NVidia (NVDA) is the Golden Ticket to AGI

NVidia’s earnings will be announced on November 19th…

And it will tell us a lot about where we are in the tech cycle for AI.

From Atari to NVidia

Remember, every tech cycle starts the same way…

A small box most people overlook becomes the machine that mints the next decade of wealth.

The “small box” this time is a rack of NVidia chips.

And the people who understand what NVidia reports next will have a chance at their own golden ticket.

I got my first golden ticket in the HOT Florida summer of 1984.

(Yes, a real human writes our newsletter. In fact, this writer grew up without air conditioning, too.)

It didn’t come wrapped in foil.

(Continued below…)

It came in a colorful box that said, “THE ATARI 800 HOME COMPUTER”.

If I close my eyes, I can smell the heated up beige plastic case and hot TV tubes as the night wore on.

I can hear clunky keys… and feel them as they struggle to smack down.

My computer had just 5,000 transistors on its graphics chip.

But to this kid, it was magic…

And the spark that eventually led to the founding of Sage Research.

Today, within arms reach in my rack sit two NVidia GeForce RTX 3090 cards running SLI.

Together, they pack 56 billion transistors.

That’s 10 million times more powerful than my Atari!

Back then, those transistors moved a few pixels on the screen.

Today, they pave the AI superhighway.

At Sage Research, we can spin up racks of the top NVIDIA GPUs in the cloud at will…

But the AI Titans?

The “AI Titans” Are Building AGI

They’re hoarding GPUs by the hundreds of thousands…

And they’re not building breakthrough AI investing models like we do.

Instead, they’re chasing the single most valuable tech breakthrough of the future.

AGI.

Whoever wins AGI wins everything.

No one wants to be second.

But they all need this ONE golden ticket.

(Continued below…)

NVidia GPUs.

That makes Jensen Huang the AI era’s Willy Wonka.

Every hyperscaler CEO wants his more of his transistors…

Including Pichai (GOOG), Musk (TSLA), Zuckerberg (META), Amodei (Anthropic) and Altman (OpenAI).

They all believe — some sooner others — that the electrons that pulse through NVidia’s GPUs could one day ignite the first AGI.

And when that happens, one last inference will run and the world will never be the same.

How soon will could that happen?

Sooner than 99% of the Doom Scrollers Think…

Pichai (GOOG) seems to think AGI is still a few years off, implying around the year 2030….

"In 2010, when Google DeepMind launched, its team would talk about a 20-year timeline for AGI."

Musk (TSLA) is closer to the tech and is more optimistic saying it will happen in 2027-2028.

"AI will superset the intelligence of any single human by the end of 2025 and maybe all humans by 2027/2028. Probability that AI exceeds the intelligence of all humans combined by 2030 is ~100%."

Zuckerberg (META) seems to guess 2026-2027… when his first AI data center comes online.

"Our long term vision is to build general intelligence…”

Amodei (Anthropic), a researcher close to the cutting edge of AI agrees with Zuckerberg (2026-2027).

"Unless something goes wrong, AGI in 2026/2027... AI could surpass almost all humans at almost everything shortly after 2027."

And Sam Altman (OpenAI)… perhaps the closest to cutting edge AI developments said 2025. In January he wrote...

"We are now confident we know how to build AGI as we have traditionally understood it... In 2025, we may see the first AI agents ‘join the workforce’ and materially change the output of companies." 

And that’s why they are all spending trillions with NVidia right now.

NVIDIA’s Latest Numbers are Staggering

Let’s break down their current backlog…¹

NOTE: Google is scaled from Blackwell and based on capex guidance. Also, Anthropic has projected that training a single frontier AI model by 2027 could require 5 gigawatts (GW) of power, equivalent to roughly 1–1.2 million Vera Rubin GPU equivalents at scale. The 200k–500k range may represent a subset for inference or phased AWS deployments, valued at $12–30B based on projected per-GPU costs (~$40k–60k including racks). AWS's $8B+ investment in Anthropic ensures priority access to cloud-based NVIDIA GPUs.

So what will happen this quarter?

Let’s start with their most recent quarter, showing $46.7 billion in revenue…

Along with guidance of $54.0 billion for November 19th.

Could it go even higher than that?

(Continued below…)

The odds are in NVidia’s favor…

Because previous guidance didn’t even take in to account recent news of NVidia’s “loopback” mega-deals.

Nvidia’s “Loopback” Deals: The AI Money Cycle Explained

Nvidia, is now teaming up with big players like Oracle to pour massive cash into AI companies like OpenAI.

This goes beyond generosity.

It’s a clever setup where the money loops back to Nvidia.

Here’s how it works, why it matters, and what happens next…

The Circular Money Machine

Nvidia (and partners like Oracle) invest billions into AI firms like OpenAI.

In return, those firms use that cash to buy Nvidia’s GPUs that run the massive data centers fueling AI breakthroughs.

It’s like Nvidia lending money to a friend who immediately spends it at Nvidia’s store.

The result is an ouroboros of revenue—a self-consuming, self-feeding loop.

In finance, it resembles vendor financing: money cycling around without necessarily creating new economic value, but making Nvidia’s earnings look stellar in the short term.

It pumps up Nvidia’s earnings in the short term, making their financial reports look stellar.

For example:

• Oracle’s Deal: Oracle commits $40 billion to buy ~400,000 Nvidia GB200 chips to power OpenAI’s data centers. That’s direct revenue for Nvidia.
  

• OpenAI’s Mega-Project: OpenAI’s “Stargate” project aims for 10 gigawatts (GW) of AI compute power, needing 4-5 million GPUs. Nvidia’s CEO, Jensen Huang, says each GW costs $50-60 billion to build, with Nvidia pocketing ~$35 billion per GW (70-80% of the total). Over time, this could mean $300-350 billion in revenue for Nvidia from OpenAI alone.

Plus, NVidia gets an equity stake in OpenAI.

At a $1 trillion OpenAI IPO, each 1% of equity NVidia owns is worth $10 billion on the balance sheet.

It’s a win-win-win deal.

This is a big deal…

(Continued below…)

Why It’s a Big Deal

AI computing costs are skyrocketing…. about $500 billion a year globally by 2030.

These loopback deals keep demand for Nvidia’s chips red-hot, ensuring factories stay busy and stock prices soar (Nvidia’s market cap hit $4.5 trillion).

But there’s a catch: this setup could inflate an “AI bubble” if the big promise — Artificial General Intelligence (AGI) — doesn’t arrive on time or if power shortages (like overloaded electrical grids) slow things down.

Why AGI Is the Key

AGI is the holy grail of AI: machines that think and reason like humans, or even better.

It’s critical because it unlocks world-changing possibilities that justify the trillions being spent.

These include:

• Autonomous Agents: AI that can run businesses, make decisions, or manage tasks without human input—like a super-smart virtual assistant for entire companies.

• Personalized AI Companions: Think AI friends or tutors tailored to your needs, revolutionizing education, health and daily life.

• Global Optimization Tools: AI systems that solve massive problems, like optimizing energy grids or logistics networks, saving billions.

• Scientific Breakthroughs: AGI could crack mysteries in medicine (new drugs in weeks), physics (fusion energy), or materials science (better batteries).

If AGI happens, these deals could bring Nvidia $340-390 billion over 5-10 years, plus spark a new “AI everywhere golden age” with even more demand for chips.

Nvidia’s CEO has called Stargate “the first 10 gigawatts,” hinting at a future where 100GW+ superclusters become reality, potentially worth trillions more.

But What If AGI Fails?

If AGI is delayed past 2027 or—worst case—turns out to be impossible, the AI bubble could burst:

• Revenue Hit: Nvidia might only see $100-200 billion of the projected $340-390 billion, as OpenAI and others scale back or default on leases.

• Market Crash: Nvidia’s stock could drop 40-60%, dragging down the broader market (AI stocks are ~25% of the S&P 500).

• Rival Chips: Companies like Google (with TPUs) or startups with custom chips could grab 20-30% of Nvidia’s market share.

• Layoffs and Losses: OpenAI’s projected $115 billion in losses by 2029 could tank partners like Oracle, forcing Nvidia to write off investments.

So, what happens on November 19th when NVidia announces earnings?

(Continued below…)

Short-Term Outlook: Nvidia’s Winning (For Now)

Right now, the “AGI believers” are in control.

Nvidia’s Q3 2025 earnings — with the added loopback revenue — would be hard to bet against:

• $40 billion from Oracle’s GPU orders, already locked in for 2025-2026.

• $300-350 billion projected from OpenAI’s 10GW buildout, starting in 2026.

• Guidance is strong, with analysts expecting $54 billion in total revenue for Q3, ~85% from AI chips.

Long-Term Outlook: Boom or Bust?

If AGI arrives by 2025-2027, Nvidia could dominate a $3-4 trillion AI market, with Stargate scaling to 100GW+ by 2035.

And, as we wrote in an earlier issue… inference might account for 90% of GPU use… so it could be even bigger.

But if power grids falter or AGI stalls (we’re following these trends for you) the circular deals could unravel, echoing the dot-com crash.

For now, Nvidia’s flying high.

We’re tracking them on our radar…

So, Subscribe to our FREE newsletter to find out what happens next.

Always be prospering,

socrAItes

Publisher, Sage Research

P.S. This looks AI created, but the content is a great companion to today’s issue…

1  https://x.com/Beth_Kindig/status/1853797104348524656?referrer=grok-com https://news.futunn.com/en/post/50649340/exposure-of-the-total-number-of-gpus-from-the-top https://x.com/firstadopter/status/1810767197582807148?referrer=grok-com https://x.com/Beth_Kindig/status/1803097821563756580?referrer=grok-com https://x.com/stocktalkweekly/status/1863582744388772234?referrer=grok-com https://x.com/wallstengine/status/1863520913729479156?referrer=grok-com https://x.com/TheTranscript_/status/1793396287649227180?referrer=grok-com https://nvidianews.nvidia.com/news/openai-and-nvidia-announce-strategic-partnership-to-deploy-10gw-of-nvidia-systems https://www.fastcompany.com/91410528/100-billion-open-ai-nvidia-data-center-energy https://x.com/Beth_Kindig/status/1884241149021524046?referrer=grok-com https://www.lesswrong.com/posts/bdQhzQsHjNrQp7cNS/estimates-of-gpu-or-equivalent-resources-of-large-ai-players

2  https://finance.yahoo.com/news/morgan-stanley-lifts-nvidia-target-163309318.html https://www.investing.com/news/analyst-ratings/keybanc-raises-nvidia-stock-price-target-to-250-on-cowos-supply-boost-93CH-4263421 https://x.com/The_AI_Investor/status/1973043370781266182 https://wccftech.com/morgan-stanley-nvidia-gb200-nvl72-racks-deliver-a-profit-margin-of-77-6-vs-64-for-amd-mi355x-while-entailing-nearly-the-same-tco/ https://x.com/T_Smolarek/status/1961146451876008052