Associate Professor Banking and Finance,Università Cattolica del Sacro Cuore, SWF,ETF,Islamic Finance, CNBC, @Natgasweather, Energy Working, Foresight Economics
Verona,Italy
Joined December 2014
- Tweets45.8K
- Following1.9K
- Followers7.9K
- Likes30.6K
Pinned Tweet
Safe havens or war hedges? #oil #gold #natgas during Iranian war. Below the link of the publication
pdf.sciencedirectassets.com/…
Andrea Paltry, Ph.D. retweeted
$NVDA | Jensen Huang Discusses Comeback As Nvidia CEO With Allies - @FT
ft.com/content/45383d86-33c4…
Andrea Paltry, Ph.D. retweeted
🚨 BofA sees a $1.58 TRILLION memory supercycle running straight through 2028.
In its November 11 note, BofA uses more conservative price assumptions than TrendForce, yet still forecasts global memory revenue rising from $970 billion in 2026 to $1.58 trillion in 2028.
When an industry adds $610 billion of annual revenue in two years and doesn't even need spot price fireworks to get there, bit volume, server mix, and multi-year contracts are doing the real work.
AI inference and expanding context windows scale straight with memory capacity. That's why Jensen said at GTC Berlin on October 22 that you can always use more memory, as $NVDA pulls mobile LPDDR into data centers alongside HBM.
Huang's personal relationships across Asia helped $NVDA lock in priority HBM allocation on favorable terms. Moving from founder handshakes to a finance-led transition while hyperscalers ramp custom ASICs turns future HBM supply into a true multi-buyer market, shifting pricing and allocation leverage back to the memory makers.
$SKHY already showed what that sellers' market looks like in its October 28 results: Q3 revenue topped 104.6 trillion won at an 80% operating margin, HBM4 becomes its mainstream product in Q4, and long-term agreements already cover more than 50% of its 2027 and 2028 capacity.
Less reliance on one dominant buyer gives $SKHY even stronger cards in HBM4 negotiations.
$MU has put enforceable cash behind that post-2028 buffer in its fiscal 2026 Form 10-K.
Its take-or-pay backlog reached about $134 billion at price-band floors, backed by $12.75 billion in customer cash deposits paid in fiscal 2026 that mostly stay locked until 2029 through 2031, plus $7 billion in bank letters of credit.
China's $CXMT proves how deep server demand runs, posting Q3 revenue of 118.71 billion yuan (+613% YoY) at an 89.53% gross margin while rolling out 32Gb DDR5 chips and 12,800Mbps MRDIMMs as servers pass half of total DRAM demand.
Without Huang at the helm after his October remark about being open-minded on Chinese suppliers, $NVDA isn't onboarding $CXMT anytime soon, leaving CXMT to capture domestic AI servers while $SKHY and $MU keep Western HBM and long-term contract economics to themselves.
$MU $SKHY $NVDA $CXMT
Andrea Paltry, Ph.D. retweeted
Bloomberg Intelligence projects the global data center total addressable market will climb from roughly $630 billion in 2025 to $1.7 trillion in 2027, then reach about $2.8 trillion by 2030.
Compute servers and GPUs would remain the largest segment at about $1.1 trillion in 2027, as accelerators require more high-bandwidth memory, faster networking, denser storage and better power delivery.
Storage and networking should grow as agentic workloads move more data, while land, power, cooling and software expand but lose share to compute. Other firms tracking capital spending, like Dell'Oro and Omdia, see $1.6–1.7 trillion by 2030.
Andrea Paltry, Ph.D. retweeted
J.P. Morgan on DC buildout and capex spend:
"Post the midterms, we are maintaining our elevated infrastructure buildout expectations for 2027 as we believe much of the spending for next year is already committed in the form of GPUs, signed power deals, & shells under construction or permitted."
"We recognize that building data centers could become more difficult and expensive, but we continue to expect spending will remain elevated to meet demand. We currently model hyperscaler capex growth of ~87% to $1.7T in 2027 & +19% to $2.0T in 2028."
$GOOGL $MSFT $META $ORCL $AMZN $SPCX $NVDA
J.P. Morgan on data center buildout & capex:
Post the midterms, we are maintaining our elevated infrastructure buildout expectations for 2027 as we believe much of the spending for next year is already committed in the form of GPUs, signed power deals, & shells under construction or permitted.
We currently model hyperscaler capex growth of ~87% to $1.7T in 2027 & +19% to $2.0T in 2028.
$GOOGL $META $AMZN $SPCX $NBIS $CRWV
Andrea Paltry, Ph.D. retweeted
One week later...
When do we deviate from just plain ol' solid El Nino look, again? 🤷♂️
#StillWaiting #Weather #ElNino #Natgas
Andrea Paltry, Ph.D. retweeted
Forward P/Es on semis dropped nearly 48% from the June peak. Next-twelve-month earnings estimates rose over 40% in the same stretch.
The Philadelphia Semiconductor Index sits 27% off its highs. Year-to-date prices are still up about 51%. NTM EPS growth ran near 139%.
Nvidia and Broadcom held up better than the pack. Their cash-flow multiples stayed resilient. Datacenter buildouts keep pushing multi-year wafer fab equipment spending higher.
The recent drop (excluding $NVDA ) comes from doubts about AI demand lasting, not from earnings cuts. Sell-side consensus numbers show the gap.
How long do investors keep pricing the demand risk when the earnings still climb?
Semiconductor equity forward price-to-earnings multiples contracted by 47.9% since their June peak, even as consensus next-twelve-month earnings estimates expanded by 40.1% over the identical timeframe. The Philadelphia Semiconductor Index has dropped 27.0% from its June highs and fell 7.2% in a single day, despite remaining up 50.9% year-to-date. This sharp correction reflects an intensifying crisis of confidence among investors regarding the sustainability of AI demand and hardware returns. Yet underlying operational delivery has moved in the opposite direction. Year-to-date next-twelve-month earnings estimates for the benchmark have been upgraded by 138.9%, driving an overall 36.8% multiple compression as fundamental profit growth outpaces stock prices. Traditional AI market leaders including NVIDIA ($NVDA) and Broadcom ($AVGO) have exhibited greater share price stability, supported by more resilient cash flow multiples relative to higher-multiple semiconductor capital equipment and CPU peers.
Sell-side analysts project that current sector underperformance is driven almost entirely by valuation derating rather than downward revisions to operational fundamentals. Over trailing six-month and one-month horizons, earnings outlooks were revised upward by 43.9% and 15.9% respectively, while forward multiples contracted by 29.3% and 27.0%. Datacenter buildouts continue to create a multi-year expansion phase for wafer fab equipment spending, positioning suppliers for sustained volume growth. The core operational risk centers on whether hyperscale capital deployment converts into durable software returns before investor fatigue restricts external funding, though current hardware demand indicators continue to firm across server infrastructure.
#Semiconductors #NVIDIA #Intel #AMD #Industrials
Report date: 2026-11-02
Andrea Paltry, Ph.D. retweeted
There have been no order cuts for Nvidia, according to four senior executives in the upstream semiconductor supply chain.
Despite rumors suggesting order cuts resulting from Jensen's scandal, orders remain unchanged for 2027 and mass production orders are looking ahead to 2028. An insider said the company expects orders to remain strong for the next 2-3 years.
Another semiconductor industry insider stated that Nvidia will not abandon the global AI market simply because Jensen Huang resigned, and Nvidia's long-term competitiveness will continue.
One industry insider pointed out that after Jensen Huang's resignation, the competitive landscape between GPUs and ASICs may partially shift towards ASICs, but the overall AI industry will not experience a bubble.
$NVDA $GOOGL $MSFT $META $ORCL $AMZN
Andrea Paltry, Ph.D. retweeted
$CRWV SEEKS $4B LOAN BACKED BY META CONTRACTS
The financing would allow CoreWeave to borrow against Meta's AA credit standing rather than its own, targeting an all-in interest rate of around 7.5%.
That's well below the nearly 16% yield on its unsecured bonds, which have sold off sharply this fall amid growing concerns about AI financing.
Morgan Stanley and Mitsubishi UFJ are syndicating the loan, which has reportedly drawn a positive early response from lenders.
Source: Semafor
Andrea Paltry, Ph.D. retweeted
Nvidia Announces CEO Transition, CFO Colette Kress Interim CEO As Huang Resigns
Andrea Paltry, Ph.D. retweeted
Saudis Tell OPEC That Output Surged 26% To New War-Time High; They Pumped 10.6M B/D In October
Andrea Paltry, Ph.D. retweeted
$CRWV | Citi 𝗿𝗲𝗶𝘁𝗲𝗿𝗮𝘁𝗲𝘀 𝗕𝘂𝘆/𝗛𝗶𝗴𝗵-𝗥𝗶𝘀𝗸 on 𝗖𝗼𝗿𝗲𝗪𝗲𝗮𝘃𝗲, raises PT to $𝟭𝟰𝟵 from $𝟭𝟮𝟴
Analyst sees strong execution, better pricing and contract economics, with available capacity sold out through 2027.
Andrea Paltry, Ph.D. retweeted
CoreWeave earnings call highlights:
> Through 2027, more than 75% of committed CapEx is tied to investment-grade and enterprise customers, with the remainder tied to the world's leading AI labs and AI-native companies.
> This year, they have collected prepayments on 70% of our deals, with customers funding up to 100% of the contract value upfront. In the next 3 months alone, they will collect approximately $9 billion in prepayments and deposits from signed customer contracts.
> They expect EBITDA less interest, which excludes the benefit of prepayments, to more than double in 2027, increasing their ability to internally fund capital needs.
> Allocating more capacity to short-term opportunities where the pricing justifies it, selling capacity at more than $40 million per megawatt annually, while pricing for new infrastructure on 4- and 5-year contracts is moving to $20 to $25 million per megawatt.
> At current pricing levels, 2027 renewals would generate approaching $4 billion of incremental annualized revenue with no additional CapEx required. This grows to nearly $10 billion of annualized incremental revenue when including 2028 renewals.
> They have 10 times as much qualified demand as available capacity through the end of 2027, and demand for 2028 is already several times larger than our available capacity, even before we have brought it formally to market. They now have more than 1.7 gigawatts of active power. They continue to expect to reach more than 1.85 gigawatts of active power by year-end.
> Revenue backlog ended the quarter at approximately $133 billion, up 138% year over year, reaching another record high. Of this backlog, more than 50% is attached to a contract where they expect customer delivery to commence by the end of this year.
$CRWV
Andrea Paltry, Ph.D. retweeted
$CRWV: "Based on the capacity we expect to deploy and the contracted revenue embedded in our backlog, we are raising our ARR guidance and we expect to exit 2026 with annualized run rate revenue of $20 to $21 billion."
Coreweave has a market cap of 26.88B!!
CoreWeave just reported an absolutely massive quarter, but there's one number that I think investors should be paying very close attention to.
The company reported $3.66 billion in Q3 revenue, up 168% year over year, while adjusted EBITDA surged 169% to $2.26 billion. Even more impressive, CoreWeave now has a revenue backlog exceeding $132 billion and raised its annualized revenue run-rate outlook to $20 billion to $21 billion. Demand for AI compute clearly isn't slowing down, and CoreWeave is positioning itself to become one of the largest infrastructure providers in the industry.
But here's where things get interesting. CoreWeave expects to spend between $35 billion and $39 billion on capital expenditures in 2026, while its quarterly interest expense alone is expected to reach $1.05 billion to $1.15 billion in Q4. That's more than $4 billion in annualized interest costs. And despite the incredible revenue growth, the company still reported an adjusted net loss of $529 million.
This is what makes CoreWeave such an interesting investment. The demand is clearly there, customers are committing billions of dollars, and the company has an enormous backlog of future revenue. But financing all this growth is incredibly expensive, and investors need to understand how much debt and capital spending it takes to generate that revenue. I still think CoreWeave has enormous long term potential, especially with AI infrastructure demand continuing to accelerate. But the biggest question isn't whether CoreWeave can grow revenue anymore. It's whether the company can eventually turn that growth into meaningful profits and free cash flow without constantly relying on additional financing.
Andrea Paltry, Ph.D. retweeted
CoreWeave $CRWV expects to turn adjusted net income positive in the first half of 2028, and for the full year 2028
It expects its adjusted net loss to "reduce meaningfully" in 2027, and EBITDA minus interest to more than double
"Adjusted operating income at $379 million nearly tripled from the second quarter, while interest expense rose about a third. That is the path to earnings," the CFO said
Andrea Paltry, Ph.D. retweeted
Coreweave $CRWV +13%:
Beat across the board: revenue +168%, backlog +138%. Raised its 2026 guide, nudged capex up.
Financing visibility looks much better:
No new unsecured debt planned through '27 at current yields & less than $5B of new parent-level capital expected.
Its next loan is expected to be investment-grade, yield 7%-7.5%, and fund over 90% of the capex it's tied to. Over 75% of committed capex through 2027 is tied to investment-grade and enterprise customers.
About $9B in customer prepayments and deposits is expected over the next 3 months from signed contracts. Over $3B is for already-paid-off GPUs, requiring no new capex or debt service.
DEMAND ISN'T THE ISSUE.
It's "largely sold out through 2027," and all 2028 capacity offered so far has sold. Qualified demand is 10x available capacity through 2027, while 2028 demand is already several times supply.
The older GPU fleet is getting more valuable, not less.
Ampere and Hopper pricing jumped 30%-50% QoQ, above their original contract rates. Over 130K GPUs roll off contract in 2027 and ~170K in 2028. Re-renting them at today's prices could generate nearly $10B in additional annualized revenue with no new capex, none of it in the backlog yet.
New 4-5 year deals now price at $20M-$25M per megawatt a year. Short-term capacity goes for over $40M and can pay back its capex in 6-18 months.
Margins are improving: adjusted operating margin went 1% → 5% → 10% across Q1-Q3. Management expects positive adjusted net income in 1H 2028.
Capex can flatten. If it only builds the committed backlog, 2027 capex stays around 2026's $37B-$39B while revenue and EBITDA grow.
WHAT'S STILL OPEN:
Q3 capex was $14.5B on $3.7B of revenue, and interest expense of $846M was over 4x GAAP operating income of $207M, with Q4 interest guided above $1B.
The parent raise could tap its new ATM for up to 35M shares, but management didn't specify. And there's no quantified downside case if GPU supply catches up and shortage pricing fades.
COREWEAVE $CRWV Q3’26 EARNINGS HIGHLIGHTS
🔹 Revenue: $3.66B (Est. $3.55B) 🟢; +168% YoY
🔹 Adj. Operating Income: $379M (Est. $234M) 🟢
🔹 Adj. EBITDA: $2.26B (Est. $2.04B) 🟢; +169% YoY
🔹 Adj. Net Loss: -$529M (Est. -$677M) 🟢
🔹 GAAP EPS: -$1.27
Raises FY26 Guide:
🔹 Revenue: $13.0B-$13.3B
🔹 Adj. Operating Income: $1.3B-$1.4B
🔹 CapEx: $37B-$39B; from $35B-$39B
🔹 Active Power: >1.85 GW
🔹 Run-Rate Revenue: $20B-$21B
Other Q3 Metrics:
🔹 Revenue Backlog: ~$132.5B
🔹 Active Power: >1.7 GW; +200 MW added
🔹 Contracted Power: >4.6 GW; +900 MW added
🔹 International Power: >1.3 GW
Announcements:
🔹 CoreWeave Forge: AI development environment launched
🔹 Vera Rubin NVL72: 4.8x greater token throughput
🔹 Convertible Notes: ~$4.2B raised
🔹 DDTL 5.5 Facility: $2.6B closed
Comments:
🔸 “This was a breakout quarter for CoreWeave, with revenue increasing by more than $1 billion sequentially and adjusted operating margins reaching double digits a quarter ahead of plan.”
🔸 “approximately $9 billion in customer prepayments and deposits expected in the next three months alone.”
Andrea Paltry, Ph.D. retweeted
CoreWeave just reported an absolutely massive quarter, but there's one number that I think investors should be paying very close attention to.
The company reported $3.66 billion in Q3 revenue, up 168% year over year, while adjusted EBITDA surged 169% to $2.26 billion. Even more impressive, CoreWeave now has a revenue backlog exceeding $132 billion and raised its annualized revenue run-rate outlook to $20 billion to $21 billion. Demand for AI compute clearly isn't slowing down, and CoreWeave is positioning itself to become one of the largest infrastructure providers in the industry.
But here's where things get interesting. CoreWeave expects to spend between $35 billion and $39 billion on capital expenditures in 2026, while its quarterly interest expense alone is expected to reach $1.05 billion to $1.15 billion in Q4. That's more than $4 billion in annualized interest costs. And despite the incredible revenue growth, the company still reported an adjusted net loss of $529 million.
This is what makes CoreWeave such an interesting investment. The demand is clearly there, customers are committing billions of dollars, and the company has an enormous backlog of future revenue. But financing all this growth is incredibly expensive, and investors need to understand how much debt and capital spending it takes to generate that revenue. I still think CoreWeave has enormous long term potential, especially with AI infrastructure demand continuing to accelerate. But the biggest question isn't whether CoreWeave can grow revenue anymore. It's whether the company can eventually turn that growth into meaningful profits and free cash flow without constantly relying on additional financing.