally sustainable solution.

Core hurdles

Synthesising and sequencing DNA remains slow and expensive, limiting their practicality. Current technologies require hours or even days to encode and retrieve small amounts of data.

In addition, errors are common during synthesis and sequencing processes. These errors require sophisticated error-correction algorithms to ensure data integrity. While developments like enzymatic synthesis have improved accuracy and speed, error rates still range from 0.1% to 1% per base sequenced.

Going forward

Researchers are improving DNA storage technology. At the University of Washington team demonstrated that DNA can identify files by similarities without decoding entire databases, potentially transforming search algorithms. New methods like photolithographic and enzymatic synthesis could enable cost-effective, efficient DNA data production, with fewer errors, and less toxic waste.

These innovations suggest that while widespread adoption may still be years away, the foundations are being laid for a future where DNA becomes more than just a molecule—it becomes humanity’s ultimate archive.

Our service

Do you need help with your data ecosystem? Reply directly to this email or reach out to [email protected]. More information about our service is available here.