mascot @thru_xyz. no investment advice.
California
- Tweets 39,326
- Following 6,254
- Followers 111,466
- Likes 112,135
Pinned Tweet
I could not put into words how much today has meant to me.
I’ve been following Intel ever since I was a child and it is not an exaggeration to say they are the reason I fell in love with chips and computers in the first place. I wouldn’t have been able to take the path I have chosen in life if it wasn’t for the thousands of brilliant hard working engineers at Intel who laid the foundation for the ground I walk on.
When I started following their bold bet on becoming a world class cutting edge foundry about a year ago I was galvanized by the vision that they set out. It wasn’t just important to me that Intel succeeded it was important to me that America succeeded. Silicon Valley was built by them and their cohort and it was critical to me that we not only maintained our standing in the semiconductor industry but for us to grow our industry so 100k other flowers could blossom.
There was no way this could be done without an American foundry on the leading edge. I believed in the plan from the start, and as an engineer myself I was absolutely confident that Intel was developing some of the world’s greatest technology and IP from 18A to backside power delivery to the amazing breakthroughs they’ve achieved in advanced packaging which I like to dub as the new Moore’s Law.
They are just getting started and there is without a shadow of a doubt more to come. Today I was vindicated for everything I’ve been talking about for over a year now. The semiconductor industry is won by those not who focus on optimizing margins but those who invest in innovation to bring value to the whole world.
Lip Bu Tan and the whole Intel team have delivered on innovation… and again this is just the beginning. I will be here waiting, watching, cheering every step of the way. Thank you from the bottom of my heart for the experience, the gifts, and your words of appreciation they mean more to me that you can possibly imagine.
With love,
Bubble Boi 🫧♥️
70
75
1,624
900,382
I extend the olive branch 🕊️ to Dwarkesh for the positive intel coverage.
Now it’s time for Intel to put me in that fab to work a shift.
8
4
182
8,487
bubble boi retweeted
Chip packaging is moving so fast it's hard to keep up. @jwt0625 built an online tool that tracks all of it, the first of its kind. It's hugely useful for both technical and market analysis.
[Advanced Packaging Atlas](o5s.org/advanced-packaging-a…)
soundtrack: Baccano! Original Soundtrack - 08 In the Speak Easy
6
77
5,644
Most of this is true but we should add some color to the last point … the industry has largely solved the smaller die problem by developing stitching (Intel) & bigger masks as an answer (TSM).
Intel 18A also already uses high NA heavily according to Intel & its used in the core ultra series 3 processors and D0 already within high volume manufacturing ranges. Stitching has already been extremely derisked with the XT260 system as well (I’m the guy who derisked it lol). The bigger mask work being done by TSM is an alternative but it’s a huge change and is projected to reach production by 2033 with all kinds of other complexity (the whole stage needs to be changed) 💀💀💀
Stitching on the other hand should be good enough for volume production by 2027 at earliest, again it’s been derisked technically a lot (I actually was the person who wrote the RTL for the stage & alignment so i would know lmao). People forget when I was first buying Intel hand over fist it was because I knew the advantage of high na and after the innovations lately it’s only become strongerrrr. Packaging was a sweetener ontop and I’m still in the minority here of betting on foundry turn around.
I don’t just say this just as an Intel shareholder but just as an engineer in general. I honestly think TSMC’s conservative culture and lack of innovation is going to be their downfall. Waiting for 2033 to adopt high NA EUV is going to be the fucking knife through their heart, CoWoS & CoPoS was already a fucking dumpster fire that TSMC is just throwing up their hands and letting the OSATs deal with it.
Rarely do I have this much of an information edge over investors lol. Today I had a great meeting with an investment firm who said they only hold TSM and never bought Intel and they said they just don’t see good ROIC on Intel fab. They did say though that if we thought there were technical reasons that wasn’t going to continue that’s a valid reason to own it.
Let me say this clearly: There are technical reasons as to why the ROI will improve a lot.
I didn’t put all my net worth into Intel 2 years ago for fun lol. I know what’s going on here better than most…
This is not true
TSMC will introduce its version of Backside Power called Super Power Rail (SPR) with the A16 node, which will enter risk production soon.
It's also more advanced than Intel's PowerVia, as TSMC directly connects the backside power to the transistors, while Intel's PowerVia uses nano TSVs to connect backside power to the transistors.
Intel will transition to a similar architecture in 2028 with the 14A node called PowerDirect.
TSMC isn't using High NA for 2-3 reasons. Multi-patterned Low-NA EUV will remain cheaper compared to High NA. The ecosystem around High NA is also not mature. TSMC doesn't want to commit a lot of CAPEX and waste cleanroom space to help mature it rn.
Third, and most importantly, the reticle field size of High NA is half (26 x 16.5mm) compared to Low NA (26 x 33mm). So the size of the monolithic die that u can print in a single exposure is halved (429 mm² vs 858 mm²). So u need to use reticle stitching to print a larger monolithic die with High NA.
7
5
146
15,957
AI relative value trade:
You short 1 billion of GE Verona stock while holding 1 billion of gas turbine equipment.
5
2
118
12,198
In hindsight shorting GE Verona when I (a dog on the internet) got access to an inventory of 2 billion turbines was probably the right move.
5
7
253
21,966
Update we booted Debian 13 and it took 5 min & 48 s after the kernel launched to reach the login prompt.
I just had AI write an entire RISC-V core, put it on an FPGA, compile C code into a custom ELF format, boot the core, and run the program.
I read ZERO of the code. Someone come up with a good name for the core.
1
2
60
7,822
oh no hes pumping intel now
I spent two days touring Intel’s most advanced fab and talking to its technical leadership to understand what the chips of the future will look like.
A lot of what's coming next – gate-all-around transistors, backside power, CFET, hybrid bonding – is hard to follow without actually seeing it, so we made all the awesome animated explainers that I wish I had while I was prepping.
I also asked Intel’s leadership how the semiconductor industry will cope with the enormous shock in demand from AI.
AI compute has grown 3x a year for the last decade, but according to Intel's CTO, Intel could only double leading-edge wafer production with today’s global stock of chipmaking tools.
What kinds of tricks can foundries pull to extend this crossover point where demand surges way ahead of supply?
This was a totally new format for us - hope you enjoy!
0:00:00 – Intro
0:00:47 – How Intel brings up a new process node
0:03:22 – What Intel learned from 10nm
0:06:28 – From planar transistors to gate-all-around
0:13:13 – Why High-NA EUV pays off for Intel but not TSMC
0:19:20 – Could chips have evolved completely differently?
0:23:33 – Inside Intel's "Hot Topics" meeting
0:29:30 – Replacing copper wires with ruthenium
0:32:41 – Why stacked transistors take a decade to ship
0:39:45 – Hybrid bonding and the chip-to-chip bottleneck
0:48:00 – There aren't enough tools to 10x chip production
0:53:14 – Can Intel catch up to TSMC?
1:01:02 – Terafab, and rethinking the fab from scratch
13
1
316
19,774
bubble boi retweeted
There are still 0 confirmed successful uses of this exploit anywhere almost a full day after its disclosure and several days after Amazon first reported seeing *attempts* (key word!) against its servers. But everyone panic! lol.
1
16
4,322
I just had AI write an entire RISC-V core, put it on an FPGA, compile C code into a custom ELF format, boot the core, and run the program.
I read ZERO of the code. Someone come up with a good name for the core.
79
23
688
38,367
