Can West African enterprises actually benefit from 100+ Tbps switching, or is this a data centre arms race for giants?
Broadcom's latest switching silicon pushes AI workload throughput to new extremes—but what does hyperscale infrastructure mean for regional organisations building out their own data strategies?
Broadcom’s release of its Tomahawk 6 switching platform—capable of handling 102.4 terabits per second of throughput—represents another milestone in the relentless push to move massive volumes of data through enterprise networks. For hyperscalers and cloud providers running distributed AI training clusters, this kind of switching capacity is becoming table stakes. But for most West African organisations, the real story isn’t the raw numbers; it’s what this infrastructure race signals about the future shape of enterprise networking.
The Throughput Treadmill
When switching capacity doubles every few years, it’s tempting to assume every organisation needs it. In reality, most regional enterprises—whether in finance, telecoms, manufacturing, or government—operate data centres and hybrid cloud environments where network bottlenecks are rarely the primary constraint. Overprovisioning switching fabric wastes capital. What matters more is matching infrastructure to actual workload patterns, then building in headroom for growth.
At GDS, we help organisations right-size their networking investments. That means auditing current traffic flows, forecasting AI and analytics workloads realistically, and choosing switching and fabric solutions that deliver performance without unnecessary expense. For many West African enterprises, that’s HPE-class switching and fabric architecture—proven, supportable, and cost-effective—rather than chasing the bleeding edge.
What This Means for Your Strategy
The broader implication is clear: AI workloads are reshaping data centre design. If your organisation is planning to run inference engines, machine learning pipelines, or real-time analytics over the next 12–18 months, your network infrastructure needs to be part of that conversation now. Undersized switching and interconnect becomes a chokepoint quickly.
That’s where a structured approach to infrastructure planning pays off. We work with regional CIOs and infrastructure teams to build data centre and cloud strategies that anticipate these shifts—assessing whether you need to upgrade switching fabric, reconfigure your network topology, or adopt disaggregated networking models.
The Tomahawk 6 is a reminder that the technology landscape keeps moving. Your job is to move with it deliberately, not reactively.