flntwd retweeted
Paywalled article I put out on October 5th:
“Bottom line and IMHO something changed last week and broke…the Fed noticed and is trying to walk back their very hawkish stance from only two weeks ago. Lower rates are coming sooner than I thought.”
flntwd retweeted
A partner of Microsoft said that AI spending from customers for 2027 will be considerably higher.
"Most customers are talking about 35-40% increases in 2027 from 2026."
$MSFT
flntwd retweeted
IRAN SIGNALS OPENNESS TO DIALOGUE AS OIL PRICES DECLINE
Regional media report that Iran's president says Tehran is not seeking nuclear weapons and is ready for dialogue.
The reportedly softer diplomatic tone is supporting market risk appetite, while oil prices move lower amid hopes of easing tensions.
The key takeaway: Any credible progress toward negotiations could reduce geopolitical risk premiums in energy markets.
flntwd retweeted
Blackwell Ultra (B300) GPU rental prices up 17 consecutive days to a new all-time high, according to Compute Desk's index of verified transactions from U.S. neoclouds.
flntwd retweeted
One of the crazy quotes from an expert call I read was from a GPU Cluster Architect at Nebius and he was asked how they model compute demand and the response was:
"Demand is essentially modeled as infinite right now. Generally, for every contract that's about to expire, there's about 15-20 other customers
waiting to bid for that."
Yes, this is basically the industry right now. Everyone is modeling compute demand as infinite, along with the spend.
$NBIS
flntwd retweeted
*NVIDIA CEO JENSEN HUANG: WE DON'T NEED TO BACKSTOP COREWEAVE, BECAUSE THEIR COST OF CAPITAL IS DECREASING AND STRENGTH IS INCREASING EVERY DAY $CRWV
flntwd retweeted
$TSM released earnings today. I was more curious about commentary given recent the market panic.
For analyst Q&A rounds:
Q: Citi explicitly challenged TSMC on AI financing, HBM, CoWoS, lasers and the risk of overbuilding capacity
A: $TSM assessed CSP capex, DC construction conditions, customer financial conditions, and supply of critical components.
Management's conclusion is still that "demand exceeds supply". As well as that AI accelerator revenue growth could exceed its previous 50%+ CAGR projection through 2029.
This is a more meaningful statement since it addresses some concerns regarding recent selloff.
Q: JPM asked if wafer supply could be tighter in 2027 than in 2026 (demand visibility)
A: Wei (Chairman/CEO) stated $TSM even sends staff to verify DC construction progress on site. And after assessing all the factors, TSMC believes gap between supply and demand remains substantial.
It could not determine whether the extent of the gap given customers continues to increase demand.
Q: On pricing power:
A: Wei responded that there is "no ceiling on TSMC's profitability" but that profits will not surge dramatically overnight.
(Source: NextApple, Taiwanese press)
Regarding earnings, very strong: as usual
- TSMC record profits of $23.6B (68% Y/Y growth), and revenue jumping to $46.72B, beating expectations.
- Q4 Guidance was $52.3-$54.3B revenue off 65-67% gross margins.
TLDR:
- No signs of any slowdown or collapse as AI demand far exceeds supply for foreseeable future. $TSM is probably the fundamental proxy/canary to how the overall AI semi trade is doing.
- On financing concerns and fear of overcapacity showing up in semiconductor demand:
$TSM specifically examined the risks investors are concerned about now and assessment still points to unmet demand.
During the Q&A session, Wei said AI demand was "very, extremely strong," with absolutely no signs of slowing.
Underlying AI semi supply chain growth is healthier than recent market panic suggest.
flntwd retweeted
Another rumor I have been told is that Anthropic has pointed to 5-6% of revenue loss from stoppage of distillation.
flntwd retweeted
President Trump had it all back in February.
The 10-Year yield was under 4%, 30-Year mortgages were under 6%, oil was under $70/barrel, and AI investment was booming, with minimal credit risk. Now, that has all reversed.
The Iran War was an unforced fumble of epic proportions.
flntwd retweeted
- Other sites protected from cost overuns, but do expect both sites (Stingray/Black Pearl) to be over budget (AWS pays)
- Could have had leases at other marketable sites (Reveille and Ulyssess), but want to be more thoughtful about final fit out before signing deal
- Half the excess costs is now higher labor ($359M to change order and the excess related to higher costs/inflation to labor and supply chain)
- All financing on the table going forward, want to remain flexible in future
- $14M is probably in line with what tenants are seeing on capex per MW
- Kept citing how beneficial this is on qualitative basis, relationship with this AI lab will be beneficial in the future
○ Wanted to help them pay up front as they are being cautious around cash spend with current part of lifecycle they are at (Anthropic IPO?)
- Sounds like BTM discussions ongoing, with potential at Barber Lake (perhaps Anthropic will be willing to help fund this BTM expansion at Barber Lake and why CIFR wants to help them out with paying this up front)
○ First time we've seen "2.5GW BTM potential" from them
flntwd retweeted
CANADA PM CARNEY’S OFFICE SAYS CARNEY SPOKE TO TRUMP AND THAT CANADA WILL MAXIMISE EXPORTS OF OIL AND DIESEL TO THE U.S., WITH OIL EXPORTS TO THE U.S. YIELDING OVER 1 MLN BARRELS OF DIESEL DAILY.
flntwd retweeted
IRAN SAYS THE US HAS ACCEPTED SOME OF ITS SEVEN CONDITIONS FOR REOPENING THE STRAIT OF HORMUZ, BUT TEHRAN CONSIDERS WASHINGTON’S RESPONSE INSUFFICIENT, WITH UNRESOLVED AMBIGUITIES REMAINING.
flntwd retweeted
I can't emphasize enough how important this @a16z chart seems and how few have internalized it.
flntwd retweeted
Falling token prices + sticky-high GPU prices is classic Jevons paradox playing out in real time.
Inference is getting dramatically cheaper (efficiency, competition, Blackwell-scale gains). That doesn’t reduce compute demand - it explodes it. Cheaper tokens unlock agentic workflows, more coding, more automation and more users. Total token volume is growing far faster than unit costs are falling, so the scarce resource (GPUs, HBM, power, data centres) stays tight.
Implications for AI infrastructure stocks:
- GPU / accelerator makers (NVDA and peers) still benefit. High utilisation + sustained scarcity supports pricing and volume.
- Memory (HBM/GDDR), networking and power infrastructure see durable demand as inference scales.
- Hyperscalers and neoclouds face a margin squeeze if token price declines outrun volume growth, but capacity constraints and rising absolute spend so far argue demand is still winning.
- The shift to inference-heavy workloads favours specialised silicon and high-throughput systems over pure training clusters.
Bottom line: AI is proliferating faster than efficiency is reducing the need for hardware. Unit prices collapse, total spend and infrastructure demand keep rising. That’s bullish for the physical layer as long as volume growth remains the dominant force. alphatarget.com
