Research-focused. Photonics. AI. Defense. Tech. Space. Optic Supercycle. NFA DYDD
New York, NY
- Tweets 2,591
- Following 575
- Followers 117,603
- Likes 4,368
OpenAI’s been hyping up Dev Day and i’m getting curious
heard rumors about GPT-6 Cyber, AEON, ultra-fast Codex… maybe Codex agents being dropped too?
tuesday could be fun.
1
13
10,169
KawzInvests retweeted
Congrats to everyone who followed me into the $CRDO fear buy on my AB fund on @joinautopilot.
The stock is up over 40% in the last 10 days.
I just bought the FEAR on $CRDO after an 18% crash.
It's one of just 4 stocks in my hyper-concentrated Asymmetrical Bets portfolio on @joinautopilot, focused on the couple stocks I think have the most upside in the AI buildout.
Nothing changed in that belief after their earnings.
I understand that gross margin slipped a bit, but if that's from the optical business taking share isn't that a core part of the bull thesis? And declining sequential growth = the copper business settling into a steadier cadence while optical gets ready to ramp.
The 2027 ramp will be insane - $SPCX is supposedly Credo's biggest customer at 34% of revenue. SemiAnalysis estimates SpaceX could spend $300B to $500B on data centers in 2027 alone as they expand from 1.4 GW toward 6 to 10 GW of AI compute.
At the end of the day, this was a strong beat and raise quarter. Revenue grew 114.7% YoY, FY27 guidance was raised from 80% to 85% growth, and new NPO design wins show the DustPhotonics integration is ahead of schedule.
Reducing cash equivalent $SGOV by ~3% to buy into $CRDO, increasing from 14.34% allocation to ~17%.
For rationale on the investment, read the Substack article: asymmetricalbets.substack.co…
To follow along these trades in your brokerage: link.joinautopilot.com/eXKr/…
7
6
129
31,180
Kind of crazy how quickly the CPU story has changed.
Last August, Lisa Su said “each token generated by a GPU triggers multiple CPU-intensive tasks.”
Three months later, $AMD put the 2030 server CPU market at $60B.
By May, it was $120B.
By August, more than $220B.
And all of that happened before $META launched Muse.
If consumer agents that run browsers, use tools, and complete tasks actually reach scale, it is hard to believe $220B will be the last revision.
13
6
106
18,401
Remember earlier this year when everyone was panicking that Meta had built too much compute and was trying to sell the excess?
Now everyone loves $META again lol.
FinX has a very short memory.
2
1
62
16,665
With the trade truce extended into January, I’m curious what else comes out of Trump’s meeting with Xi today.
Key times to watch for today
• 10:15am - Both leaders speak
• 11:30am - Private talks
• 12:00pm - Talks with both delegations
• 7:55pm - Evening remarks
• 8:10pm - State dinner
3
28
13,561
I find it so funny that FinX spent all of last week panicking about Pacing the Frontier while Anthropic was signing a compute deal with $RUM
Like bruh.
Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family.
It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.
6
41
20,162
$META up 7%.
Muse has the entire timeline in a chokehold.
$META's Muse can search, compare, and help complete purchases for a user.
That raises a simple question: what happens when a purchase begins with an AI agent instead of a platform like Booking com, Expedia, or Uber?
The answer is more complicated than “agents replace aggregators.”
So we followed the process from the initial prompt through checkout to understand which companies remain involved, who gets paid, and why agent adoption creates additional CPU demand.
$META has spent years facing questions about what its enormous AI investment will ultimately become beyond better ads and recommendations. Muse may be the first glimpse of a broader consumer AI business.
The Age of Agentic Consumers. Link below
8
4
179
44,966
$META's Muse can search, compare, and help complete purchases for a user.
That raises a simple question: what happens when a purchase begins with an AI agent instead of a platform like Booking com, Expedia, or Uber?
The answer is more complicated than “agents replace aggregators.”
So we followed the process from the initial prompt through checkout to understand which companies remain involved, who gets paid, and why agent adoption creates additional CPU demand.
$META has spent years facing questions about what its enormous AI investment will ultimately become beyond better ads and recommendations. Muse may be the first glimpse of a broader consumer AI business.
The Age of Agentic Consumers. Link below
17
13
236
69,500
Big week ahead 👀
So we’re already seeing testing for:
Sonnet 5.2
Opus 5.2
Fable 5.2
Gemini 4 Pro
And we have confirmation or hints of upcoming releases for:
Grok 4.7
GPT-6-Sol (Sam and Tibo teased a big release on Tuesday, so this could be it)
Kimi K3.1 (a vague post from Kimi seems to hint at 3.1)
This week was pretty boring. But next week is probably gonna insane.
3
56
37,502
Why am I so focused on networking over raw compute?
Look at GPU utilization 53% of surveyed data center operators report running at only 50–70% capacity during peak demand.
GPUs aren't sitting idle because they lack compute power, they're sitting idle because they're waiting for data. They wait on storage, they wait on CPUs, and above all, they wait on other GPUs to send results back across the network.
If you add more compute to a cluster without upgrading the pipes, you just get diminishing returns on your capex.
$LITE $COHR $AAOI $CRDO
9
10
133
20,208
So much for pacing the frontier.
Per the NYT, Anthropic plans to go from roughly 1.5 GW of compute last year to 5 GW by year-end, then roughly 10 GW next year, while preparing for an IPO.
Nearly 7x the capacity in two years.
6
3
68
15,740
A decade-long thesis with a two-week expiration.
He's baaaaack: "Somebody" just spent $100MM premium on 2 week AI Stock Calls for the Oct2 expiration:
INTC Oct2nd 115c bot up to 3.65 20k $7.3MM
MRVL Oct2nd 250c bot 11.00 3500x $3.85MM
SNDK Oct2nd 1600c bot up to 97.60 4200x $41MM
MU oct 2nd 1000c bot up to 44.00 10k $44MM
26
23
1,217
123,991
$NOK gathering hyperscaler partnerships like infinity stones
NOKIA $NOK & MICROSOFT $MSFT ARE BRINGING AGENTIC AI INTO TELECOM NETWORK OPERATIONS
The companies are expanding their partnership to help carriers automate how network data is collected, analyzed and acted on.
Nokia Data Suite will plug into Microsoft Fabric, combining telco-specific data products with OneLake, Copilot, Foundry and Power BI.
Nokia says the setup can cut access to usable network data from weeks to minutes.
Early use cases include predictive maintenance, automated root-cause analysis, radio-network performance monitoring and closed-loop network operations.
The platform is available now and supports multi-vendor networks across cloud, hybrid and on-prem environments.
13
9
214
27,235
