Pinned Tweet
I was frustrated trying to compare GLP-1s... Every comparison looks only at weight loss, but the real question is what you pay in side effects to get there.... So I made an GLP1 index and quickly vibe coded a website.
So how it works: I took each drug’s main Phase 3 trial, averaged the rates of nausea, vomiting, diarrhea and constipation, and added the share of people who quit because of side effects. That gives a side effect burden. Then I divided weight loss by that burden. One number per drug, higher is better.
No surprise here. Tirzepatide is at the top with 8.8, and Foundayo (orforglipron) is at the bottom with 3.4....
Full ranking and data: glp1compare.io/glp1-drug-ind…
$VKTX, $NVO
Gasoline futures are literally falling off a cliff. November inflation could come in significantly weaker than expected, which would be a very positive catalyst for the market.
$QQQ
1
147
$NVDA Here is why a return of Jensen Huang is still possible:
Nvidia's board fight is more mechanical than it looks.
Huang holds about 3.6%, above the 3% for 3 years that Nvidia's proxy access bylaw requires. That lets him put 2 nominees on the company's own ballot at the 2027 meeting, the greater of two or "up to 20% of the number of directors in office".
Nvidia's election would then be contested, so plurality applies and the top 10 of 12 names win. A CEO return needs 6 of 10 directors.
It seems 3 of the 10 directors already back a return. Two wins in 2027 would take him to 5, and one swing vote or a settlement does the rest.
Nvidia's cap of two proxy access nominees applies per meeting, so he can hold two seats at a time this way, not add two more each year.
As a shareholder, I hope Nvidia's board comes to its senses and finds a settlement, but I have concerns about the new chair.
1
180
$NVDA I feel the market is overlooking the possibility of a massive share buyback announcement on the earnings call. At these prices, that would be a very smart allocation of capital. A real CFO move.
1
176
Florian Chevallier retweeted
Jensen The White? Huang Reportedly Weighs Nvidia Comeback As Silicon Valley "War Room" Of Allies Assembles zerohedge.com/ai/jensen-whit…
Florian Chevallier retweeted
This letter was sent to the Nvidia Board today which we believe outlines the minimum acceptable standards the Board owes shareholders as it manages this difficult situation. Onward. 🫡
Florian Chevallier retweeted
Cheapest bubble ever. Forward P/E and PEG ratios are hovering around their 35-year historical averages.
zerohedge.com/the-market-ear…
Shareholders feel the same way. This board is incompetent. They could have given him an advisory role while the dust settles. Instead, they fired him as if he'd been caught stealing money from the cash register. Absolutely ridiculous. $NVDA
88% OF 232 NVIDIA $NVDA EMPLOYEES SURVEYED WANT JENSEN HUANG TO STAY CEO
1
309
"US intel isn’t sure if businesswoman spied on Nvidia for China, doesn’t mean she wasn’t a risk" CNN coping hard here. $NVDA
185
Can Jensen Huang come back as NVIDIA CEO even if the board says no?
Short answer: yes, but only by replacing the board. Shareholders can't appoint a CEO. Directors do.
Nvidia's charter closes the fast routes. Written consent is banned. A special meeting needs 15% of the votes, held for at least a year. Jensen owns about 3.56%, so he'd need roughly four times his stake in allies.
That leaves the 2027 annual meeting. Every director is up for election each year. Jensen could fund a full rival slate, win a majority, and the new board reappoints him. Proxy access alone (3% held 3 years) only gets him up to 20% of seats, so it would be a real proxy fight.
The swing votes are Vanguard, BlackRock and State Street. They back whoever they think maximizes value, and they will wait for the Wachtell investigation.
Clean findings plus a sliding stock: a comeback becomes plausible, maybe even without a fight, Altman style. Bad findings: no slate wins.
Watch the investigation, not the headlines. $NVDA
1
257
Jensen Huang will make a comeback. A beautiful American story. I'm sure. $NVDA
1
248
Finally... we can start focusing on something else. I'm sure Huang will come back as chairman or CEO at some point in the future. it's entirely up to him. Shareholders love him, and rightly so! $NVDA
JENSEN OUT
*NVIDIA CEO HUANG RESIGNS, EFFECTIVE IMMEDIATELY
*NVIDIA NAMES CFO COLETTE KRESS INTERIM CEO
145
Chinese buyers will simply continue to turn to the second hand market and buy chips that are barely used.... That's exactly what will happen, potentially even at a premium. $NVDA
COMMERCE DEPARTMENT IS FINALIZING A RULE TO BLOCK CHINESE AI FIRMS FROM REMOTELY ACCESSING ADVANCED AI CHIPS THROUGH DATA CENTERS ABROAD: THE INFORMATION
1
171
$NVDA The forward P/E is just 10x next year's earnings. I like Jensen, but at that valuation, anyone else could step in and steer the ship. This is getting ridiculous...
I think he should be Chairman, and returning as CEO later, or simply stay CEO. It is his company after all..
219
You can almost feel the dirty hands of the closed-source AI lobby behind this $NVDA story.
If it turns out that the woman was indeed a spy, the fact that the CIA never alerted Jensen Huang is quite telling.
The whole thing increasingly looks like an attempt to manufacture leverage to advance a broader agenda: restricting chip exports, tightening regulations on open source AI models, and protecting the competitive advantage of closed source players.
Let’s see what the evidence actually shows.
245
If you want to unlock the full potential of your MacBook or PC and have some freedom with open source models, I highly recommend connecting Hermes to OpenRouter.
You can directly manage your laptop and the files inside, like you take almost control of the OS vs just scrolling the web with Claude agent. It’s seriously impressive and cheap, you pay only what you consume in tokens.
hermes-agent.nousresearch.co…
openrouter.ai/
189
$GOOG Is flexing its muscle by selling TPUs to Anthropic, Meta and Oracle, with four new chips this year. Impressive today. But I don’t think Google can keep up in the long run.
Start with the addressable market. By my estimate, only about 45% of AI compute, mainly inference for text and video, is realistically open to TPUs and other custom chips. The other 55%, training, R&D, robotics and self-driving, stays on GPUs.
Not because TPUs lack power. Google trains Gemini and Veo on its own chips, which proves the silicon works. The issue is everything around it. The rest of the market builds on GPUs first, new models are tuned for CUDA from day one, and robotics and self-driving need physics simulation that custom chips simply cannot run.
Even in video, where TPUs can compete, GPUs remain more flexible and far easier to use. Most open video models are developed on Nvidia, the tooling works out of the box, and Nvidia chips have built-in video encoding hardware. For anyone outside Google, choosing a GPU is the path of least resistance.
Kimi K3, the largest open frontier model ever released, makes the point. Moonshot launched it to run on GPUs, with day one support for Nvidia and AMD. TPUs were not part of the launch. That is what an ecosystem advantage looks like.
Then there is focus. Selling accelerators is not a side business you run from a holding company. It demands 100% focus: a new chip every year, networking, software, developer support and a full ecosystem around it. Nvidia does exactly that and nothing else.
Google is juggling search, ads, cloud, YouTube, Waymo and a frontier AI lab. TPUs are one priority among many. It also relies on partners like Broadcom and Marvell for key parts of its chips and networking, while Nvidia builds the full stack in house.
Software is the biggest gap. Google’s tools are excellent for Google, but the rest of the industry works in PyTorch and CUDA. Moving to TPUs still takes real engineering effort, and customers accept that only as long as the discount is big enough.
Inference is the fastest growing slice of AI, so the TPU opportunity will keep expanding, and TPUs will remain central to Google’s own AI. But as a merchant business competing with GPUs over the next decade, I expect Google to fall behind. The moat is the ecosystem, not the silicon. Broadcom, meanwhile, wins either way....
234
$NVDA is now trading at 11x forward earnings. It's a value play at this point... This is had to believe. Even assuming zero growth (versus +68% consensus estimates), it's hard to see much downside at this valuation.
173
$VKTX the FDA's new vision loss warning on GLP-1s is worth knowing about... but keep the numbers in perspective. In the FDA's review of 134,253 trial patients, there were only 15 probable NAION cases in total: 10 on GLP-1s and 5 on placebo.
Even if the risk is two to three times higher, as several registry studies suggest, that works out to roughly one extra case per 5,000 to 20,000 users a year.
IT IS EXTREMELY RARE... I feel it excites journalists cause being blind on a GLP-1 is an easy sell for an article IMO....
The most likely explanation is two things happening together.
First, anatomy: NAION mostly affects people with a small, crowded optic disc, where the nerve fibres have little room to swell.
Second, a drop in blood flow to the nerve. During dose escalation, appetite and thirst fall, nausea or diarrhoea can add fluid loss, and blood pressure tends to drop, especially overnight. In a vulnerable disc, that may be enough to tip the nerve into ischemia.
This is still a hypothesis, not a proven mechanism. The practical takeaways: stay well hydrated during dose escalation, be careful combining with blood pressure medication, and see a doctor right away for any sudden, painless vision loss in one eye.....
As predicted... Iran's Supreme Leader Mojtaba Khamenei has told the country's leadership to put an end to the “no war, no peace” situation, according to the President @drpezeshkian
Waiting until after the elections was the punishment Iran had in store for Trump.
The CPI numbers are now in the rear view mirror. What matters from here is reopening the Strait. And the sooner, the better.
$QQQ
143













