I went to NVIDIA GTC and saw the future

Observations on the Next Big Moment

I just returned from my first NVIDIA GTC, touted as the “Woodstock of AI.”

When I started my technology career, I attended mostly hardware conferences. Back in the mid-90s and early 2000s, hardware was the center of the technology universe. Since the rise of the cloud—and the concept, “every company is a software company”—I haven’t attended a hardware conference. It’s mainly been software and cloud events, with hardware mostly abstracted.

On Monday, walking into the SAP Center in San Jose for the NVIDIA GTC (GPU Technology Conference) keynote, I realized that not only have we entered a new era of technology and business, but also the next era of hardware prominence.

I’ll break down what I saw and share my observations. First, a disclaimer. I recognize that I attended a vendor conference whose purpose is selling new things—it was clear that the “sunshine pump” was on full blast. That said, I haven’t felt the same volume, variety, and velocity of participation, interest, and ideas in decades.

Research excitement

NVIDIA GTC is a developer conference that showcases the academic advances underpinning its offerings. I took a stroll through the research poster gallery, which had approximately 100 entries in competition. There was such an exciting range of projects, from computer science to medicine, from oceanography to edge-computing in space—even AI-driven generative music for karaoke!

Scanning the various papers, I saw two common themes:

Being able to accomplish something new that wasn’t feasible before.
Greatly accelerating something that was once slow and complex.

New feasibility excitement

I attended a panel session of AI startup leaders. When asked about their challenges and opportunities, one of the panelists said something profound: “We’ve created something that didn’t exist before. There isn’t an existing category or business or pricing model.”

In that same panel, I learned about Synthesia, a startup that creates AI-generated videos with avatars based on actors or business leaders. The media industry has never seen anything like this: it represents a revolutionary new way to quickly and cost-effectively create new content of certain kinds—allowing resources to be applied to higher-value, higher-impact production that will always require live human talent.

At every session I attended, I continued to hear how generative AI and its surrounding capabilities are moving us forward into the next industrial revolution—making what was once impossible possible.

Use case excitement

A big draw for me to attend GTC was learning about new use cases enabled by AI. The show’s keynote illustrated a dizzying array across different industries—with a heavy emphasis on industrial applications like digital twins and robotics.

In other sessions, I found that democratizing access to AI tools across the enterprise creates many horizontal solutions. For example, the finance team at OpenAI uses ChatGPT to speed up reconciliation. The software engineering team at NVIDIA built a security tool to automate the detection of anomalies and potential security threats. The compelling takeaway is that every department in any organization can benefit from generative AI.

Alliances excitement

I had to laugh while hearing the keynote and recognizing the strategic position in which NVIDIA finds itself. They have partnerships with all of the hyperscalers—and all of the hyperscalers were platinum sponsors of the conference, with big booths and large swaths of interested parties at their respective booths. Is this a new virtuous cycle?

Nevertheless, I was excited to see the many strategic partnerships forming that take advantage of each other’s respective strengths. One example is ServiceNow and NVIDIA’s telecom partnership to drive better customer outcomes through advanced service automation. NVIDIA is also partnering with top security vendors to take advantage of recently developed AI-powered threat detection capabilities.

I heard startup leaders talk about their interest in partnerships that deliver great value to mutual customers. It’s still very early; as AI matures, these partnerships will drive strong and diverse ecosystems.

Now back to reality

Despite all the otherworldly excitement, news, and momentum, I experienced a few down-to-earth moments at the conference.

Enterprise realities: I attended several panels focusing on the enterprise and AI innovation opportunities. What I heard was consistent with what I’ve been hearing for three decades: Current enterprise cultures and siloes—and existing technical debt—are holding back innovation. Culture transformation and change management are priorities in enabling experimentation and production of new AI-powered capabilities. And that’s hard work.

Stuck in POC land: I heard a lot of participants seeking guidance on how to move beyond proofs of concept. I saw a lot of heads nodding across the audience when people complained about not being able to transition experiments into production.

Robots and digital twins: Much emphasis was placed on advancements in industrial digital twins and robotics. I get the power and importance of these advancements, but we’ve been working on them for a while, and they’re very niche in nature. I’m more interested in expanding the digital twin concept outside industrial applications and less in robots.

On-premises vs. the cloud: Here we go again. There was plenty of banter about whether to build your own AI infrastructure on-prem or use the cloud. What I heard from various panelists in ISVs and other organizations is that they’re doing both. Many organizations—for customer applications or research purposes, for example—are choosing to build their own AI infrastructure on-premises.

What else this old exec comms guy saw

I’ve been doing keynotes and supporting events for over twenty years. It was really nice to attend a conference and not have to worry about being backstage stressing about a presentation. It allowed me to observe the event and look for improvement opportunities. Here are some quick takes:

A two-hour keynote is too long: No. Matter. What. I spent an hour in queue to get to my seat, then another hour waiting for the show to start. After two hours of keynote, I was fatigued. NVIDIA CEO Jensen Huang was great but tried to cover too much ground.
Session titles should match the actual session content: Is this stating the obvious? Content and discussion topics during several of the sessions didn’t line up with their titles and abstracts. If I’d known, I would have chosen to go to different talks.
Enterprise topics should include enterprise players: Again, a no-brainer? I attended many of the enterprise-oriented sessions and found that most panel members were enterprise tech vendor leaders. Ideally, you’d have actual Fortune 100 business leaders on panels talking about their AI perspectives, not vendors.
Provide places to sit please: This was one of the busiest conferences I’ve attended in years. The challenge was not enough seating for moments between sessions. I often found myself wandering around, just wanting to sit down.
Plan queue and capacity logistics for popular/featured sessions: Queues for many sessions were very long, snaking through the convention center, disrupting flow, and causing frustrations. I abandoned several promising sessions with lines hundreds and hundreds of people long.

Parting thoughts

It was fantastic to feel the energy at this conference. It’s been quite a while since there was so much excitement and potential.

The industry needs this moment of innovation. I think people have grown tired of talking about “cloud migration” and “risk mitigation.” Jensen said it well: “I’m glad I’m going to be part of this next decade—it’s going to be the best yet.”

Back in Seattle, at Sappington, we’re already at work on our Next Big Thing. I couldn’t be more optimistic to be part of this watershed moment in tech—indeed, the beginning of the next industrial revolution.

Get in touch