To The Moon

If you ever needed proof that the AI race has entered a new, brutal phase, this is it. An image circulating from Bloomberg and public statements reveals OpenAI’s staggering $1.4 trillion commitment to chips and data centers.

This isn't just a spending plan; it's a "shock and awe" declaration. It confirms a thesis we’ve been discussing: the primary bottleneck in AI is no longer just a chip shortage. The real war is for power.

For two years, the market has been obsessed with one company: Nvidia. But this $1.4 trillion plan shows that while chips are the "fuel," the new kings are the companies that can provide the power plants, land, and infrastructure to actually run them.

Let's deconstruct the deals.

The New AI Metric Isn't Dollars—It's Gigawatts

The most revealing details in this plan are not the dollar amounts, but the power metrics. Look at the deals with $Oracle(ORCL)$   and$Advanced Micro Devices(AMD)$  

Oracle: A $300 billion agreement for 4.5 Gigawatts (GW) of data center capacity.

AMD: An undisclosed deal to deploy 6 Gigawatts (GW) of its graphics processing units.

That is a combined 10.5 Gigawatts of power from just two of these deals.

To put that in perspective, a single gigawatt is the output of a large nuclear power plant and can power a city of about 750,000 homes. OpenAI is committing to consume the power equivalent of more than 10 nuclear reactors.

This is the ultimate validation of "power bottleneck" thesis. OpenAI isn't just buying chips. It's buying entire power plants' worth of capacity to ensure those chips don't sit idle.

Who Wins in a $1.4 Trillion Arms Race?

This spending spree isn't just about OpenAI; it’s a roadmap for investors, showing who the real long-term winners of the AI build-out will be. The spoils are being divided among three key groups.

1. The "AI Landlords" (Cloud & Infrastructure)

OpenAI is making massive commitments to the "hyperscalers"—the companies that own the clouds. This is the "rent."

$Microsoft(MSFT)$ : $250 billion. As OpenAI's primary partner, this solidifies Azure's position as the premier AI platform.

$Amazon.com(AMZN)$ : $38 billion. This is a crucial "multi-cloud" move, giving AWS a massive contract to supply OpenAI with Nvidia chips.

$Alphabet(GOOG)$ : An "undisclosed" amount, proving that OpenAI is diversifying its cloud providers to prevent any single one from having too much leverage.

2. The "Power Plugs" (Data Center Specialists)

This is the most critical and undervalued group.

Oracle: The $300 billion / 4.5 GW deal is transformative. It catapults Oracle from a database company to an indispensable AI infrastructure giant.

CoreWeave: An "up to $22.4 billion" expansion. This confirms that specialized, agile data center providers are essential.

3. The "Silicon Diversifiers" (The Chipmakers)

This is perhaps the biggest strategic shift. The $1.4 trillion plan is not just a blank check for Nvidia.

AMD: The 6 GW deal is a monumental victory for AMD. It proves their GPUs are ready for the absolute top tier of AI training, breaking Nvidia's monopoly.

Broadcom: "Tens of billions" for custom chips and networking. This shows the bottleneck is also in networking—how the chips talk to each other. Broadcom is the leader here.

My Takeaway: Follow the Gigawatts

OpenAI has laid its cards on the table. The $1.4 trillion figure is designed to send a message to the market: we will out-build, out-spend, and out-power everyone.

This marks the transition of AI from a software business to the largest industrial project in human history. The "chip shortage" narrative is over. The "power and infrastructure" era has begun.

For investors, the strategy must now evolve. Don't just look at who makes the chips. The real, durable money will be made by the companies who can plug them in.