- Return the Fund
- Posts
- The Truth About GPU Overspending
The Truth About GPU Overspending
Learnings from big tech earnings & the supposed AI hardware "bubble"
Return the Fund 🚀
The frontier of tech-focused VC research
In today’s edition
What big tech earnings had to say about GPU overspending & the AI hardware “bubble”
Big rounds incoming for GPU providers & the Middle East’s GPU opportunity
MADE POSSIBLE BY
Startup PromoTix ($48M in traction) is saving the events industry
Solving high ticket fees and low attendance issues
Offers low-fee, SaaS, and patented marketing tools
Raising funds for expansion; profitable with 656k users
None of our startup picks are ever sponsored. So, simply checking out Promo Tix’s website above is the easiest way to keep Return the Fund alive and free. 🤝
Thanks for another round of thoughtful responses, company suggestions, and collaboration requests after last week’s edition.
Please drop a note to team@returnthefund.vc if you have something to say. I read everything!
BIG TECH
Learnings from Earnings
Credit: Andrej Sokolow/Getty Images; F. Carter Smith/Bloomberg via Getty Images; Alyssa Powell/BI via Business Insider
Amazon, Google, Microsoft, Meta, Apple, and Intel all reported earnings this week. The theme? AI infrastructure spend—a presently unescapable topic.
It seems everyone is talking about when the hardware “bubble” will burst.
Throughout big tech’s earnings cycle, one critical theme emerged: the continuing flow of money to Nvidia.
Here’s how earnings went for these 6 companies, from the perspective of AI hardware spending. See if you can spot recurrences.
Amazon (results)
Capital Spending: Increased spending on AI and data centers.
Key Points:
AWS cloud business exceeded revenue targets.
Other segments underperformed, leading to a disappointing overall revenue and earnings projection for the September quarter.
Stock fell over 7% in after-hours trading.
Google (results)
Capital Spending: Increased spending on AI and data centers.
Key Points:
Revenue growth deceleration from the previous quarter.
Underperformance in several key business units.
Stock declined following the earnings announcement.
Microsoft (results)
Capital Spending: Increased spending on AI and data centers.
Key Points:
Noticeable slowdown in revenue growth.
Underperformance in key segments, including cloud services.
Stock price dropped post-earnings report.
Meta (results)
Capital Spending: Increased spending on AI and data centers.
Key Points:
Financial results exceeded Wall Street expectations.
Strong future forecast.
Stock gained nearly 5%.
Apple (results)
Capital Spending: Building small on-device models and investing in AI integration (Apple Intelligence, etc.).
Key Points:
iPhone revenue declined 1% year-over-year.
Modest 5% revenue growth projection for next quarter.
Stock slightly increased in after-hours trading.
Intel (results)
Capital Spending: Facing challenges in AI and data center investments.
Key Points:
3% drop in data center and AI segment revenue.
Adjusted earnings 80% below expectations.
Announced over 15% workforce layoffs.
Suspended dividend paid since 1992.
Shares dropped 26%.
Every single relevant hyperscaler (sorry, Intel) remains steadfast on building the infrastructure for the future of computing.
During his earnings call, Amazon CEO (and former AWS head) Andy Jassy said that despite concerns about infrastructure overspending, Amazon is investing “a significant amount” in AI.
We have a lot of demand right now.
Big tech leadership at large echoes this perspective.
What this means
Take a look at Nvidia’s revenue breakdown.
A breakdown of Nvidia’s revenue by customer. Data from Bloomberg.
Big tech’s cloud’s providers—Amazon’s AWS, Google’s GCP, and Microsoft’s Azure—are all-in on GPU-enhanced data centers.
The resulting infrastructure spend is flowing directly into Nvidia’s pocket.
But if the ROI isn’t there yet, why are these companies playing “who can write the largest check to Nvidia?”
There are many theories to this question—my own being that hardware ROI at this scale takes a long time to materialize.
The most popular theory in venture circles is the Red Queen effect, as characterized by Jamin Ball in Clouded Judgment (and referenced in an earlier edition of RTF).
The Red Queen effect is a concept that originates from evolutionary biology and is often applied to business and economics. It describes a situation where entities must constantly adapt and evolve, not just for progression, but simply to maintain their current position relative to others who are also evolving.
Big tech executives have echoed this sentiment.
In circumstances like this, the risk of under-investing is significantly higher than the risk of over-investing.
If you’re a weapons manufacturer, an arms race driven by the Red Queen effect is nothing short of a goldmine. Both sides want to be your biggest customer.
Our version of the weapons manufacturer is “he who supplies the GPU.” The aforementioned earlier edition of RTF made the following claim about GPU macro factors.
GPU shortages are less common than they were last year, with distributed stockpiles increasing
Disappointing revenue from app-layer companies doesn’t justify the enormous, trend-induced hardware arms race
Huge hardware-in-the-cloud players like Lambda Labs are re-upping, seeking $800 million
This is AI’s $600 Billion Question, as written by David Cahn of Sequoia. In the AI arms race, Nvidia sits pretty supplying nukes to both sides.
We have plenty of questions to answer.
Will the AI arms race last another year? How about another 5? What does that mean for infrastructure companies?
Do hardware infrastructure companies (Nvidia + GPU suppliers) survive an AI “bubble” bursting? What is their staying power over the next decade?
My short answers, to be further elaborated as we examine incoming GPU funding rounds, are (1) it doesn’t matter, and (2) yes. These are, of course, hypotheses, but ones I believe to be true given historical trends, future potential, and market dynamics emerging over the past year.
Taking the market
In the long run, GPUs will proliferate just as CPUs did over the last 5 decades. There will always be competing chip architectures worth analyzing as they arise. But as the landscape currently stands, Nvidia is to GPUs what Intel has been to CPUs.
They’ve been ahead of the game for decades.
In the last 50 years, an Intel CPU has been in at least every other computer manufactured. In 2024, they accounted for 71% of all CPU benchmarks, a proxy for computers in use.
Nvidia currently owns 88% of the GPU market, with AMD occupying the remaining 12%. As economic profit is competed away over time, it’s reasonable to expect Nvidia’s portion of the market to gradually slide, just as did Intel’s.
Nonetheless, they are a formidable company. Nvidia is constantly innovating and somehow always in the right place at the right time (gaming, crypto, AI). And while a hardware “bubble” bursting would certainly bring their multiples back in check, it wouldn’t undermine the fundamental rationale for their proliferation.
Read more on Nvidia’s staying power.
Overall
My hardly-substantiated hunch is that the hardware bubble, insofar as there is one, will remain intact through much of next year. A new wave of GPUs is coming, “small” models still require tens of thousands of units to train, and cloud reservations have not slowed.
The only short-term concern about a hardware “bubble” is entry pricing for infrastructure deals. The best time to fund a hardware-reliant company you believe in is the same moment everyone believes they have no macro tailwinds.
The ROI, EV, and selection risk are optimal, assuming your conviction is irrespective of popular consensus. These deals will return the fund.
Beyond deal pricing, though, given the decade-long lifecycle of a venture fund, my focus is on the players building my expected vision of the next 10 years.
COMING SOON
Part 2 of This Edition
To prevent this week’s edition from inadvertently turning into a short novel, I decided to split it in half.
In the coming days, expect a follow-up on the same overall topic, discussing considerations in analyzing hardware-centered prospects, along with seldom-discussed trends on which GPU companies will capitalize (energy, the Middle East and more).
See you then.
ABOUT RTF
Return the Fund 🚀
One startup a week poised for 10x growth; market deep dives with actionable insights for builders and investors.
Technical breakdowns of advanced new tech to empower informed decision-making
Find your next prospect, product, job, customer, or partner. 🤝 Written by undercover pioneers of the field; trusted by builders and investors from Silicon Valley to NYC. 💸
Last week, we dove into a wild week in AI, the OpenAI Escapism Stack, and technical considerations in revolutionizing search.
As you know, we’re hell-bent on uncovering future unicorns cruising under the radar. Preeminent companies are lean, quiet, and driven before reaching their watershed moments. By the time people start talking about them, it’s too late.
In a nutshell—we pitch you startups like you’re an esteemed VC. If you’re interested in them as a partner, product, or prospect, we’ll make a warm intro. Humbly, our network knows no bounds!
We’ll also intuitively break down advanced tech so you can stay ahead of trends and critically analyze hype in the news and in your circles (regardless of your technical prowess).
Periodically, we’ll propose niche market opportunities. These are tangible ways to extract alpha from the private markets.
You won’t get editions from us very often. Weekly at best. Two reasons:
We’re balancing full-time VC/startup work with Return the Fund.
We prioritize depth, insight, and value. This is not a daily news publication… We hope that when you do get an email from us, it’s dense, valuable, actionable, and worth saving.
Thanks for reading today’s RTF. Feel free to reach out to us at team@returnthefund.vc. 🤝
Psst: None of our company picks are ever sponsored. All research and opinions are completely held by the Return the Fund team.
Reply