Buzzwords like ‘high-capacity’ and ‘low-latency’ have become meaningless — so what now?
Beyond Buzzwords: Redefining Network Capacity Terms
In the world of telco networking, buzzwords like “high-capacity” and “low-latency” have become ubiquitous marketing terms, often divorced from their practical implications. As we stand on the brink of a data revolution, it’s time to challenge our industry’s shared definition of network performance. The metrics that once defined network excellence are not just becoming obsolete — they’re actively hindering our ability to prepare for the future of data processing and transmission.
The (Realtime) Data Tsunami: A New Reality
Before we delve into why certain over-used terms are falling short, let’s consider the scale of the challenge we’re facing. According to IDC, global data volume is set to explode to a staggering 175 zettabytes by 2025, with global data doubling every 2–3 years. 
Even more astounding, IDC predicts that 90 zettabytes of this data will come from IoT devices alone. This isn’t just a quantitative shift — it’s a fundamental change in how data is generated, processed, and consumed.
Forbes adds another layer to this prediction, estimating that 150 zettabytes of real-time data will need analysis by 2025. This emphasis on real-time processing has profound implications for industries like VFX, AR, and HFT, where split-second decisions and interactions can make or break projects, live experiences, and trading strategies.
The Illusion of Speed: Why Traditional Metrics Fail
As we face this impending data tsunami, it’s crucial to reconsider what we mean by “high-capacity” networks. The traditional metrics we’ve relied on are increasingly inadequate in the face of these new challenges.
When you hear about 100G or 400G networks, what comes to mind? Blazing speed? Unlimited potential? The reality is far more nuanced. These figures refer to port speeds, not actual service levels. It’s like describing highway performance by the maximum speed limit, ignoring factors like traffic, road conditions, and the number of lanes.
Real-world example: A leading VFX studio working on a blockbuster film found that their “high-capacity” 100G network was bottlenecking during peak rendering hours. Despite the impressive port speed, they were achieving less than 15% of the theoretical maximum throughput due to network congestion and inconsistent latency. This resulted in missed deadlines and increased production costs, ultimately impacting the film’s post-production schedule and budget.
Redefining Standards for the AI Era: Beyond High-Capacity and Low-Latency
As we enter a technology phase dominated by distributed AI and machine learning, our network needs are fundamentally changing. AI workloads require not just high bandwidth, but consistent, low-latency performance across distributed systems. This new reality demands a complete reassessment of what we mean by “high-capacity” and “low-latency.”
At Stelia, we’re pushing the boundaries of what’s possible with our AI Availability Zones. Here, network performance is defined not by port speeds, but by practical, real-world metrics:
- Maximum Round Trip Time (RTT) of 0.1ms
- Minimum of 120Tbps of redundant on-platform interconnectivity between any two locations
To put this in perspective, a 0.1ms RTT is about 50 times faster than a blink of an eye. This level of responsiveness is crucial for applications like:
- Real-time collaborative VFX work across global teams
- Seamless, lag-free AR experiences in large-scale environments
- Ultra-low latency execution in high-frequency trading algorithms
The 120Tbps interconnectivity isn’t just a big number — it’s a new paradigm. It allows for the seamless transfer of massive datasets, enabling:
- Instantaneous sharing of high-resolution VFX assets between studios
- Real-time processing of complex AR environments with millions of data points
- Simultaneous analysis of global financial markets for HFT firms
To illustrate, this capacity could transfer an entire feature film’s worth of 8K raw footage (about 500TB) in under 35 seconds.
But it’s not just about raw speed. True “high-capacity” and “low-latency” in modern networks must also consider:
- Consistency: Fluctuations in performance can be as problematic as high latency or low bandwidth.
- Scalability: Maintaining performance as data volumes and network complexity increase.
- End-to-end performance: Ensuring high capacity and low latency across the entire data path, not just in isolated network segments.
As we move into the zettabyte era, with 150 zettabytes of real-time data needing analysis by 2025, our definitions must evolve. In the petabit era, truly low latency will be measured not in milliseconds, but in microseconds and nanoseconds.
The Future is Measured in Petabits
By 2025, we’re aiming to make 1 Petabit per second (Pbps) the new standard for high-capacity networks. This isn’t just an incremental improvement — it’s a quantum leap necessary to handle the 175 zettabytes of global data and 150 ZB of real-time analysis predicted by IDC and Forbes.
To put this in perspective, 1 Pbps could:
- Render a feature-length 3D animated film in real-time
- Support millions of simultaneous high-fidelity AR experiences
- Process market data from every global exchange simultaneously with zero lag for HFT applications
By 2025, we project our network aggregate to surpass 100 Pbps. This capacity will enable a new era of innovation:
- VFX studios could collaborate on global projects as if they were in the same room, handling the massive datasets required for next-generation visual effects
- AR could evolve into seamless, city-scale experiences with zero perceptible lag, processing the 90 zettabytes of IoT data in real-time
- HFT firms could execute trades based on global market conditions in real-time, analysing vast swathes of market data feeds requiring instant processing
From Theory to Practice: Reimagining Possibilities
Imagine a world where:
- VFX artists can manipulate complex 3D environments in real-time, with changes instantly visible to team members across the globe.
- AR applications can process and render entire cityscapes in real-time, creating truly immersive and responsive urban experiences.
- HFT algorithms can analyse and react to global market shifts in nanoseconds, potentially stabilising volatile markets.
These scenarios aren’t science fiction — they’re the near-future enabled by truly high-capacity, low-latency networks.
Preparing for the Petabit Future: Steps You Can Take Today
While the full realisation of petabit-scale networking may still be on the horizon, there are steps businesses can take now to prepare for this high-capacity future:
1. Audit Your Current Infrastructure:
- Conduct a thorough assessment of your existing network infrastructure.
- Identify bottlenecks and areas where current “high-capacity” solutions are falling short.
2. Embrace Edge Computing:
- Start exploring edge computing solutions to reduce latency and bandwidth demands on your central network.
- Consider distributed compute and store solutions that leverage edge nodes.
3. Invest in AI-Ready Infrastructure:
- Even if you’re not fully leveraging AI now, ensure new infrastructure investments are AI-ready.
- Look for solutions that offer flexibility and scalability to accommodate growing AI workloads.
4. Prioritise East-West Traffic Optimisation:
- Re-evaluate your network architecture to better handle machine-to-machine and server-to-server communication.
5. Explore Quantum-Resistant Security:
- Start investigating quantum-resistant encryption methods to future-proof your data security.
- Consider implementing post-quantum cryptography in critical systems.
6. Develop a Data Strategy:
- Create a comprehensive data strategy that accounts for the exponential growth in data volume and real-time processing needs.
- Implement data governance policies that will scale with your growing data needs.
By taking these six steps, you’ll both improve your current network performance and also position your organisation to take full advantage of petabit-scale networking when it becomes available.
The Road Ahead
Redefining network standards is not without challenges. It requires rethinking everything from physical infrastructure to protocols. But the potential rewards are immense. As IDC predicts, by 2025, nearly 30% of data generated will be real-time, demanding networks that can keep pace.
We’re not just building faster networks — we’re creating a foundational layer for the next wave of cross-industry innovation. It’s time for businesses to reassess their network needs, looking beyond marketing buzzwords to the practical implications of true high-capacity networking.
The future of connectivity is here, and it’s measured in petabits. Are you ready to redefine what’s possible in your industry?
