By: Brandon Ross

What happens in Vegas, stays in Vegas. Or so the saying goes. But when more than 3,000 of the sharpest minds in compute, networking, energy, and AI descend on the MGM Grand for the annual Yotta conference, keeping a lid on the flow of ideas is well-nigh impossible. Billed as being “at the intersection of AI, energy, and infrastructure”, the event provided myriad opportunities for delegates to ponder, plan, and propel the future of our industry. Which is why my team took the opportunity of reading the room and doing a pulse survey to gauge responses to a few provocative questions about the future of connectivity.

Over 200 attendees weighed in, from hyperscalers to enterprise IT leads, network engineers to hardware vendors. They shared what they see as probable or improbable in the coming years, as our industry continues to evolve. It wasn’t a scientific sample, but it caught the attention of enough people to gain a valuable snapshot into how insiders see the Internet, the cloud, and even outer space evolving by 2030. And while the methodology might not pass peer review, the findings were surprisingly consistent. Trust is shifting. Cloud strategies are diversifying. Satellites are moving from science fiction to infrastructure. And latency, the invisible force of digital performance, is fast becoming the currency of the next decade. And now, we break the rule and let you know a bit about what happened in Vegas.

What Happens When Trust Runs Out?

Surprisingly, one of the liveliest debates wasn’t about AI ethics or cooling systems or how grand the Internet will become in the future. It was actually about whether enterprises will even trust it by the end of the next decade. Nearly half of those my team stopped between sessions (48%, to be exact) said no: they believe organizations will move away from relying on the public Internet altogether for enterprise traffic by 2030. The other third – a defiant 34% – weren’t ready to give up on it just yet. You can imagine the two camps forming: the pragmatists, armed with horror stories of outages and breaches, and the idealists, holding onto the Internet’s open, democratic promise. For many, the issue came down to control. The public Internet, for all its reach, still operates on a “best-effort” model. It’s a gamble that increasingly few enterprises are willing to make when it comes to mission-critical workloads. The rise of zero-trust frameworks, regulatory pressure around data sovereignty, and the sheer cost of downtime have made reliability and predictability non-negotiable.

Yet, there’s a paradox here. The same enterprises losing faith in the open Internet still depend on it for global reach, innovation, and agility. Their skepticism isn’t about abandoning the Internet, but about redefining how they use it. Private interconnection, direct cloud access, and secure network fabrics are becoming the norm for organizations that want the benefits of connectivity without the risk. In other words, the Internet isn’t dying, it’s evolving. The blind trust that once powered its early years is being replaced by something far more deliberate: a new kind of pragmatism that treats connectivity as an asset to be architected rather than a basic utility to be taken for granted.

The Cloud Conundrum

If there was one phrase that keeps rearing its head – whether at Yotta or anywhere else – it’s “multi-cloud.” Nearly half of respondents (48%) agreed that enterprises will abandon single-cloud strategies by 2026, signaling that the honeymoon phase of one-cloud-fits-all is officially over. For years, the industry has preached the gospel of consolidation: centralize your workloads, simplify your management, streamline your costs. But the reality hitting home for many IT leaders is that every cloud has its own quirks, strengths, and trade-offs, and betting your business on just one can feel a little too much like spinning the wheel. From outages and pricing shifts to regional compliance rules and AI-specific GPU availability, too much now rides on flexibility. Enterprises are learning to spread their chips, mixing public and private cloud with edge resources and colocation to build hybrid ecosystems that play to each environment’s strengths.

Still, this new era of “cloud roulette” won’t be plain sailing. Multi-cloud architectures bring incredible agility, but they also multiply the points of failure and the number of integrations to manage. AI is making that even trickier: training might happen in one