 How to win out against the competiton by a pivot to new areas - head off further the deep seek challenge they hope....
Tackling real-world tasks
OpenAI’s early testing highlights several areas where GPT-4.5 excels. OpenAI’s early testing highlights several areas where GPT-4.5 excels:
Creative writing & design: The model demonstrates a stronger aesthetic intuition, making it a more effective tool for writing assistance, storytelling, and brainstorming ideas.
Programming & problem-solving: GPT-4.5 improves its ability to follow complex multi-step instructions, making it a more reliable coding assistant.
Factual knowledge & research: Thanks to its refined training, the model hallucinates less, meaning users can expect more accurate and reliable responses in knowledge-based queries.
Emotional intelligence: OpenAI has incorporated more human-like conversational skills, allowing GPT-4.5 to respond empathetically and provide better user support, whether for educational guidance or personal encouragement.
hxxps://www.tomsguide.com/ai/chatgpt/openai-chatgpt-4-5-is-its-most-human-like-chatbot-yet-heres-how-to-try-it
nai/dyor etc |
NVDA CC this week seems consistent narrative with the OAI 4.5 compute demands...
"Hugging Face alone hosts over 90,000 derivatives freighted from the Llama foundation model. The scale of post-training and model customization is massive and can collectively demand orders of magnitude, more compute than pretraining. Our inference demand is accelerating, driven by test time scaling and new reasoning models like OpenAI's o3, DeepSeek-R1, and Grok 3. Long-thinking reasoning AI can require 100x more compute per task compared to one-shot inferences.
Blackwell was architected for reasoning AI inference. Blackwell supercharges reasoning AI models with up to 25x higher token throughput and 20x lower cost versus Hopper 100"
Our manager seems to have picked up on the TCO difference between blackwell, hopper and competings ASICs with a post on twitter this week
nai/dyor etc |
Sam Altman -twitter yesterday...more chips still needed :)
GPT-4.5 is ready!
good news: it is the first model that feels like talking to a thoughtful person to me. i have had several moments where i've sat back in my chair and been astonished at getting actually good advice from an AI.
bad news: it is a giant, expensive model. we really wanted to launch it to plus and pro at the same time, but we've been growing a lot and are out of GPUs. we will add tens of thousands of GPUs next week and roll it out to the plus tier then. (hundreds of thousands coming soon, and i'm pretty sure y'all will use every one we can rack up.)
this isn't how we want to operate, but it's hard to perfectly predict growth surges that lead to GPU shortages.
a heads up: this isn’t a reasoning model and won’t crush benchmarks. it’s a different kind of intelligence and there’s a magic to it i haven’t felt before. really excited for people to try it!
nai/dyor etc
[...] |
Synopsys also reported last night
nai/dyor etc |
Nvidia $NVDA on Blackwell: "We delivered $11.0 billion of Blackwell architecture revenue in the fourth quarter of fiscal 2025, the fastest product ramp in our company’s history."
Nvidia $NVDA data center revenue in FY25 rose 142% YoY to $115.2 billion.
For Q1, Nvidia $NVDA guided for $43 billion in revenue, +/- 2%, ahead of estimates for ~$42 billion.
Twitter tech analyst comments
nai/dyor etc |
Twitter update yesterday
OpenAI also said that paying enterprise users have roughly doubled since September to 2 million, while developers using o3 have quintupled
·
10h
OpenAI says that it now has 400 million weekly active users, a 33% increase from 300 million in December.
nai/dyor etc |
More irony, on a bad day for the NDQ100 the three MAG7 names in the green (as I write this) are the three that we hold -NVDA, Apple, MSFT. NVDA results next week which may clarify things a bit more here.... nai/dyor etc |
"KeyBanc analysts noted NVIDIA's challenges with manufacturing constraints for its GB200 NVL server racks but expect the impact to be more than compensated by several factors. Customers have reportedly been substituting orders for the GB200 with HGX-based B200 servers, which have x86 head nodes, due to lower initial yields for the GB200 NVL. Additionally, a surge in demand for NVIDIA's H20 GPUs from China's cloud service providers (CSPs) has been driven by limited supply of Huawei's Ascend AI ASIC and the introduction of DeepSeek." how ironic ...
nai/dyor etc |
NEWS: Elon Musk said tonight that he expects xAI's next training cluster to have 1 million GPUs, worth about $25B-$30B, 5x more than their current 200k GPU cluster."We're already working on the next training cluster. We anticipate it to be roughly 5x bigger than the current cluster."
twitter user
Update on two of our holdings reporting last night
TipRanks Feb. 18, 2025, 03:58 PM Arista Networks (ANET) has reported fourth-quarter 2024 financial results that topped Wall Street forecasts on the top and bottom lines.
Cadence Design Systems (CDNS) reported strong Q4 and fiscal year 2024 results, with Q4 revenue reaching $1.356 billion, up from $1.069 billion in Q4 2023. The company achieved a record backlog of $6.8 billion and current remaining performance obligations (cRPO) of $3.4 billion.
nai/dyor etc |
We have stated shorts on the following 3 stocks-interesting choices for an AI fund to be short NIKE and others , but all seem to be down over 6 months - are these positions therefore in profit?
nai/dyor etc |
Our 38% NVDA holding is up to over $140 in overnight trade back to almost exactly where it was before the deepseek news (up over 25% from the lows) - proof if we needed it that the market has now moved entirely on from the story. In contrast, MNL share price seems to remain in a state of "shock" almost unable to budge much at all in 2025.
After nearly 9 months of consolidation, and punters leaving after deepseek scares recently, what will it take for any sort of revival here. Quite some change in near term fortunes after the past two years of success.
nai/dyor etc |
Microsoft CEO Satya Nadella: “We have more than doubled our overall data center capacity in the last three years and we have added more capacity last year than any other year in our history.”
Twitter tech analyst quote today nai/dyor etc |
www.trustnet.com/fundswire/13437393/deepseek-and-learning-about-diversification/?fe=1&utm_campaign=Fundswire%20Alert&utm_content=Forensic-News&utm_medium=Email&utm_source=FE%20Trustnet%20Alerts
Liontrust view published today on deepseek and portfolio management - different perspective..nai/dyor etc |
MNL fact sheet on deepseek noise: "Sadly, it has shaken out a lot of investors in the US Ai names who, we guess, will repent at leisure on their decision to exit".
Why Morgan Stanley thinks the market has called it wrong on Nvidia in compelling detail. Some interesting stats on market share between NVDA, AMD and the cloud asic providers..Best quote I could see:
One exasperated cloud executive recently remarked, “Every two years, our ASIC team delivers technology that is 2–3 years behind Nvidia. It is economically not that useful.” Of course, this is not everyone’s view, nor is it the objective at the outset of the design process. However, it is a fairly common complaint among cloud vendors, who—even while considering it an investment in the future—are not entirely convinced that adopting inferior ASICs will form the foundation of a differentiated long‑term strategy.
nai/dyor etc |
Alphabet $GOOG CEO Sundar Pichai: “Today, cloud customers consume more than 8x the compute capacity for training and inferencing compared to 18 months ago.” twitter update today from tech analyst
nai/dyor |
As luck would have it, located some sharpe ratios (not to be taken as gospel of course)..
MNL - 1.1
PCT - 2
ATT - 1.8
SMT - 1.7
nai/dyor etc |
A picture paints a thousand words...are your success comments referring to a 5 year time horizon...and do you look at risk adjusted returns...
The approach here is to use high beta to outperform (c.3 times the volatility of the FTSE). We had a great couple of years. Over the past period since the high in June 2024 the performance on a relative basis has dwindled imvho - so you have to ask the question. Will this continue and do I want to burn up my risk bucket vs. other tech investments.Be interesting to report sharpe ratios to clarify better.
nai/dyor etc |
You seem to discuss everything but your trading online. This is one of the MOST successful funds in the UK !!! I suggest that if you think you are better do it yourself. The reason I ask was I AMa successful retired ex City trader and I got sick to death of Commentators! Mr Sheppard has huge skin I. The game, I have put significant funds in too. In my experience the ones who comment most Comit the Least! |
"Three of the four most negative attributors to Nasdaq’s performance for the month were our Top 3 holdings, a bad month both absolutely (~-2.4 per cent) and relatively (~-4.6 per cent v Nasdaq Composite). The DeepSeek distraction not only caused us performance pain but also chewed up about 7 days of our time. Our forthcoming Newsletter will cover in more detail why we do not think this event will have any unexpected long-term consequences apart from being an excellent lesson in propaganda from the CCP. Sadly, it has shaken out a lot of investors in the US Ai names who, we guess, will repent at leisure on their decision to exit. We also saw a lot of our own shareholders throw in the towel...."
Says it all really ....fact sheet out yesterday...the suggestion is that those bailing may well regret it...I wonder if they are moving on the basis that they think they will outperform this strategy...I would never disclose my trading online but I struggle now to understand their investment strategy here any more...there is an important AI value chain in other parts beyond chips e.g. cyber/health. We seem to focus heavily on a part of the market with names struggling and under performing stocks eg AMD/ASML. Perhaps over a longer time line things may come back...all depends on your time scales...all imvho of course...
nai/dyor etc |
nai/dyor etc |
How much stock do you own? |
We are c.16% down vs. SMT just in 2025 NAV YTD according to my Morningstar site - total mean reversal from the AI boom of prior years. Hedge fund shorts agains the chip sector the highest for many years according to an analyst this week.
The tactical asset allocation of the fund over the last year has surprised me .....
I have revised my views considerably as we are lagging at the moment at the time of higher volatility than peers.
nai/dyor etc |
Very good point TIE. That's why I'm out of MNL for now and have slimmed down drastically on most AI stocks while Trumps tariff and Ukraine policies play out. There's huge scope for market upheaval right now so I've taken most of my profits in TSLA and PLTR off the table for now. Waiting in cash and silver while this plays out. Dull maybe, but I'm not losing any sleep! |
Register for Free to get streaming real-time quotes, interactive charts, live options flow, and more.
Hot Features
Premium
