OUR CUSTOMERS
Cloud Gaming Platforms

The only way to profitable cloud gaming
ASICs deliver lowest cost per concurrent user (CCU) in the industry.
Cloud gamers compete on almost any internet-connected device via a subscription service. With monthly subscription rates as low as $4.99, margins are razor-thin, and profitability requires the lowest possible capital and operating costs.
In cloud gaming, the cost per concurrent user (CCU) is the key metric, and ASICs enable the lowest possible CCU.
Choosing the right combination of hardware for graphics rendering and video encoding is essential. With an ASIC-based encoder like the Quadra T2 VPU coupled with a third-party GPU, a single server can deliver up to 200 simultaneous 720p60 gameplay sessions.
* SOURCE: bit.ly/ASIC_CloudGaming
Cloud gaming is expected to be the future of the gaming industry.
Market Growth Factors:
5G technology’s low-latency capability will drive global cloud gaming growth.
Market Restraining Factors:
Latency and responsiveness issues will hinder market potential.
SOURCE: www.fortunebusinessinsights.com
Cloud gaming is expected to be the future of the gaming industry.
Market Growth Factors:
5G technology’s low-latency capability will drive global cloud gaming growth.
Market Restraining Factors:
Latency and responsiveness issues will hinder market potential.
SOURCE: www.fortunebusinessinsights.com
Head to head comparison
ASICs for the win!
ASIC-based encoding makes immersive cloud gaming experiences economically feasible. As shown on the right, combining an ASIC-based encoder with GPU-based rendering delivers the lowest CCU available.
Data based on 720p output with maximum power consumption.
SOURCE: bit.ly/ASIC_CloudGaming

Smart VPU
We created a new category of Smart VPUs to disrupt the previous limitations of video encoding for streaming delivery platforms. It is unique because it’s dense, cost effective and AI-powered making it the smart technology for platforms to future-proof their services and hyperscale profitably.
200 gamers at once
Cost per concurrent user (CCU) is the key metric for cloud gaming
When you encode using a Smart VPU, you decrease your CCU by 90%, enabling profitability at a much lower subscription price. How is this technically feasible? First, Quadra VPUs are extraordinarily efficient encoding devices solely dedicated to high-quality video encoding.
Second, in cloud gaming workflows, NETINT implements peer-to-peer direct memory access between the GPU and VPU. This speeds the transfer of rendered frames from GPU to VPU, reduces latency, and increases overall system performance.

The secret sauce to minimizing latency
Peer-to-peer DMA relieves the CPU from managing GPU to VPU data transfers.
Cloud gaming involves high-quality GPU frame rendering and high-bandwidth transfers to the VPU for encoding. Without efficient transfers between the two devices, processing bottlenecks can occur. To avoid this, in cloud-gaming architectures, NETINT deploys peer-to-peer direct memory access (P2P DMA) between GPU and VPU.
Briefly, P2P DMA enables the GPU to VPU transfer without involving the host CPU. This accelerates the transfer, reduces latency, and improves overall system performance, resulting in a CCU that cannot be matched by any competing architecture.
Additionally, edge servers reduce latency.
The strategic placement of servers at the edge of the network, and within the carrier’s infrastructure, helps ensure low latency performance.


High density and low power in one
10 VPUs in one server is the functional equivalent of
10-20 high-end, CPU-based encoding servers*.
The Quadra T2 consumes only 40 watts, which is 5-10% of the consumption of a server encoding with software. Yet the T2 delivers the same or higher throughput.
* Depending on the host server configuration
Public cloud is great for proof of concept
Undisputedly, the fastest go-to-market path is to deploy in the public cloud. This enables gaming services to test their games, collect various metrics, and gather usage data. However, the public cloud quickly becomes an unsustainable operating expense. Ultimately, you need to build your own infrastructure with ASICs to rein in OPEX costs and turn a profit.

ASIC Benefits
Highest Density
Lowest Cost
Ultra Low Latency
720p30 - 4Kp60
Encode up to 4Kp60.
AI Engine
On-board AI rendering supports advanced gaming features.
Scalable
Smart VPUs for Cloud Gaming
Smart VPUs are engineered for live streaming, interactive gaming, and the metaverse, with AI-powered features enabling a whole new world of online gaming.
Here’s a glance at NETINT’s line of Smart VPU products for gaming:
![]() | ![]() | ![]() | ![]() | |
---|---|---|---|---|
2nd Generation Smart VPUs | ||||
Server | Modules | |||
G5 Smart VPU
Quadra | Smart VPU
T1U | Smart VPU
T1A | Smart VPU
T2A | |
Performance | ||||
ASIC Codensity chip | G5, T1Us (10x) | G5 | G5 | G5 (2x) |
Price | starting at $19,000 | $1,500 | $1,500 | $2,750 |
Form Factor | 1RU Server | U.2 | AIC, HHHL | AIC, HHHL |
Power Consumption | ~500W | 17W | 20W | 40W |
Real-time Throughput Up to: | 320x 1080p30 80x 4Kp30 20x 8Kp30 | 32x 1080p30 8x 4Kp30 2x 8Kp30 | 32x 1080p30 8x 4Kp30 2x 8Kp30 | 64x 1080p30 16x 4Kp30 4x 8Kp30 |
Latency | 8 ms | 8 ms | 8 ms | 8 ms |
Encode Codecs | H.264, HEVC, JPEG, YUV, AV1 | |||
Decode Codecs | H.264, HEVC, JPEG, YUV, VP9 | |||
Audio Codecs | MP3, AAC-LC, HE-AAC | |||
Features | ||||
Artificial Intelligence | 150 TOPS | 15 TOPS | 18 TOPS | 36 TOPS |
New Capped CRF | ||||
Peer-to-peer DMA | ||||
Cropping and Padding | ||||
Scaling | ||||
Video Overlay | ||||
Audio decode / encode | ||||
YUV / RGB Conversion | ||||
Configurable tuning of
quality/throughput |
Feature supported on VPU
Feature runs on host CPU

Reference Cloud Gaming Video Server
This Supermicro 2RU chassis is loaded with Quadra T2A video processing units giving it 40x processing power over a software based comparable CPU.
Simple architecture

Our Customers
NETINT ASIC VPUs lead the streaming technology in these markets also: