Updated 2026-08-20: the figures in this scenario have been re-derived from Sephora's live catalogue and itemised. The previous version stated ₦3.8 million per render node against a real ₦15,004,658 — 3.95x — with no component breakdown to check it against. If you scoped a budget from this page before this date, it was understated by that margin.
The Scenario
Consider a 3D animation studio in Victoria Island, Lagos, producing animated content for children's television, branded explainer content, and increasingly, short-form animated segments for streaming platforms. A team of seven: three animators, two riggers/technical artists, a compositor, and a studio manager, producing four to six minutes of finished 3D animation per month — a respectable output for a team of that size.
The bottleneck is the render farm. Or more precisely, the thing being called a render farm, which is actually two gaming PCs and one workstation running Blender's Cycles renderer overnight and on weekends. Those machines produce the rendered frames that become the finished content, and they produce them far too slowly to support the studio's growth ambitions.
The Challenge
3D animation rendering is one of the most computationally intensive workloads that exists in commercial creative work. A single frame of a finished 3D scene in Cycles — a path-traced renderer — can take anywhere from 2 minutes to 4 hours, depending on scene complexity, lighting, resolution, and render settings. Delivering at 1080p with full global illumination and subsurface scattering is a cinematic quality standard that requires real compute to achieve in reasonable time.
An ad-hoc "farm" of two RTX 3070 gaming PCs and one RTX 3080 workstation might collectively render around 45 frames per hour on complex scenes. A one-minute animation at 24fps is 1,440 frames. That puts the most complex sequences at over 32 hours of farm time per minute of finished animation — a 90-second branded animation occupying the farm for roughly 48 hours. Every weekend goes to rendering what was animated during the week.
The problem compounds when client revision requests arrive. A lighting change on a 90-second animation means re-rendering affected scenes — potentially another 24–48 hours. Meeting tight revision turnaround times becomes impossible.
The Assessment
The conversation to have is with whoever handles the pipeline — the studio's technical director — about Blender workflow, typical scene complexity, and growth targets. A studio wanting to double output in 12 months, to eight or twelve minutes of finished content per month, cannot do it on that render capacity without working around the clock. The arithmetic decides this, not opinion.
What it calls for is a dedicated GPU render farm: nodes purpose-built for Blender Cycles rendering, with high-VRAM NVIDIA GPUs (Cycles is CUDA-optimised), no unnecessary components (no display output, no gaming peripherals, minimal CPU — the GPUs do all the work), managed by a render farm software package. Flamenco, Blender's own open-source farm manager, is the natural fit.
Power and cooling need designing at the same time, not afterwards. Running multiple high-TDP GPUs in a Lagos office requires thinking carefully about the electrical load, the heat output, and what happens when NEPA cuts power mid-render.
The Build
4 Render Nodes — ₦15,004,658 each
- GPU: 2× RTX 4090 24GB per node (₦5,405,068 each, ₦10,810,136 the pair) — Cycles renders to VRAM first; 24GB per GPU handles extremely complex scenes
- CPU: Intel Core i5-13400 (₦446,142) — minimal, just enough to manage the render process; the GPUs do 95% of the work
- Motherboard: B760 ATX DDR5 (₦457,363)
- RAM: 64GB DDR5-5600 (₦1,560,006) — scene data is loaded to RAM before dispatch to the GPUs
- Storage: 512GB PCIe 4.0 NVMe (₦125,991) for OS and Blender, plus a 10GbE NIC for receiving jobs from the studio NAS — the NIC is specified per deployment rather than priced here
- Chassis and cooling: our largest case is an ATX tower (₦350,010) with a premium air cooler (₦155,000). The 4U rackmount chassis and high-static-pressure server fans this scenario describes are not catalogue parts; a rack-mounted farm is a sourced-to-order build.
- PSU: 1200W 80+ Platinum (₦1,100,010) — and this is the honest constraint. Two RTX 4090s draw 900W between them before the CPU, which puts a dual-flagship node at the very top of what our range supports. A farm built this way is a consultation, not a configurator selection.
What the Farm Actually Costs
Four nodes at ₦15,004,658 come to ₦60,018,632, before the 40kVA UPS, the 10GbE switch and structured cabling, and the rack room itself — none of which we supply, and all of which the article previously folded into a single indicative figure.
An earlier version of this page put each node at ₦3.8 million. The real figure is close to four times that, and almost all of the gap is the eight RTX 4090s: at ₦5,405,068 each they are ₦43,240,544 of the farm on their own, or 72% of the hardware. That is the number a studio should actually be deciding about. If it is too large, the productive question is not which cheaper card to substitute but whether fewer nodes with the same GPUs, or renting cloud render capacity for peak weeks, fits the studio's actual load better than owning eight flagship cards.
Infrastructure:
- 40kVA online UPS for the entire farm — keeps rendering through power outages up to 45 minutes
- 10GbE switch and structured cabling between farm nodes and studio NAS
- Flamenco render farm software configured and tested
- Temperature monitoring and automatic shutdown triggers if the rack room exceeds 38°C
What Changes
Cycles render throughput scales with GPU count and GPU capability together, and both move here. Eight RTX 4090s in place of three consumer cards is not a percentage improvement — it is a different order of capacity, and it collapses the render window for a 90-second branded animation from something that eats a weekend to something that fits inside a working day.
That changes what the studio can promise. Revision turnaround is the commercial product of render speed: when a lighting change no longer means another 24–48 hours of farm time, tight revision windows stop being a reason to decline work. Render capacity is what sets the ceiling on how many projects a studio of seven people can accept.
The 40kVA online UPS protects all of it. A NEPA cut partway through a long render is wasted GPU hours, and on a farm this size those hours are the asset.
Key Takeaway
For animation studios, render capacity is revenue capacity. A render farm that keeps up with your artists enables you to take on more projects, meet tighter deadlines, and offer revision turnaround times that command premium rates. A render farm that can't keep up becomes the ceiling on your business growth. The economics of a proper GPU render farm — particularly with current-generation RTX GPUs — have shifted dramatically in the last two years. The investment pays back in billable capacity faster than most studios expect.
Running an animation or VFX studio that needs more render capacity? Talk to our team about a render farm consultation. We design, build, and install.