Manifold 2.0: Think Bigger

2024 was a defining year for Manifold. We navigated volatility, learned valuable lessons, and emerged with deeper conviction in both our mission and the potential of Bittensor. We’ve entered 2025 more aligned, more experienced, and more focused than ever. Manifold 2.0 represents the beginning of that next chapter.

We believe the most powerful protocols are those that ultimately disappear into the background—where the user experience speaks for itself, and the infrastructure quietly does the work. Our mission is to make Bittensor that kind of protocol. And to do that, we must think bigger.

From Protocol to Product

Since our founding in 2023, much of our focus has been internal to Bittensor. Subnet design, incentive modeling, and protocol-level experimentation consumed much of our energy. This work laid the foundation for Subnet 4, which coordinates anonymous clusters of compute to serve real-time inference via Targon. As of writing, the miners on Targon can afford approximately 1,400 H200s a day. This places us in hyperscaler territory, on par with top-tier AI companies that have raised significant amounts of venture capital.

Targon began as a proof of concept. It’s now a production-grade inference platform, helping thousands of users access powerful models at lower cost and higher speed. With the release of Version 6, we are introducing confidential compute - enterprise-grade privacy delivered through our decentralized infrastructure.

For model developers, today’s distributed networks offer no protection for private model weights. For enterprise consumers, these networks expose all data - allowing any participant to observe potentially sensitive information. As a result, enterprises have been unable to leverage the cheaper, more scalable AI provided by networks like Bittensor. That all changes today.

Centralization of Value, Not Infrastructure

Like every project in Bittensor, Manifold has benefited from persistent TAO inflation. That’s not a weakness—it’s a feature of any early network. As we emerge from this bootstrapping phase, we understand the critical importance of intentional design in order to build a sustainable ecosystem.

In traditional companies, loss leaders are essential. Products like Google Chrome are free not because they’re unimportant, but because they drive value elsewhere. Bittensor is no different. Some layers in the Targon stack may never be directly profitable—but they’re strategically critical to the health of our system.

That’s why we’re building Targon to support multiple core functions: inference, pre-training, post-training, and research. All within a single, value-aligned surface.

Standalone subnets can serve important experimental purposes. But over time, spreading value too thin risks weakening the very foundations we’re trying to build. We believe consolidating key functionality into SN4 maximizes our chances of long-term durability.

Introducing Manifold 2.0

This week, we’re introducing a new wave of products—each designed to showcase what’s possible on Bittensor while delivering real utility to users. We’ll reveal one new product every day.

Targon V6 Whitepaper – Live Now

Targon is now faster, more secure, and more production-ready than ever. The Targon Virtual Machine (TVM) ensures that proprietary model weights remain confidential and are executed solely within attested, cryptographically secure environments.

Sybil – Live Now

Sybil is our reimagined decentralized search engine. It now features a redesigned interface, support for multiple models, and a dramatically improved user experience. Fully powered by Targon, Sybil is fast, intuitive, and AI-native by design.

Tao.xyz + Extension – Live Now

Tao.xyz is a focused dashboard for anyone active in Bittensor. Think of it as the Bloomberg Terminal for subnet operators, validators, miners—and all TAO holders. It filters out noise and surfaces the signal that matters most.

To accompany Tao.xyz, we’re launching a lightweight browser extension. It allows seamless switching between subnet validators and makes navigating the Bittensor ecosystem effortless.

Targon.com Refresh + V6 Live on SN4 – Live Now

The Targon Virtual Machine (TVM) establishes a secure, confidential computing framework specifically tailored for AI workloads, providing robust hardware and GPU attestation through integration with NVIDIA's nvTrust SDK. By ensuring confidentiality and integrity, TVM enables trustworthy monetization of AI models, fostering secure, decentralized AI marketplaces.

Looking Ahead

We believe we have a small window of opportunity to build the world’s first decentralized frontier AI lab. That means building a full-stack platform: decentralized compute, production-grade research, and premium user-facing products. Doing so will allow us to solve technology’s most difficult challenges, and if successful, we have the potential to play a significant role in ushering in an AGI utopia for all of humanity. As always, thank you for your continued support. We are incredibly excited for Manifold 2.0, as we think bigger, and usher Bittensor into the mainstream.