PostHeaderSection

Symbol

Global NavBar

PostContentSection

Global Footer

Blog Detail Page

Beirit Harvey

At Web Summit 2025, TensorWave CEO joined a panel to unpack what’s really happening behind the headlines.

Inside the AI Infrastructure Gold Rush: Key Takeaways from Web Summit 2025

At Web Summit 2025, TensorWave CEO and co-founder Darrick Horton joined a panel titled “Inside the AI Infrastructure Gold Rush” to unpack what’s really happening behind the headlines of record AI funding, GPU shortages, and mega-rounds.

Alongside fellow founders building in AI software and models, Horton brought the perspective of a hardware-heavy, cloud infrastructure company: one that has to buy GPUs up front, build data centers, and think in gigawatts. The conversation covered how capital is being allocated, why infrastructure plays look very different from AI apps, and what it takes to survive if the AI funding wave slows down.

This blog breaks down the key themes from that discussion, through the lens of AI infrastructure and TensorWave’s AMD-first strategy.


1. Profitability vs. Growth Looks Different in Infrastructure

A lot of venture conversations over the last few years centered on “the rule of 40” and a clear path to profitability. That narrative has softened for many software companies chasing growth at all costs.

Infrastructure is a very different story.

TensorWave operates in one of the most capital-intensive corners of AI: cloud infrastructure and data centers. That means massive up-front investments; real hardware, real buildings, real power contracts.

“We have to build the data center. We have to buy all the GPUs up front,” Horton explained. “But it also means we’re generally profitable very early in the cycle.”

In AI software, many companies are still spending heavily on compute while trying to figure out sustainable unit economics. In infrastructure, the model is closer to traditional, long-lived assets: you invest heavily up front, but your economics and path to profitability are designed into the business from day one.

The takeaway:

 * AI software often optimizes for growth first, profitability later.
 * AI infrastructure has to balance both early, because the capital outlay and risk profile are much higher.


2. Scale is the New Table Stakes

One of the most striking points Horton made on stage: the definition of “big” in data centers has changed almost overnight.

“Two or three years ago, 10 megawatts was a big data center,” he noted. “Now you need 100 megawatts just as entry level, and you need to be talking about gigawatts to even have a seat at the table.”

That shift has massive implications:

 * Power density and capacity: AI training and inference workloads demand unprecedented power and cooling, especially for GPU-dense clusters.
 * Capex requirements: Hitting gigawatt-scale capacity translates directly into tens of billions of dollars of capital deployed over time.
 * Investor expectations: For infrastructure providers, the core investor question becomes: How fast can you get to meaningful scale and can you do it efficiently?

For TensorWave and other infra players, this isn’t a “nice to have” ambition; it’s existential. If you can’t get big enough, fast enough, you risk being locked out of the largest and most demanding AI workloads.


3. The Funding Deluge is Real and Highly Concentrated

The panel moderator highlighted a sharp trend: a huge percentage of venture dollars are now concentrated in a very small number of AI deals. According to recent data they cited, more than 40% of funding in a recent period went into just the top 10 deals, a big jump from the 15–20% range of years past.

Horton sees the same pattern playing out in infrastructure, perhaps even more intensely:

“In our space, deploying data centers for AI is incredibly capital intensive. Investors aren’t going to make small bets on 20 or 30 different clouds. They’re going to pick a small number and make very large bets.”

In other words:

 * Capital is abundant, especially for AI.
 * But it’s not evenly distributed. A handful of players absorb most of it.
 * Infra amplifies that effect because the minimum viable bet is already very large.

That concentration can feel daunting if you’re not one of the “winners”, but it also reinforces the importance of differentiation.

4. Differentiation: Why TensorWave Chose to be AMD Cloud #1

On a stage full of AI founders, one of Horton’s most memorable points was about strategic differentiation.

Rather than become “NVIDIA cloud number 27,” TensorWave took a deliberate contrarian route:

“We early on decided we wanted to bring something slightly different to the market. Instead of being NVIDIA cloud number 27, we’re AMD cloud number one and we’re still the only ones in the world that have that focus.”

That decision matters to investors and customers for a few reasons:

 * It avoids a crowded, look-alike market. Many GPU clouds are fighting over the same NVIDIA roadmap and supply dynamics.
 * It aligns with an open, alternative ecosystem. Backing AMD-centric infrastructure gives investors and customers a differentiated AI platform with less dependence on a single vendor.
 * It creates a “primary provider” slot. Investors who missed early waves of AI infra can now back a leader in a distinct, fast-growing segment.

For TensorWave, that AMD-first bet isn’t just a branding statement; it’s a structural way to stand out in an environment where capital and attention gravitate toward perceived category leaders.


5. Surviving the AI Bubble

Everyone on the panel acknowledged some version of the “AI bubble” question: is this overfunded? What happens if the tide goes out?

That’s one reason TensorWave raised a substantial round while running a capital-efficient business. With a strong balance sheet and comparatively low burn, the company has the ability to:

 * Continue building long-term assets like data centers and GPU clusters even if funding markets tighten.
 * Be selective and strategic about future capital instead of dependent on it.

Horton mentions, “Being able to achieve a certain amount of scale and a certain amount of profitability also protects you. In the infrastructure space, we think what will happen is over time, the players that have achieved some minimum level of scale; a gigawatt of capacity or more over the next couple of years are going to be much more protected. They’re going to have better margins, more profitability, and be able to weather those storms a lot better. And if you haven’t achieved that scale and something happens, you might not be in a good spot.”

The core message: if you want to build in AI infrastructure, you can’t assume the good times will last forever. You have to engineer resilience into the business model.


6. Talent: Winning Without Outspending Big Tech

One theme that comes up in every AI conversation is talent. With major tech companies offering eye-watering compensation packages, how can startups compete?

For infrastructure players like TensorWave, the answer isn’t “outbid Meta.”

Horton’s perspective:

 * You don’t win top talent on cash alone.
 * You win by aligning people with a mission and giving them meaningful ownership of the roadmap.
 * Culture and fit matter just as much as raw brilliance.

Someone who doesn’t align with the culture, even if they’re exceptionally talented, likely isn’t a fit. At the same time, there’s a minimum bar of capability. Sustainable teams are built at the intersection of those two: solid talent and strong culture alignment.

The result is a hiring approach that prioritizes mission, values, and long-term growth over short-term prestige.


7. Is AI Funding Still “Abundant”? Short Answer: Yes

Despite constant talk about bubbles and corrections, all three founders on stage agreed on one thing: there is still a lot of money looking for a home in AI.

“Is the funding environment still abundant?” the moderator asked.

Horton and the other panelists didn’t hesitate: yes.

The nuance is where that capital goes:

 * It’s abundant, but highly selective.
 * It flows disproportionately toward clear winners, differentiated positions, and companies showing real traction and margins.
 * It is increasingly creative, beyond classic ventures, including strategics and other structures, but still power-law driven.

For TensorWave, that means continuing to do what got them to the stage in the first place: build a capital-efficient, differentiated AI infrastructure platform at serious scale, anchored in AMD GPUs and long-term customer value.


Building for the Long Game

The “AI infrastructure gold rush” frame is accurate, but only partly. Yes, there’s a deluge of funding and a scramble for GPUs, land, and power. But beneath the hype, the companies most likely to endure share a few traits:

 * Clear differentiation (like being AMD-first instead of yet another NVIDIA cloud)
 * Capital discipline, even when money is flowing
 * A realistic path to scale in a world where gigawatts are the new benchmark
 * Resilient economics that can survive a market correction

From the Web Summit 2025 stage, Darrick Horton’s message was ultimately optimistic but grounded: this is a once-in-a-generation opportunity, but only for teams willing to think in terms of decades, not funding cycles.


About TensorWave

TensorWave is the AI neocloud purpose-built for performance. Powered exclusively by AMD Instinct™ Series GPUs, we deliver high-bandwidth, memory optimized infrastructure that scales with your most demanding models–training or inference. 

Ready to get started? Connect with a sales engineer. 

Inside the AI Infrastructure Gold Rush: Key Takeaways from Web Summit 2025

Product

Solutions

Resources

Company

Product

Resources

Solutions

Company