Big Technology Podcast

Episode: Coreweave: AI Bubble Poster Child Or The Next Tech Giant?
Date: January 7, 2026
Host: Alex Kantrowitz
Guests: Michael Intrator (CEO, CoreWeave), Brian Venturo (Chief Strategy Officer, CoreWeave)

Episode Overview

This episode delves into CoreWeave's meteoric rise in the AI infrastructure world, examining whether the company is the quintessential symbol of an AI bubble or a genuine future tech giant. Host Alex Kantrowitz interviews CoreWeave CEO Michael Intrator and Chief Strategy Officer Brian Venturo, exploring their business evolution, operational challenges, market positioning, and the underlying realities regarding AI infrastructure, demand, risk, and financial structure.

Key Discussion Points & Insights

1. The AI Infrastructure Surge: Scale & Pace

• Unprecedented Growth:
  • CoreWeave IPO’d eight months ago, now valued at $42B.
  • Built eight U.S. data centers in Q3 alone; owns ~250,000 Nvidia GPUs.
• Exhausting but Exciting:
  • "It's exhausting. All right, so let's start with that. It's been exhausting." — Brian Venturo [03:10]
  • "It has been an unbelievable year... it feels like it's been two lifetimes." — Michael Intrator [03:15]
• Purpose & Privilege:
  • Founders align their passion with the societal impact of AI:
    "It's a privilege to come into work and focus your energy, your creativity every day on building a component of this, of artificial intelligence..." — Michael Intrator [04:22]

2. Handling Hypergrowth & Internal Culture

• Company Growth:
  • Went from ~100 to ~2,500 employees in three years.
• Pressure & Resilience:
  • "We're doing something at a scale no one's ever done before, at a speed no one's ever seen before. Of course things are going to go wrong, but take perspective..." — Brian Venturo [05:21]
• Speed Bumps:
  • Mistakes happen, but the company maintains momentum and morale.

3. Customer Base & Market Position

• Diversifying Beyond Microsoft:
  • Microsoft is a major customer but now <30% of backlog, reflecting broader client diversification.
  • "We've done a really good job bringing on other wonderful clients..." — Michael Intrator [07:33]

4. Building Data Centers: Tech & Differentiation

• Shifting from Leasing to Bespoke Buildouts:
  • "Now we're doing much more bespoke in house design..." — Brian Venturo [09:02]
• The Real “Secret Sauce”:
  • Not in physical plumbing, but in software for managing provisions, validation, and failure recovery.
  • "The differentiation comes after you turn it on and how you control those systems." — Brian Venturo [10:23]
  • CoreWeave’s software enables clients to extract more value from commodity hardware:
    "We have built allows us to take the commodity GPU and deliver a decommoditized premium service..." — Michael Intrator [11:28]

5. Training vs. Inference: Shifting Demand

• Changing Use Cases:
  • Historically served model “training”; now much more “inference”—actual deployment and monetization of AI.
  • "Our customer base for the last three years has primarily been the largest AI labs... it's now shifted... to people that want to use those capabilities to change business outcomes." — Brian Venturo [14:42]
• Balance of Workloads:
  • "Six months ago I would have said it was 2/3 training and one third inference. It's probably close to 50, 50 now." — Brian Venturo [16:12]
  • Inference is rapidly growing, reflecting real business adoption and monetization.

6. Why Does CoreWeave Exist? — Strategic Positioning

• Birth from Crypto into AI:
  • Originated as an Ethereum mining company; pivoted to AI as demand spiked.
  • "We recognized that compute was going to be valuable. We didn't necessarily know at the time what it was going to be valuable for." — Brian Venturo [18:36]
• Building for Cloud 2.0:
  • "As you moved from sequential computing into parallelized computing... it stands to reason that a fundamental change in how compute is used will also require a fundamental change in how you build the cloud..." — Michael Intrator [19:37]
• Product Differentiation:
  • Optimized parallel compute, building proprietary software and infrastructure superior to legacy “Cloud 1.0.”

7. Demand, Risk, & Financing

• Narratives vs. Reality:
  • Large tech users (Microsoft, Google) still build as fast as possible—even as they buy from CoreWeave.
  • Companies like CoreWeave “take the risk” so hyperscalers don’t have to preemptively over-invest.
• Risk Mitigation Approach:
  • Long-term, creditworthy client contracts facilitate stable growth: "85% of our exposure is to deliver compute to investment grade or AI labs..." — Michael Intrator [24:14]
  • Only a minority (15%) is speculative, uncontracted exposure.

8. Strategic Use of Debt

• How Financing Actually Works:
  • Debt for infrastructure is underwritten by top financial institutions (Blackstone, BlackRock) with strict waterfalls.
  • Customer payments go into a "box" that pays opex/interest/principal before profits are released—minimizing default risk.
  • Compression of loan interest rates over time reflects lenders’ increasing confidence:
    "It went from SOFR +1350 down to SOFR +400..." — Michael Intrator [41:29]
• Low-Risk Leverage Model:
  • "Low-risk" use of debt akin to how power plants or railroads have been built historically.
  • "We didn't invent anything new here. We just took a tried and true method and applied it to the specifics..." — Michael Intrator [30:18]

9. Skepticism and AI “Bubble” Concerns

• GenAI as a Developing Category:
  • Some concerns around the ability to turn AI investment into profit.
  • CoreWeave manages exposure via diversification and risk-aware contracting.

10. AI Hardware: Depreciation & Usefulness

• Lifespan of GPUs:
  • Popular narrative: GPUs "burn out" or become obsolete in 2-3 years is "bunk".
  • Real-world data:
    "Hyperscalers... finally retire their Nvidia K80 fleets... which were active for 10 years." — Brian Venturo [45:14]
• Value Preservation:
  • New term contracts being signed for existing, older GPUs (A100s, H100s) at 95% of original value.
  • "The most important tool that I have... is not what I think... it’s what are the buyers... willing to pay for today?" — Michael Intrator [49:11]
• Software-Driven Long-Term Value:
  • Thanks to their software, older GPUs stay useful throughout contract terms.

Notable Quotes & Memorable Moments

• On the Emotional Toll of Growth:
  "Sometimes since the IPO, we've been under this spotlight... and inside the company... we always set the highest bar for how fast can we do something, how high of a quality can we do it at?"
  — Brian Venturo [05:07]

• On CoreWeave’s Real Differentiator:
  "The differentiation comes after you turn it on and how you control those systems..."
  — Brian Venturo [10:23]

• On Skepticism and Market Analysis:
  "I feel like the depreciation narrative is being spun up by folks... who couldn't spell GPU two years ago and now they are out there as experts..."
  — Michael Intrator [49:07]

• On Real-World Value of Old GPUs:
  "We signed a term contract for the A100s... within 95% of an original price range... Those GPUs are already five years old."
  — Brian Venturo [53:04]

• On Managing Risk & Contracts:
  "When we think about compute, we think about what is the option value associated with it. When we think about the data centers, we think about what is the option value to be able to build, to be relevant in the future."
  — Michael Intrator [24:44]

Important Timestamps

• 03:10: Exhaustion and excitement of scaling CoreWeave
• 05:07: Managing company culture under pressure
• 07:33: Customer concentration and diversification
• 09:02: Building bespoke data centers; shifting strategy
• 10:23: The secret sauce—data center control software
• 14:42: Shift from training to inference in AI demand
• 16:12: Current workload is about 50/50 training vs. inference
• 18:36: Origin story—pivot from crypto mining to AI
• 24:14: CoreWeave’s approach to risk and contract structure
• 30:18: Use of debt and financial structure
• 41:29: Interest rate compression as confidence grows
• 45:14: Accelerated depreciation debate; GPU lifespan reality
• 49:11: Real market data refutes rapid depreciation claims
• 53:04: Five-year-old GPUs still holding contract value

Conclusion

CoreWeave is not simply an AI boom beneficiary but has built a highly differentiated—and resilient—business at the intersection of AI hardware, software, and finance. Their approach combines infrastructural expertise, a deep understanding of parallel compute, and conservative risk management—all the while navigating narratives about bubbles and skepticism with real-world data and customer demand as their north star.

For those wondering if AI infrastructure is a bubble or a sustainable sector, CoreWeave’s founders make a strong argument for the latter, not least because of their flexibility, rigorous financial discipline, and ongoing value-add beyond just possessing hardware.