# data center revenue — X 热门讨论 (2026-09-23 10:05 UTC)

## @wallstengine (Wall St Engine) · 09-23 09:20 · ♥63 ↻2 💬8 Nvidia-backed Firmus is in talks to raise $10B to buy $NVDA chips for its 360MW AI data center in Batam, Indonesia, ahead of an Australian IPO that could raise up to $5B

Financing: $7.5B debt + $2.5B equity

Nvidia is also expected to provide revenue-sharing and credit support https://t.co/z0Gm5g5SA9 https://x.com/wallstengine/status/2102689676272509152

## @spcxGOD (Panke) · 09-22 23:54 · ♥32 ↻5 💬2 A lot of Tesla is priced in , but this is not.

The reason why Tesla is producing thousands of Cybercabs without deploying them is because of distributed Ai Inference.

Regardless of whether regulation allows them to scale as quickly as they want or not…

Each cybercab is equipped with the tools to become a driving Ai Data center providing around 1 KW of compute per car per year.

At 10 million cybercabs that is 10 GW additional compute priced at 30-50$ is 300-500 billion in potential revenue for $TSLA in 3 years.

This means each person who has a Tesla can choose between letting it join the Robotaxi fleet OR earn up to 30,000 - 50,000 dollars a year by letting their car train AI when it sits idle.

SpaceX will acquire them before this

$SPCX https://x.com/spcxGOD/status/2102547127843082650

## @bert_gilfoyle (Reasonably Approximating 🇺🇸 🇺🇦 🔋 🅰️) · 09-23 05:16 · ♥30 ↻0 💬1 $EOSE

This was an interesting white paper, for sure, and I soaked up a lot of factoids from it.

But, frankly, not much new information stood out to me, because I already knew so much about this and had inferred other things.

However, one thing stood out to me. They have referred to a 3 high, 6 high, and 12 high configuration since last January, and did a again today. But in this white paper, they referred to 12 high as a future thing. As if 3 and 6 are already present, which they aren't. I took that to mean the engineering design (e.g. seismic qualifications) are completed and certified for the 3 and 6 configurations, and maybe they have some promising things in the pipeline for that, but maybe not yet for the 12 high. Otherwise, I'm not sure what they mean by 12 being future, as opposed to the others. 12 high is an enormous height. Almost 100 feet or 30 meters high. The concept rendering building for the Brooklyn Navy Yard was nowhere near 100 feet tall.

The other thing I learned was not really worth mentioning. It was about airflow, and where it exits. Except that it has a filter. And I'm wondering why it needs a filter, and how often it needs replacing.

The real questions you all want answered are what projects will use Indensity and when will that first happen? Well, they answered that for us earlier this year. They said multiple NYSERDA projects were planning on using Indensity. And some of that formed the top end of the revenue guidance at $400m (as opposed to $300m without NY). I'm personally prepared for NYSERDA being less than I originally hoped for, and in some cases coming in excruciatingly long stages.

Fast forward to late September, and I don't believe any of that NY stuff is happening this year, which is par for the course, even with their then sandbagged and gunshy guidance from that disastrous Q4 call. And that kind of explains why they decided to move Line 1 this year, rather than waiting.

How will Indensity first play out and when? I don't see Eos producing Indensity cores in large numbers for customers anytime soon. I have reason to believe that will start at the very earliest Q2 of next year, and more realistically Q3. I believe the first data center applications for Indensity involve Turbine-X and NUAI during phase 2 of the Texas Critical Data Center. That's closer to a 2028 thing. I have written about all this multiple times before.

If 10 years from now Indensity cores become an everyday household item, as nearly ubiquitous as Tesla Megapacks or Safeway stores, you can look back and tell stories about how you knew the handful of investor SOBs in this space, were there and up close, and were early, and you waited for wen. > 引用 @PoweredByEos: Energy storage needs to go where power demand is growing fastest. Data centers, urban substations, military installations, and commercial campuses all present new constraints on space, labor, deployment, and infrastructure. The assumptions built into conventional systems don’t transfer easily to these environments.

That’s the challenge behind Eos Indensity™.

Our latest white paper explores how Eos reimagined energy storage architecture from the ground up, moving beyond the traditional “big box” BESS model toward a modular, stackable, high-density system designed around the real-world constraints of space, speed, constructability, reliability, and resilience.

The paper breaks down three generations of compounding insight that led to what the Indensity architecture delivers: a complete plant-level energy storage solution for places it hasn’t been practical before.

Read the full white paper: https://t.co/UV4WZKZsD2 https://x.com/bert_gilfoyle/status/2102628234911506778