This episode unpacks a critical BitTensor network disruption and recovery, underscoring the drive towards decentralization, before pivoting to an in-depth look at VIDAIO (Subnet 85), a project aiming to revolutionize video processing with decentralized AI-driven upscaling and compression.
BitTensor Network Disruption: A Stress Test and Catalyst for Decentralization
- The episode began with F (likely Jacob Steeves/Const) recounting a recent critical incident where the BitTensor chain halted, ceasing to finalize blocks. This occurred due to validator nodes being unable to reach consensus.
- Substrate: A modular framework for building blockchains, which BitTensor utilizes. Out-of-the-box Substrate chains often start as Proof of Authority (PoA), a system where a select group of pre-authorized nodes validate transactions, offering a degree of decentralization but still centralized control.
- F emphasized that despite such "extremely traumatic events," building on a blockchain like Substrate provides a "consistent verifiable history," ensuring transparency even when emergency measures are taken.
- The root cause was a vulnerability allowing the creation of transactions that led to a loss of consensus. An anonymous individual, after requesting a bounty, provided information that helped identify and patch the issue.
- F stated, "We were very skeptical at first... but after speaking with this individual... we realized that they actually had some legitimate information for us."
- The fix involved a deep patch in the Substrate codebase and a manual "soft fork" to bring the network back into consensus.
- Soft Fork: A type of blockchain upgrade where only previously valid transaction rules are made invalid, maintaining backward compatibility.
- F highlighted the dedication of developers, specifically VO from Leighton and JIP from F, who worked tirelessly for 48 hours to resolve the crisis.
- Strategic Implication: This event accelerates BitTensor's planned transition from its current 20 F-controlled PoA validator nodes to a more distributed set of organizational validators, and ultimately to a full Proof of Stake (PoS) system. PoS is a consensus mechanism where validators are chosen based on the number of coins they "stake" or lock up, offering greater decentralization and security.
- Investors and researchers should monitor this transition closely, as increased decentralization can enhance network resilience, reduce single points of failure, and potentially impact tokenomics and governance.
Introducing VIDAIO (Subnet 85): Decentralized AI Video Processing
- Alla (likely Ala Shaabana/Mog Machine) introduced Gareth and the VIDAIO team, a new subnet on BitTensor (SN85) focused on AI-driven video upscaling and compression.
- Alla, with his business development acumen, noted his initial surprise and subsequent strong impression of VIDAIO's business plan, deciding it was "definitely worth it for the benefit of the network."
- Gareth, from VIDAIO, set the stage by highlighting a critical market pain point: "85% of traffic on the internet is video based," and every piece of this content requires processing, leading to spiraling costs for streaming platforms and content owners due to delivery and storage.
- CDN (Content Delivery Network): A geographically distributed network of proxy servers and their data centers, which work together to provide fast delivery of Internet content.
The Problem: Spiraling Video Costs and Inefficient Processing
- Gareth explained that content platforms face uncontrolled delivery and storage costs as more video content is continuously added to CDNs and storage systems.
- The current centralized video processing industry is often "closed off, slow to innovate, and really expensive to use."
- VIDAIO, launched on April 17th, aims to address this by offering AI-driven video upscaling (enhancing low-resolution content) and compression (reducing file size without quality loss).
- Actionable Insight: The massive scale of video content and the associated costs create a significant market opportunity for decentralized solutions like VIDAIO that promise efficiency and cost reduction.
VIDAIO's Team and Addressable Market
- Gareth introduced the VIDAIO team:
- Gareth: Product and operations background, with experience at Netflix, Disney, Sony.
- Ahmed: Machine Learning Engineer, focusing on the compression model.
- Mog Machine (Alla): Providing incubation and business guidance.
- Gopy: Video technology expert (not present).
- Medville: Subnet and incentive mechanism architect.
- Akin and Chinaza: Front-end and product designers.
- The addressable market is estimated at $400 billion, with initial targets being video streaming ($150B) and cloud storage/CDN providers ($150B). Other potential areas include post-production, surveillance, and medical imaging.
- Strategic Implication: VIDAIO's diverse team expertise and focus on a large, growing market indicate a serious attempt to build a commercially viable decentralized service.
Deep Dive: Video Upscaling on VIDAIO
- Gareth detailed the current live feature: video upscaling (still in beta). The workflow involves sending low-resolution files to miners, who upscale them.
- Verification uses:
- VMAF (Video Multi-Method Assessment Fusion): A Netflix-developed perceptual video quality assessment algorithm.
- PIAQ (Perceptual Image Quality Assessment): Another metric used to score miners, as VMAF alone was found to be gameable.
- For organic uploads without a reference, CLIP IQA+ is used, focusing on perceptual quality as perceived by the human eye, trained on large datasets.
- F clarified that upscaling, like going from 4K to 8K, involves generative AI models "filling in the gaps" by predicting correct pixel values.
- Mog provided a real-world example: upscaling old SD skate movies for inclusion in modern HD/4K documentaries, making previously unusable footage watchable by analyzing pixels and "creating new pixels" based on perceptual quality.
- Benchmark Claims: Gareth presented benchmarks showing VIDAIO's base miner model for upscaling performing slightly better on their chosen perceptual visual quality metrics (0.4697 score) compared to competitors like Topaz, FFmpeg, and HeadPaw. He noted miners have already improved upon this base model.
- Actionable Insight: If VIDAIO's decentralized upscaling can consistently outperform or match centralized competitors on quality and speed (currently comparable for offline processing), it could attract niche markets like archive restoration or user-generated content enhancement.
- Privacy and Content Moderation: The discussion touched upon privacy, with videos currently broken into one-second chunks for miners. Mog confirmed that content filters for inappropriate material (e.g., CSM, nudity) will be implemented, emphasizing the need to protect the network's reputation. For sensitive, NDA-bound content, future solutions might involve trusted B2B relationships with specific miners.
Groundbreaking Video Compression Model
- Ahmed, VIDAIO's ML engineer, showcased their proof-of-concept video compression model, which he described as "completely groundbreaking."
- The core idea is to move beyond one-size-fits-all encoders (like standard H.264, a widely used video compression standard) by using ML to analyze video content (scenes, complexity, motion, contrast) and predict optimal encoder parameters for codecs like H.264, HEVC (High Efficiency Video Coding), or AV1 (AOMedia Video 1).
- Ahmed explained, "Our model will take all of these information and then it will give us the best encoder parameters that we can apply."
- The process involves pre-processing to remove artifacts, splitting video into scenes, analyzing scenes for metrics, classifying frame types, and then using an ML model to determine the best encoding parameters for widely supported encoders.
- F questioned why not use a neural network for the entire end-to-end encoding/decoding. Ahmed cited compatibility issues (requiring all users to have the specific decoder model) and the vast data needed to train such a model to outperform current, highly optimized traditional encoders.
- Demonstration: Ahmed showed the app compressing an ~800MB uncompressed video to ~40MB (a 4.8% compression rate of the original size) while maintaining high VMAF scores, significantly reducing bitrate.
- Actionable Insight: VIDAIO's intelligent, content-aware compression, if scalable and effective, could offer substantial cost savings in storage and bandwidth for content platforms, representing a major disruptive potential. Investors should track the development and real-world performance of this compression technology.
Miner Competition and Subnet Performance
- The current live upscaling synapse sees miners competing to improve perceptual quality scores (VMAF, PIAQ). Mog noted a "week-on-week performance increase" as miners optimize beyond the base model.
- Gareth confirmed miners are "tweaking the models," using open-source models and other techniques, while VIDAIO actively checks against gaming the system.
- F remarked on the "classic phenomenon of subnets on BitTensor where you're like, 'So, how did you achieve this?' And the owners are like, 'We have no idea.'"
- For the upcoming compression synapse, competition dimensions will likely include perceived quality, compression ratio, and speed/compute power.
- Strategic Implication: The ability of the BitTensor incentive mechanism to drive rapid, emergent optimization by miners is a core strength. For VIDAIO, this means their platform's capabilities can evolve faster than a centralized R&D team might achieve alone.
Business Model, Revenue, and the Role of Alpha
- Mog, speaking from his experience incubating VIDAIO via Talstats, emphasized a "business first approach." The subnet is currently funded but aims for self-sustainability within 90-100 days through revenue generation.
- The priority is securing a solid customer base through reliable services before focusing heavily on revenue. Miners are viewed as crucial "contracting suppliers."
- Mog believes "minor reputation is going to be paramount moving forward," beyond just Alpha (a BitTensor-specific term likely referring to stake, influence, or rewards share). He envisions a system where top-performing, trusted miners could handle sensitive B2B contracts, even those requiring NDAs.
- Mog stated, "You need to not be afraid to foster business-to-business relationships."
- He stressed that subnets "cannot be solely emissions dependent" for long-term viability.
- Actionable Insight: Investors should look for subnets with clear strategies for achieving revenue and self-sustainability beyond network emissions. VIDAIO's focus on B2B enterprise clients and a tiered approach to miner engagement for sensitive data is a pragmatic strategy.
Inter-Subnet Collaborations and Ecosystem Synergies
- VIDAIO plans collaborations with other BitTensor subnets:
- Hippius (SN21): For decentralized storage.
- Bitmind (likely referring to an AI content detection subnet): For identifying AI-generated content.
- Soundsight (SN105): For audio restoration, creating a combined video/audio enhancement product.
- Bitcast: content creation marketplace, for promotional content.
- Potential for Score (SN44, computer vision): For metadata tagging.
- Mog likened this to an automotive company outsourcing gearbox manufacturing to specialists, stating, "Why not go to the experts in that?"
- Gareth added, "There's so much value in the community... we want to keep it in within it as much as possible."
- Strategic Implication: Composability and collaboration between specialized subnets can create powerful network effects within the BitTensor ecosystem, offering more comprehensive solutions than standalone projects. This is a key differentiator for decentralized AI platforms.
Go-To-Market Strategy and Target Customers
- Gareth outlined VIDAIO's go-to-market strategy, initially targeting emerging markets, particularly in Africa, where bandwidth and connectivity issues make efficient compression highly valuable.
- Gareth, drawing on his 10 years of experience in VOD in Africa, said, "If we can make these files 50, 60, 70, 80% smaller and still maintain the quality, then that's going to be a huge win."
- The primary customers will be large enterprise clients: content owners, CDN providers, VOD platforms – anyone storing and streaming large volumes of video.
- Mog clarified that while consumers might use VIDAIO for occasional upscaling/compression, the "real value will come from business-to-business clients."
- Gareth mentioned engaging with Jan Ozer, a respected figure in streaming and video compression, for independent benchmarking and validation once the compression model is ready.
- Actionable Insight: Targeting enterprise clients in specific geographies with acute needs (like emerging markets) is a smart entry strategy. External validation from industry experts will be crucial for gaining credibility.
Future Roadmap: On-Demand Architecture and Live Streaming
- VIDAIO's future roadmap includes:
- Video compression synapse launch (estimated ~1 month for good progress).
- Transcoding optimization.
- Longer-term: On-demand video architecture and live streaming through the subnet. This would require decentralized storage, edge servers, and a decentralized CDN.
- Gareth believes a "decentralized streaming service that runs on the BitTensor network, it would be incredible."
- Strategic Implication: Successfully developing decentralized live streaming would be a significant technical achievement and could open up vast new markets, further solidifying BitTensor's utility in media processing.
Conclusion
This episode highlights BitTensor's resilience and commitment to decentralization, alongside VIDAIO's ambitious plan to disrupt video processing using decentralized AI. For investors and researchers, VIDAIO's progress in AI-driven compression and upscaling, coupled with its B2B focus and inter-subnet strategy, presents a compelling case study in monetizing decentralized compute.
Key Takeaways for Crypto AI Investors/Researchers:
- Monitor BitTensor's PoS transition for network stability and governance implications.
- Evaluate VIDAIO's compression technology benchmarks and enterprise adoption as key indicators of its disruptive potential.
- Track the development of inter-subnet collaborations as a measure of the BitTensor ecosystem's maturity and network effects.
