comfy

@0xMerp sheesh

Networth 2x in the last month while everybody yapping about the end of the world Hyperliquid

@chrisgrx_ enjoying the chop i see 😝

All eyes on today and tomorrow, especially tomorrows daily close, as crypto tradfi boomer land stops their trading. 70k still LIS for me after we get constructive PA/volume (aka not going to overreact if it gives a brief dip lower today).

@chang_defi wdym y

citrini is literally ansem to the bone

My flight attendant oversaw me checking the markets on my phone and just casually told me with a smirk: “It’s just going down down down. That’s why I have puts on the SPY for tmrw.”

bouncy ass march so far

@thiccyth0t thiccy pick up line goes hard eh

dad? how did you meet mom? I thought she was sexually attractive even though I thought her interests and aesthetic choices were symptoms of America's decline

One of the best ads I've seen in a while. So good.

@lBattleRhino /chumbaworld

What will become the next event that causes everyone to be driven into insanity buying puts

Equities shorters going down a schizo rabbit hole analyzing the Strait of Hormuz watching the admin lie about the length of the war and the cost to the American people

@anidesh7 oklo's team headquarters location:

@comfyleverage Some real P&D GOATs

what's goin on with $oklo hmm

@realroseceline on the q4 cc, green emphasized "trust, transparency, objectivity, and aligned incentives" on why $ttd deserves a premium 3 weeks later, their largest holdco client publicly says $ttd failed an audit for hidden fees, unauthorized opt-ins, and blocking transparency bad look..

@comfyleverage No

10 reasons why Green bought $150m of $TTD stock Jeff Green announced he personally bought about $150m of $TTD stock, the largest purchase of his life, because he believes strongly in the company’s long term future. His main arguments: 1. AI advantage $TTD has been building AI for over a decade. Their platform analyzes about 20M ad opportunities per second and uses huge datasets to help advertisers find the best ad placements. He believes their objective data platform gives them a major edge. 2. The ad market is getting bigger New types of advertising inventory are emerging, especially AI chatbots and sponsored shopping listings. As search becomes more fragmented beyond Google, $TTD could gain access to more ad inventory and expand its TAM. 3. Wall Street is mispricing software Some investors fear AI will make software companies easier to replicate. Green argues this misses the real moat of great companies: integrations, trust, data, ecosystems, expertise, etc. 4. The ad tech media narrative is wrong He believes trade publications exaggerate controversy around $TTD and ignore major developments like $GOOG antitrust loss or his own $150m stock purchase. 5. Open internet will win vs. closed platforms Green believes the future internet will favor open ecosystems rather than walled gardens, which benefits $TTD neutral platform. 6. OpenTTD and agentic AI $TTD launched OpenTTD, allowing other companies to build tools on top of its platform using AI agents that help optimize campaigns. 7. Measurement innovation Advertising measurement is broken today. $TTD is working with retailers and partners to build better systems using real purchase data. 8. Amazon DSP is overrated Green believes $AMZN DSP will eventually disappear because $AMZN makes far more money from its own advertising inventory. 9. Talent and culture He believes $TTD attracts top employees because they believe in the company’s mission. 10. Strong ecosystem Brands, retailers, publishers, and partners are increasingly working with $TTD because they want alternatives to big tech platforms. His message is essentially the market is underestimating the long term opportunity for $TTD. I agree with him and have been saying this same thing for a long time (years). 🌹 Here is a link to his post: thecurrent.com/opinion-jeff-g…

We issued a response to an inaccurate short seller report. Read more here: investors.sofi.com/news/news-deta…

@Alex_Intel_ @jukan05 if u had to put a date range on this happening when would that be? q3 2026?

@jukan05 No we sell Intel when the chip stack is going to have a bunch of 1GB SRAM chips in hottest growing sector of AI demand (Also ignore whole Morgan Stanley event with Dave)

《GF Overseas Electronics & Communications》 ☄️ March 16 GTC 2026 Keynote + Silicon Valley Channel Check Call Summary ☀️ Key Takeaways from the GTC Keynote - AI scaling laws have evolved from the traditional pre-train / post-train / inference stages toward Agent AI, with the longer-term direction pointing to Physical AI. We expect the Agent AI era to drive both GPU and CPU demand, and NVIDIA’s architectural messaging at GTC was clearly focused on improving inference performance. - Revenue visibility for NVIDIA Blackwell and Rubin from 2025 to 2027 has reached $1 trillion, implying roughly $500 billion in data center revenue by 2027, well above the market’s general expectation of $350–390 billion. - NVIDIA unveiled Groq 3 LPU, featuring 500MB of SRAM, built on Samsung 4nm, and offering extremely high bandwidth (150TB/s). The specifications were broadly in line with expectations. The system uses a 32 Compute Tray architecture, with 8 chips per tray, for a total of 256 chips. - The LPX Rack has been integrated into the Vera Rubin system product lineup, and this will also extend to Feynman later on, strengthening NVIDIA’s competitiveness in the inference market. Demand forecast: LPX racks are expected to reach 500 racks in 2026 and preliminarily 10,000 racks in 2027, which is positive for leading domestic PCB vendors. - Highlights of the NVIDIA Vera Rubin NVL72 architecture include the aforementioned LPX integration, a strong emphasis on inference, and mention of urgent demand for storage, especially SSDs. - NVIDIA Vera Rubin NVL144 uses an orthogonal backplane in place of copper cables, in line with expectations, and this is also beneficial for PCB and connector vendors. - NVIDIA ETL256 adopts an MGX design with 8 GPUs per tray, for a total of 256 GPUs. The middle 8 switch trays use Spectrum X rather than NVLink. - CPO scale-up came in below bullish expectations: NVIDIA indicated that both copper and optical solutions can be used for scale-up in the Oberon architecture (via Spectrum X CPO) and the Kyber architecture (up to NVL1152), rather than the market’s previous optimistic assumption that everything would shift fully to optical. Optical interconnect-related names may see some short-term profit-taking, but the medium- to long-term outlook remains positive. ☀️ Feedback from U.S. Institutional Roadshows - Investors had previously become more risk-averse due to the Middle East conflict, rising oil prices, and renewed inflation expectations. But after TACO, confidence gradually recovered, and capital has flowed back into optics (though GTC could trigger short-term profit-taking), memory, and CPU-related names such as substrates. - The broader tone remains constructive on AI, supported by: 1. accelerating commercialization of applications this year, 2. sufficient capital market funding for AI infrastructure, 3. strong infrastructure buildout through 2027, 4. patience for lagging AI names to strengthen, and 5. continued focus on TSMC, CPUs, CPO, optics vs. copper, memory, as well as the opportunities higher memory prices may create for Apple. ☀️ Feedback from U.S. Supply Chain Checks - Google TPU demand remains very strong, and 2027 capacity is already sold out. Shipment demand is far above our previous forecast of 6 million cards, with upside coming from external sales of Broadcom-related projects. - HBM demand is extremely strong, with visibility extending through 2027. Samsung is making the fastest progress on HBM4. Architecturally, AI still depends heavily on HBM’s high bandwidth and speed, though it remains worth monitoring how NVIDIA software optimization and the launch of LPUs may affect HBM demand. - This year, the core theme of AI model iteration is still commercialization and Agentic AI. Groq is well-suited for 1-on-1 multi-turn conversational inference, while CUDA remains the core moat. The KV cache required for inference continues to drive demand for DRAM, SSDs, and context memory. ☀️ Feedback from U.S. Company Meetings - Coherent (COHR, Hold): NVIDIA’s $2 billion investment is mainly focused on CPO, in addition to supply agreements worth several more billions of dollars. The company is optimistic on the CPO opportunity and believes scale-up is accelerating, with both market opportunity and penetration potential higher than in scale-out. Its transition to 6-inch InP capacity, compared with the early stage of 3-inch production, has progressed more smoothly, and full-year capacity is expected to double. The outlook for OCS demand is also positive, and in addition to CSP customers, another CSP is currently under discussion. COHR occupies an important position in the optical module and CPO supply chain and is worth monitoring. - Intel (INTC, Buy): Yield progress on the 18A process is going smoothly, and it is expected to reach industry-standard levels by year-end. 14A is being developed in close cooperation with customers and is expected to progress even more smoothly than 18A did at the same stage. Structural demand for server CPUs remains strong, with double-digit growth expected this year. EMIB is expected to begin contributing revenue in 2H27. We remain positive on the stock, supported by CPU tailwinds, capacity conversion, and solid foundry progress. - SMCI (Hold): Revenue guidance has been raised, but shipment timing is being affected by major customers (xAI) and data center facility readiness. Its advantage lies in rapid deployment and engineering support. The company is relatively conservative on memory procurement in the second half. Given competitive pressure from Dell, we remain cautious on both its margins and revenue growth. ☀️ In addition, we have conducted supply chain and listed company checks across multiple non-expert companies. More details will be shared in follow-up calls and research reports.

@Wild_Randomness @notpratty jfc u are trying to short this into the 150s??

$NVDA As expected, price drifted up into the beginning of NVDA GTC. The 3-day telegraphed bull-posting event now set to begin. While there will certainly be winners, returns out of this event aren’t that surprising to me. (Final image provided by @notpratty)

ceo buying $500k worth of $sofi right into the short report

@Investor_NICK_ @shortbus_ace u are a doctor?? wtf lol

@shortbus_ace 👀… I have my medical school student loans with SOFI. Do i just stop paying? Hehe

guess who buys A LOT of shit paper from SoFi? Blue $OWL

@PegasusFund vol doing weird things

Really seems like we’re sleepwalking into fuel shortages with no plan whatsoever

holy athleticism

forget the draft stuff for a moment, my goodness this is a disgusting reel. this kind of dribble timing and change of height is art.