auisssa

I love South Carolina and think they've fit in great in the SEC since '92, but replace them with Vandy so we get back to the pre 1992 SEC.



Don't really care about any other conferences.

Thank you!

It's nice to have someone to trust that wouldn't

quote:

---

use units, axes, and underlying data

---

deceptively.

Nice chart!

It's Game Day

This is NJ Democrats playing games. One way they can hurt Musk and Tesla.

I'm in both Tesla and OUST, so maybe Tesla will add 2 OUST LIDAR sensors to just detect Looney-Toon walls.

Why use "perinatal" in the bill?

From Websters: "The term "perinatal" refers to the period immediately before and after birth. While definitions vary slightly, it broadly spans from conception to up to a year after childbirth. It encompasses both antenatal (before birth) and postnatal (after birth) stages."

Why not use "antenatal" instead?

Yea, I just dumped my bonus in yesterday. Happy it was on a red day. Oh well, it always comes back. But would have been nice to start with 8% more shares today.

quote:

---

Who knows, keep in mind Elon said lidarr was useless for self driving.

---

I was wondering the same thing so ran it through gpt:

quote:

---

What you're seeing is not necessarily LiDAR being added to the production vehicle. The post even says "ground truthing equipment," which is different.

The likely explanation:

Tesla often equips test vehicles with external sensors (including LiDAR, high-precision GPS, cameras, and mapping equipment).
Those sensors collect extremely accurate measurements of the environment.

Engineers then compare what the Tesla camera-only system thought it saw against what the external sensors measured.

The LiDAR data acts as the "ground truth" during development and validation.

Think of it like this:

A student takes a test without a calculator.

The teacher uses a calculator to verify the answers.

The student isn't using the calculator, but the calculator is still useful.

This wouldn't contradict Musk's anti-LiDAR stance because he has primarily argued against using LiDAR on production vehicles for autonomous driving, not against using it as a development or validation tool.

---

quote:

---

So I’ll ask Gemini, what’s a reasonable revenue estimate based on 5GW

---

Ran it through Claude as well. Claude thinks it'll be 4GW but has similar price ranges:

Where Nebius actually stands
Current state: 2026 revenue guidance of $3.0B–$3.4B, with $7B–$9B ARR exit rate (sec) , against contracted power guidance that's risen from >1GW to >2.5GW to >3GW to now >4GW (sec) . Market cap is $70.69B (Robinhood) on 253.9M shares (Morningstar) at $291/share (Robinhood) .
So Nebius is tracking toward ~4GW contracted, not 5GW, and importantly almost none of that is "active" yet — they expect to reach 800 MW to 1 GW of connected power by the end of 2026 (sec) , with the rest coming online over 2027 and into 2028. Contracted ? live. That gap matters a lot for a 3-year projection.

Building the 3-year case (year-end 2029)
Using the framework from before, but anchored to Nebius's real trajectory rather than a generic 5GW hyperscaler:
If Nebius gets ~4-5GW connected and substantially monetized by 2029 (plausible if the Pennsylvania site and other buildouts land on schedule), and using the $10-15B revenue-per-GW-equivalent logic from earlier:

Revenue: roughly $35-55B (this is a big if — it assumes high utilization, not just construction completion)
At 25% net margin: $9-14B net income
At 20-25x earnings (a "matured hyperscaler" multiple, since by 2029 the market would likely no longer pay 70x+ for Nebius given slower growth): $180-350B market cap

Per share on ~280-320M shares (assuming some dilution from capital raises, which the model notes will require significant equity and debt raises to fund the huge capital expenditure (Morningstar) ): roughly $550-1,200/share

That's a wide range, and honestly the spread reflects real uncertainty, not just rounding. The key swing factors: whether contracted GW actually converts to active, monetized capacity on schedule; whether 25% margins are achievable at scale given Microsoft/Meta deals are likely low margin (Morningstar) ; and whether the market still assigns growth multiples or reverts to mature-infra multiples by 2029.

I'd treat this as a back-of-envelope scenario, not a forecast — three years is a long time for a company whose guidance has moved 4x in under a year.

Sort of...Grok's response:

**Nebius is expanding from GPU-heavy AI cloud into full-stack CPU capacity.**

In the podcast, Ben Bajarin (post talking to mgmt) means they're adding serious CPU infrastructure — specifically deploying NVIDIA Vera CPUs (purpose-built for agentic AI) alongside Rubin GPUs in their AI factories.

Why? Agentic AI needs massive high-core CPUs for orchestration, running code/agents, and inference — not just GPUs. Neo-clouds like Nebius have had a big CPU deficit vs hyperscalers. This + the expected AI memory tuck-in rounds out their platform.

I'm waiting until the long term shareholders are able to sell their shares. Might miss the rocketship but that's my strategy.

Effin kill me part is from Ozark show