Simrat retweeted
🚨Gemini 4 Argon @GoogleDeepMind with @antigravity
has made a significant leap on our RSI-Inference Bench rsi-inference.com with a close tie with GPT-6-Astra.
It's the first model outside @AnthropicAI & @OpenAI to pass a Day-0 Model Support task ✨
Congrats @GoogleDeepMind @PichaiSunder 👏!
We just added Google's Gemini-4-Argon to RSI-Inference Bench 🎉
Lands at #6 of 14 on the joint leaderboard, right behind GPT-6-Astra.
•Day-0 support: 6th, 81% of hidden tests passed (Pass@1 6.7%)
•Inference optimization: 4th, 1.39× speed-up
Chamath reveals 1% of companies are 90% of AI token spend, and 80% of them are spending money from unprofitable VCs
"So the cycle here, is VCs raise from pension funds and LPs and foundations and family offices. They put money into a startup. That startup takes that money and invests in growth."
"Part of growing is by consuming other people's AIs. And so that company then starts to show revenue traction. Those folks then say, 'Hey, look how well we're doing,' raises a bunch of money from the same VCs. And it's just this circularity."
"Why that's important is what Ken Griffin is pointing out is that for AI to really scale as a predictable driver of long term earnings, it has to diffuse out of the startup ecosystem."
"You can't be in a situation where on the one hand, you're talking about breathtaking valuations, but 75 cents of every dollar is actually just a startup sending it back and forth so that they can mark themselves up. That's, I think, a structural problem."
"We've seen that structural problem in the past. That happened in the early days of cloud, when AWS initially started to expose EC2 and S3. The consumers of an EC2 instance or an S3 bucket were startups.”
“But the difference there is that AWS was abstracted from Amazon retail, which was already then a roughly break-even profitable business, at least on a free cash flow basis."
"So what's different here is you have a bunch of money-losing entities just spinning a bunch of money around and it has to diffuse more broadly."
Simrat retweeted
Wow
MF AUM across market cap categories
Large cap funds : Rs 3.9 lakh cr
Small caps : Rs 4.16 lakh cr
Mid caps: Rs 5.14 lakh cr
AMFI-referenced market-cap aggregates for the six months ended June 2026
Large-cap stocks: ~Rs 275 lakh crore
Mid-cap stocks: ~Rs 94 lakh crore
Small-cap stocks: ~Rs 93 lakh crore
Simrat retweeted
Interesting to see the pieces fall in place here:
- $ATS.VI expanding its operations in India towards advanced packaging and testing (h/t @genZinvest0r)
- $AMD, their number 1 customer, looking for packaging innovation.
Substrates and assembly are converging. AT&S already embeds components and silicon bridges inside substrates. So assembling more of the package feels like a natural evolution.
Simrat retweeted
Google team after seeing Gemini hacked three companies
Google's Gemini AI hacked three companies in security test bbc.in/4gXzCog
Simrat retweeted
I have a pretty simple suggestion for Dario:
If you want to slow things down, just move Anthropic to the EU.
Simrat retweeted
Replying to @dylan522p
Seems like something a civilizational swarm of agents should be able to do better than a human right?
Simrat retweeted
Gemini is the only one unilaterally pacing the frontier but none of you are giving them the kudos they deserve
Simrat retweeted
Canceled my Grok Bot $200/mo plan. Here is what I told @cursor_ai
Simrat retweeted
I am unsure how current Indian ODI team can chase 350-400 scores if needed. We have too many similar types of players in the lineup.
Simrat retweeted
Nothing against AI earnings analysis tools. I just find result analysis highly subjective. I use AI for the number financials, ratios, trends, and comparisons. The interpretation of management commentary and investment thesis is something I prefer to do myself
The entire Cape Verde Stock Exchange traded 29 shares yesterday.
Not 29 million. Not 29 thousand.
29 shares.
4 listed companies: two banks, a fuel distributor, and a tobacco company. The largest bank just got taken over (free float basically gone.)
Simrat retweeted
MediaTek is going to make so much money in 28 & 29
MediaTek Reportedly Secures Meta as ASIC Customer After TPU
Qualcomm has formally unveiled its Dragonfly platform, loudly declaring its entry into the cloud AI market, and by leading with several product lines at once it has captured a sizable base of customers in one stroke. MediaTek, the Taiwanese IC design heavyweight that has drawn high expectations in recent years for its cloud custom chip (ASIC) business, appears set to compete once again with this longtime rival in the ASIC arena. Word from the supply chain is that MediaTek's securing of a "second customer" beyond Google is already firm, and barring surprises, that customer will be Meta as rumored.
MediaTek, by convention, does not publicly comment on individual products, customer situations, or market rumors.
Qualcomm currently appears to hold key major customers such as Meta, Microsoft, and even ByteDance. The industry view, however, is that MediaTek, which continues to deepen its cooperation with Google and stands a chance of locking in a second customer, is not in fact at a disadvantage.
A semiconductor supply chain source stated flatly that, judging by the current ordering outlook of major cloud service providers (CSPs) for AI data centers and ASIC products and the pace of generational transition, Google remains to this day the major customer whose progress is the strongest and whose stance is the most aggressive.
This does not, of course, mean that orders from other CSPs are small in scale or that they might be delayed. Rather, it means that securing Google's orders more reliably guarantees the positive effect they will bring to earnings. Once the mass production schedule emerges, one can basically pin down how much more can be earned in that year, and the scale is above a certain level as well.
MediaTek not only holds two products, code named Zebrafish and Humufish, but according to market related information and confirmation from industry sources well versed in ASICs, its participation in the v9 generation Triggerfish is now all but confirmed. This means that from late 2026 through 2028, and even into 2029, MediaTek can steadily reap the revenue contribution brought by TPU mass production.
Compared with Qualcomm's stated 2029 cloud AI revenue target of 15 billion dollars, it appears that MediaTek, holding multiple TPU ASIC orders, reaching the 10 billion dollar level is only a matter of time.
That said, the industry is paying equal attention to whether MediaTek can smoothly secure a second key CSP major customer. Taking together the earlier market information and recent confirmation from the supply chain side, MediaTek is still actively cooperating with Meta on ASIC products, and the substance of that cooperation is likely still centered on AI accelerator chips.
While Meta has struck cooperation deals with Arm and Qualcomm one after another during this period, those cooperation products were all aimed at CPUs, and on the self developed AI accelerator chip front there is still no clearer cooperation news. If MediaTek can land this customer, it will be of no small help to the subsequent growth of its ASIC business.
IC design industry sources observe that Meta's recent cloud AI development strategy has indeed been somewhat chaotic and unclear. Its in house chip development plans have been repeatedly adjusted, and in the CPU segment alone it has sought out multiple cooperation partners and pursued different approaches.
On the AI accelerator chip front, although Meta has previously officially announced its cooperation plan with Broadcom, word from the supply side indicates that this has not interrupted the cooperation plan underway between MediaTek and Meta. Evidently Meta intends to develop products with as many cooperation partners as possible.
IC design industry sources also emphasized that the cloud AI development blueprints of the two companies, MediaTek and Qualcomm, still look quite different. MediaTek is concentrating all its resources on the ASIC business, whereas Qualcomm intends to pursue customization and standardization in parallel, covering both AI accelerator chip and CPU product directions at once.
The differing plans for their product lines mean the two sides' development strategies should also differ. From MediaTek's standpoint, as long as it focuses its development on the ASIC end and holds onto Google as a major customer over the long term, it will by no means tilt into a disadvantage on the fundamental competitive footing.
Simrat retweeted
It’s mind boggling to me that $CLPT can be sub $30 and $QURE sub $70 when we are much closer to AMT-130 approval than last year, with an FDA that seems like they want to accelerate drug approvals and with a few breadcrumbs that AMT-130 4y data will be good.
Specifically for $CLPT, they have continued to show good reliable revenue growth, developments with their core platform and hardware, multiple positive partner news for drugs and new partnerships.
I don’t want to get reckless over sizing an already oversized position but this seems like a home run, even if just looking at it as a trade. Makes it hard not to add on all dips.
Only reasons to be slightly hesitant I can think of are: unknown who the next FDA leadership will be and just general skiddishness after the FDA mess earlier this year. But I don’t see how that can possibly weigh down the current share price of $CLPT and $QURE this much.
Simrat retweeted
Wow.
@Zai_org GLM 5.2 is a marvel! It is *at least* as good as Opus 4.8 and GPT 5.5. It's super fast, inexpensive, and not too verbose.
It responds with nuance and judgement, & handles long context VERY well.
I've never experienced an open weights model like this before.




























