gab
9.3K posts
gab
@gabclgr
Digital Acquisition Marketing expert. SEM, SEO, paid social, display and affiliate 💻
เข้าร่วม Kasım 2018
512 กำลังติดตาม536 ผู้ติดตาม
gab รีทวีตแล้ว
@sierraaaliyahh Restaurants:
Italian - Torissi, Cafe Carmelini, A Pasta Bar, Cipriani
Greek - Avra, Nerai, Kyma
Bars: Nubeluz, Baccarat, Bar Penelope
Italiano
If you went to school before the 2000's, how many kids did you know with gender dysphoria?
English
@tywoosley23 Magnesium + eye mask + weighted blanket + white noise/guided sleep meditation
English
I’ll Venmo someone $1000 if they can fix my sleep permanently. I havent slept through a night in 10 years. Tried every sleep supplement under the sun , several peptides. Dropped 50 lbs. I still wake up 8-10x a night
English
gab รีทวีตแล้ว
gab รีทวีตแล้ว
AI chips thread time 🍿
1/
Everyone’s talking about AI chips, but nobody’s talking about the chips that feed the chips.
When you plug in your phone, it charges.
When an AI data center plugs in… it eats a small power plant.
Let’s talk about the companies that make that possible — and why I’m quietly rooting for the little one in the group. ⚡️🧵
⸻
2/
The world’s power grid wasn’t built for AI.
These GPU racks from $NVDA now draw 1 megawatt each — that’s like a thousand houses.
So the new mission isn’t “make more power.”
It’s move the power smarter, without wasting half of it as heat.
That’s where this crew comes in. 👇
⸻
3/
$IFNNY (Infineon): the German giant — ~$50 B market cap.
They make the big industrial power chips that run EVs, factories, and servers.
$STM (STMicro): ~$30 B — strong in automotive and energy conversion.
$TXN (Texas Instruments): ~$170 B — the analog king.
$MPWR (Monolithic Power): ~$50 B — the power-supply whisperer for GPUs.
$NVTS (Navitas): ~$1.7 B — the new kid with big dreams.
⸻
4/
All of them build pieces that turn electricity into the right kind for chips to actually use.
Think of the power grid as a river.
These companies build the dams, valves, and filters that slow the water down just enough so the computers don’t drown.
⸻
5/
Here’s the twist:
Old chips that manage power (made from silicon) are good…
But new materials — GaN and SiC — are better.
They waste less power, stay cooler, and switch electricity faster.
That’s the upgrade path the whole industry is moving toward.
⸻
6/
$NVDA’s new 800-volt architecture for AI data centers will need tons of these GaN and SiC parts.
They literally can’t run efficiently without them.
And in $NVDA’s own announcement, they name-dropped:
$IFNNY, $STM, $TXN, $MPWR, $ON, $ADI … and $NVTS.
That’s a big table for a small player.
⸻
7/
So what makes $NVTS special?
They’re the only one on that list fully focused on next-gen materials — GaN + SiC. (Photo from investor deck attached)
No legacy silicon baggage.
Just clean-sheet designs aimed right at the new 800 V world.
They’ve already shown off working 8.5 kW and 12 kW power supplies at ~98 % efficiency.
⸻
8/
Here’s the part that gets me:
$TXN and $IFNNY can move the needle a little and it’s billions in revenue.
For $NVTS, even a small win — one big data-center customer, one socket — could double or triple their business.
That’s what “asymmetric upside” really means.
⸻
9/
I want them to win.
Not because they’re perfect or guaranteed — they’re not.
But because they’re swinging at the right pitch: making every watt count in a world running out of watts.
If they execute, they go from $1 B to something with a “tens-of-billions” neighbor. Also 30% short…
⸻
10/
So next time you hear someone talk about “AI power shortages,” remember:
It’s not just about building more.
It’s about using what we already have better.
That’s the mission for $NVTS — the smallest name in the room trying to make the whole room more efficient. ⚡️
Source: ir.navitassemi.com/static-files/f…
$NVDA $IFNNY $STM $TXN $MPWR $ON $ADI #AI #GaN #SiC #DataCenter
English