Levitation Memory (RAM)

White Horse is a PCIe co-processor card that expands your system's effective memory to 120GB and beyond — without replacing your motherboard, your RAM, or your patience.

Slide it into a standard PCIe slot and your system's effective memory expands immediately. Minimum 120GB on a 8GB system. - (Our claims under estimate and over deliver.)

How It Works

White Horse runs AO8's proprietary Levitation algorithm on dedicated RISC-V processors — The algorithm encodes your data into high-dimensional Levitation signatures that achieve a 30:1+ emergence ratio. That means 8GB of physical RAM operates like 32GB.

But White Horse doesn't just process data on the card. It recruits your existing hardware.

Your CPU becomes a Levitation engine. The AO8 Levitation Manager runs on your system's processor alongside the card's dedicated RISC-V engines. Your CPU handles overflow, background indexing, and predictive data staging — work that would otherwise require dedicated silicon.

Your RAM becomes shared Levitation workspace. The card accesses your system memory over the PCIe bus via Direct Memory Access. Your existing RAM becomes an extended index and staging ground for Levitation signatures. More system RAM means a bigger index. A bigger index means faster access. Your hardware investment compounds.

Your storage becomes the persistent cache. There's nothing stored on the card that can be lost. All persistent Levitation data lives on your own drive, managed by the Levitation Manager. Pull the card out and your data is still yours, on your machine, exactly where you left it.

Every component in your system contributes to Levitation performance. Nothing is wasted.

The RISC-V Advantage

White Horse is built on RISC-V — We chose RISC-V because the instruction set includes native vector extensions that align directly with the mathematical operations at the core of Levitation encoding. It's not a workaround. It's an architectural match.

What You Get

• Effective Memory: 120GB+ minimum (scales with your system)
• Processor: Dedicated RISC-V Levitation engines with vector extensions
• Interface: PCIe 4.0 x4 (backward compatible with Gen 3.0 and 2.0)
• Cooperative Processing: Card + your CPU + your RAM work together as a unified Levitation system
• Power Draw: Under 5W typical — passive heatsink, no fan, no external power cable
• Operating Range: 0–70°C
• Form Factor: Half-height and full-height brackets included — fits any chassis
• Data Safety: Nothing stored on the card. All data persists on your system drive. Card failure = zero data loss.
• Price: TBA

Cooperative Processing — Your System, Amplified

Traditional add-in cards are islands. They sit in your PCIe slot, do their job, and ignore the rest of your system. White Horse is different.

The AO8 Levitation Manager software turns your entire system into a Levitation processing pool. The card's RISC-V engines handle the real-time hot path — encoding and decoding data at hardware speed. Your system's CPU handles the deep work — background indexing, predictive warming, and overflow processing. Your RAM holds the extended Levitation index so the card can find what it needs faster.

The more powerful your system, the more White Horse can do. A gaming PC with 32GB of RAM gives the card a massive index to work with. A workstation with 64GB+ turns White Horse into a terabyte-scale memory accelerator. Same card. Same price. Your hardware determines the ceiling.

This is why White Horse is a co-processor, not a storage device. It doesn't replace your RAM. It makes your entire system better at using it.

Who It's For

Gamers. Textures, shaders, level geometry, physics data — modern games want more RAM than you have. White Horse pushes the wall back far enough that your system stops being the bottleneck. Load times shrink. Stuttering disappears. Your existing rig plays like a machine that costs three times as much.

AI and Machine Learning. Model inference, knowledge retrieval, and training pipelines hit the RAM wall constantly. White Horse delivers effective memory expansion at a fraction of the cost of additional physical RAM — and the Levitation algorithm's binary processing architecture moves signature data 257x faster than text-serialized alternatives.

Scientific Computing and Simulation. Molecular simulation, signal processing, physics modeling, and any workload operating on dense floating-point arrays. Levitation encoding operates in the same mathematical domain as the data it processes. The card doesn't approximate your computation — it speaks the same language.

Enterprise and Virtualization. Memory-constrained VMs, database servers hitting swap, and infrastructure teams tired of waiting for DDR5 prices to come back to earth. One PCIe card per server. Immediate capacity expansion. No downtime.

Anyone hitting the RAM wall. If your systems are memory-constrained and you can't wait two years for the supply chain to recover, White Horse is the answer that ships now.

The card you buy today becomes the platform for tomorrow's breakthrough. Your investment is protected.

Compare our products

© 2026 AO8 Inc. All rights reserved — Privacy Policy and Legal