Releases·Model Release
TBC's Rat-Brain-Derived AI Model for Video Generation Enters Limited AWS Preview
The Biological Computing Company (TBC), a Baltimore startup that derives compute patterns from rat neurons, launches a limited AWS preview of its video-generation AI starting Tuesday. The company claims its model runs up to five times faster than the open-source frontier video model it builds on, and has raised over $50 million, including an unreported second $25 million round. Amazon's backing signals growing cloud interest in biological computing, though scaling fidelity to longer videos remains unproven.
0 engagement1 sourceTue, Sep 22, 2026, 01:00 PM
AI Overview
- TBC, a Baltimore startup, enters a limited AWS preview Tuesday for its biological AI model, claiming up to 5x faster video generation than the open-source model it runs on
- TBC raised over $50 million total this year, including two $25 million rounds led by Primary Venture Partners, plus an unreported second round in March
- TBC trains rat neurons and human stem cells on 3Brain multi-electrode silicon arrays, translating neural responses into software for video models on Jeff Dean's advice
- Amazon also distributes Cortical Labs' 'wetware as a service,' which pairs lab-grown neurons with silicon chips and sells a biological computer keeping neurons alive six months
- TBC cofounder Alexander Ksendzovsky says the company codes images onto biological material, observes how neurons process them, then mimics that process in software
- Amazon's Deap Ubhi questions whether the technology will scale, asking about fidelity losses in longer-form videos of ten minutes or an hour
Entities
Related
News·Company News
Recursive Superintelligence signs $410M compute deal with AWS
Recursive Superintelligence announced a $410 million compute deal with Amazon Web Services to support its open-ended self-improving AI systems. The company emerged from stealth in May with $650 million in funding, and this multi-year deal represents the bulk of its fundraising to date.
Recursive Superintelligence announced a $410 million compute deal with Amazon Web Services to support its open-ended self-improving AI systems. The company emerged from stealth in May with $650 million in funding, and this multi-year deal represents the bulk of its fundraising to date.
0 engagement1 sourcerssTue, Jul 28, 2026, 01:19 PM
Releases·Tool Release
Amazon launches SageMaker HyperPod Inference Gateway, a GPU-aware Kubernetes router cutting first-token latency up to 82%
Amazon announced SageMaker HyperPod Inference Gateway, a Kubernetes-native addon that routes LLM inference requests using real-time GPU signals instead of round-robin load balancing. Benchmarks show P95 first-token latency cuts up to 98% and throughput gains up to 50% under mixed or bursty workloads. It deploys as a single EKS addon with no application or SDK changes.
Amazon announced SageMaker HyperPod Inference Gateway, a Kubernetes-native addon that routes LLM inference requests using real-time GPU signals instead of round-robin load balancing. Benchmarks show P95 first-token latency cuts up to 98% and throughput gains up to 50% under mixed or bursty workloads. It deploys as a single EKS addon with no application or SDK changes.
0 engagement1 sourcerssFri, Sep 18, 2026, 01:08 PM
News·Company News
Nvidia expands AWS GPU deal to 2M chips and invests $3.5B in MediaTek amid custom-silicon push
Amazon and Nvidia announced an expanded partnership adding 2 million Nvidia GPUs — Blackwell Ultra, Rubin, and Rubin Ultra — to AWS data centers in 2027 and 2028, tripling an order made five months earlier. Days later, Nvidia committed $3.5 billion to MediaTek, which will adopt Nvidia technology to design custom AI chips for hyperscalers that plug into Nvidia-based data centers. Together the moves show Nvidia locking in hyperscaler demand while positioning itself inside the custom-silicon buildout by Amazon, Google, Microsoft, OpenAI, and Anthropic.
Amazon and Nvidia announced an expanded partnership adding 2 million Nvidia GPUs — Blackwell Ultra, Rubin, and Rubin Ultra — to AWS data centers in 2027 and 2028, tripling an order made five months earlier. Days later, Nvidia committed $3.5 billion to MediaTek, which will adopt Nvidia technology to design custom AI chips for hyperscalers that plug into Nvidia-based data centers. Together the moves show Nvidia locking in hyperscaler demand while positioning itself inside the custom-silicon buildout by Amazon, Google, Microsoft, OpenAI, and Anthropic.
Updated Mon, Aug 31, 2026, 03:15 PM — Nvidia invests $3.5B in MediaTek to design custom AI chips for hyperscalers compatible with Nvidia data centers.
0 engagement2 sourcesrssWed, Aug 26, 2026, 11:47 PM
Products·launched
BrainCo: AI platform for brain-to-robot control
BrainCo launched an integrated AI platform that enables direct brain-to-robot control using EEG signals and machine learning. The platform translates neural activity into commands for robotic devices, aiming to assist people with paralysis or motor impairments. It uses LLMs for natural language interpretation of user intent alongside brain signals.
BrainCo launched an integrated AI platform that enables direct brain-to-robot control using EEG signals and machine learning. The platform translates neural activity into commands for robotic devices, aiming to assist people with paralysis or motor impairments. It uses LLMs for natural language interpretation of user intent alongside brain signals.
53 engagement1 sourceredditMon, Jul 20, 2026, 02:48 PM
Releases·Model Release
Accelerated Understanding Inc launches new AI model that ditches transformers for neural operators
Accelerated Understanding Inc has launched a new AI model that replaces transformer architectures with neural operators, a shift that could impact efficiency and scalability in AI systems. The move is notable for practitioners exploring alternatives to attention-based models.
Accelerated Understanding Inc has launched a new AI model that replaces transformer architectures with neural operators, a shift that could impact efficiency and scalability in AI systems. The move is notable for practitioners exploring alternatives to attention-based models.
76 engagement1 sourceredditWed, Aug 26, 2026, 02:08 AM