- AI-driven traffic demand has been “grossly underestimated” across the industry, Smith told Fierce Network
- The shift from telecom carriers to hyperscalers is reshaping network investment and ownership
- Smith said 5G was “a bust” from a carrier economics perspective
Ciena CEO Gary Smith stepped into the role at the height of the dot-com bubble — a moment he now describes as the start of a 25-year journey defined by volatility, structural change and, increasingly, the central role of networks in the digital economy.
“I took over as CEO just in time for the dot-com bubble… that was kind of my first phase… basically survival,” Smith said.
That volatile period reshaped both Ciena and the broader telecom industry. What began as high growth quickly turned into what he called a “telecom nuclear winter,” with revenues collapsing and vendors scrambling to stay afloat. “We went from… revenues of a billion and a half down to… $200 million in the space of four months,” he said.
Looking back, Smith said that one of the most consequential moments of the past 25 years was the industry’s overbuild of fiber capacity (aka the fiber glut) during that era — painful at the time, but foundational in hindsight.
“That was without doubt… the biggest defining moment of the last 25 years,” he said, referring to the “build it and they will come” mindset that initially failed.
While the glut triggered a collapse, it also enabled the rise of the modern internet. Smith said that many digital business models — from e-commerce to ride-sharing — would not have scaled without “copious amounts of low-cost bandwidth.”
At the same time, that reset how value is distributed across the ecosystem. “It really determined the value piece… between the carrier and the content,” Smith said, adding that bandwidth came to be treated as “pretty much free… or close to it.”
That imbalance continues to shape the industry today, particularly as networks face a new wave of demand driven by artificial intelligence.
Not prepared for what's ahead
However, Smith also said the industry has not fully grasped the scale of what is coming. “The amount of capacity… required just to facilitate AI… has been grossly underestimated," he said.
In his view, AI represents a convergence of earlier technological shifts, such as compute, storage, mobility and networking, into a single platform. “That’s the platform for AI… that’s the killer app on top,” Smith said.
But translating that into operational networks introduces new constraints. Training large models across multiple data centers, moving data globally and supporting inference workloads all increase pressure on optical infrastructure. “Nothing operationalizes in AI until it comes out of the data center,” he said.
The result is a supply-demand mismatch that is still playing out. Component supply, manufacturing capacity and fiber deployment are all struggling to keep pace. “We can’t get enough fiber in the ground fast enough,” Smith said.
At the same time, the balance of power in telecom infrastructure has shifted. Where carriers once dominated long-haul and submarine investment, hyperscalers are now leading.
“The biggest owners of submarine cable capacity in the world are now the hyperscalers… it used to be all carriers,” Smith said.
That transition is also changing how operators position themselves. Smith pointed to the rise of wholesale models and managed fiber networks tailored to cloud providers and emerging “neoscaler” players focused on AI infrastructure.
“The network is a critical element now… as important as power,” he said.
Overhyped innovations
Despite that growing importance, Smith was blunt about areas where expectations have not matched reality. Asked about the most overhyped innovation of the past 25 years, he pointed to 5G.
“The promise of 5G… and the economics… it’s been a bust,” he said.
He also raised questions about long-standing policy debates, noting that network neutrality has played a key role in shaping how value is distributed between infrastructure providers and content platforms.
Looking ahead, Smith expects the pace of change to accelerate, not stabilize. “The velocity of change is going to outstrip our wildest expectations,” he said.
But whether the industry can keep up remains an open question. The risk, in his view, is not a lack of demand but the ability to scale supply quickly enough to meet it.
“The challenge… is how do we scale up as quickly as we possibly can… otherwise we become the constraint,” he said.
After 25 years marked by boom-and-bust cycles, Smith sees the current moment differently — not as another temporary surge, but as a shift tied to AI’s reliance on networks.
“If you’re not excited about what we’ve got in front of us now… I don’t know what excites you,” he said.