bubble boi
Semiconductor Market Commentary, Thru Project Mascot
Nvidia's accelerator shipments can meet estimates because Rubin can replace Blackwell in existing racks, reducing the need for new US power. About 30% of shipments go abroad; if a third of the Blackwell base upgrades, only about 21 GW of new US power is needed by 2028. ERCOT additions, backup engines, converted mining sites and smaller sites should more than cover that need. The estimated share of non-Nvidia accelerators is a weaker assumption. x.com x.com x.com x.com x.com x.com
Dow Jones
Real-Time Financial News Wire
The SOX (semiconductor index) suffered its longest losing streak since 2008, and unsecured debt of neocloud provider CoreWeave fell into distressed territory. ... Yet despite recurrent setbacks amid lofty expectations, we wouldn't write off the prospects of a strong AI buildout prematurely. ... demand for AI applications and computing power should keep growing fast tradingview.com
AlvaMixedGenerali reports CoreWeave unsecured debt entering distressed territory, tightening the financing backdrop for neocloud capacity even as it expects demand for AI applications and compute to keep growing.
Bloomberg
Business and Financial Information and News Platform
Amazon has set up supplier financing arrangements to borrow as much as $80 billion against the value of hardware it buys, the company’s latest move to preserve financial flexibility as it spends hundreds of billions on AI infrastructure x.com
AlvaSupportsSupplier financing of up to $80 billion against purchased hardware strengthens Amazon's ability to fund AI infrastructure and sustain accelerator purchases.
Fireside Alpha
Business, Technology and Investing Interview Digest
AWS CEO Matt Garman ... “I think we shared recently, we’re installing an accelerator chip, whether a GPU or a Trainium chip, every 14 seconds. It is a massive and rapid scale up. And today, we’re still supply constrained, and we’re scaling really rapidly for customers.” x.com
AlvaMixedInstalling a GPU or Trainium chip every 14 seconds demonstrates rapid accelerator absorption, while continuing supply constraints limit how quickly that deployment can expand.
Hedgie
Financial Markets Newsletter and Commentary
Bloomberg says the bank is in early talks with insurers who would cover part of its losses if borrowers default. Barclays wants to hand off more than half the risk on some of its own AI loans. Some insurers are already close to the most data center risk they’ll take x.com
AlvaChallengesReports that some insurers are nearing their data-center exposure limits weaken confidence that banks can keep transferring risk to fund additional construction and accelerator purchases.Bloomberg
Business and Financial Information and News Platform
Chevron expects to power up a Microsoft data center project in West Texas on schedule despite Governor Greg Abbott’s suspension of digital infrastructure permits x.com
AlvaSupportsExpected on-schedule energization of Microsoft's West Texas project supports the capacity pipeline despite the state's permit suspension; this remains a project forecast.
Sravan Kundojjala
Semiconductor Industry Analyst at SemiAnalysis
ASIC revenue was up > 50% q/q and >2x y/y; tripled over the last three years and is guided to triple again over the next three x.com
AlvaChallengesThe reported growth and three-year tripling outlook for Intel's ASIC business strengthen the case for expanding custom-chip competition, challenging the thesis's non-Nvidia share assumption.Sravan Kundojjala
Semiconductor Industry Analyst at SemiAnalysis
2027 ASIC is now at the high end of the $16B target (20% of the AI ASIC TAM) ... Expects 2028 is to be another strong year but supply isn’t finalized ... second ASIC customer is confirmed (not from China); revenue begins in 2028 x.com
AlvaChallengesThe reported $16 billion 2027 ASIC target and second customer starting revenue in 2028 strengthen the case for greater non-Nvidia accelerator share; 2028 supply remains unfinalized.
The Transcript
Weekly Earnings-Call Quotes Newsletter
$GOOGL CFO: ... In Google Cloud, we expect TPU system sales to continue to scale in Q4, followed by a meaningful revenue ramp in 2027 x.com
AlvaChallengesA meaningful TPU system-sales ramp expected in 2027 strengthens the case for a larger non-Nvidia share of accelerator deployments.
Reuters
Global News Agency
Rick Tsai, CEO of Taiwan's largest chip designer MediaTek ... Tsai said demand for its first AI accelerator remains strong, with initial shipments starting last month to generate $2.5 billion of revenue in Q4 ... For the second AI accelerator project, high-volume production is on track for Q1 of 2028, he said tradingview.com
AlvaChallengesInitial accelerator shipments targeting $2.5 billion of Q4 revenue and a second project's planned volume production in early 2028 strengthen the case for a larger custom-chip share of accelerator deployments.Reuters
Global News Agency
PJM FILED AN AMENDED PROPOSAL TO FERC WITH THE GOAL OF COMMENCING A RELIABILITY BACKSTOP PROCUREMENT ON MARCH 1, 2027. ... THE PROCESS IS INTENDED TO FILL THE APPROXIMATELY 6,800 MW GENERATING CAPACITY SHORTFALL tradingview.com
AlvaMixedA proposed March 2027 backstop procurement offers a route to fill PJM's roughly 6.8 GW shortfall for 2028–29, while the gap weakens confidence that US generation will comfortably cover new demand.Bloomberg
Business and Financial Information and News Platform
Indian AI infrastructure firm AM Intelligence is seeking to raise as much as $3 billion to fund a previously announced purchase of Nvidia’s next-generation chips, according to sources x.com
AlvaSupportsThe reported effort to raise up to $3 billion for Nvidia chip purchases in India supports the overseas shipment pipeline; the financing remains sought rather than secured.
Walter Bloomberg
Financial News Headlines Aggregator, Not Affiliated with Bloomberg
NVIDIA IS IN TALKS TO PROVIDE NSCALE WITH BILLIONS IN CREDIT GUARANTEES FOR DEBT TO DEVELOP A 460-MEGAWATT DATA CENTER x.com
AlvaSupportsThe Information reports Nvidia is discussing billions in credit guarantees for nScale's 460 MW data center, a potential financing route for new accelerator capacity that remains under discussion.Fireside Alpha
Business, Technology and Investing Interview Digest
"AI continues to be a tailwind for Graviton because post-training, reinforcement learning, and agent tool use is mostly done on CPUs versus AI accelerators, and Graviton is the strongest CPU chip, offering up to 30 to 40% better price performance than alternatives... x.com
AlvaChallengesJassy says post-training, reinforcement learning and agent tool use are mostly CPU workloads, with Graviton offering up to 30–40% better price performance than alternatives, reducing the share of AI compute that requires Nvidia accelerators.
First Squawk
Financial Audio News Service
AMAZON CEO SAYS THE AWS BACKLOG IS $578 BLN, GROWING TRIPLE DIGITS YOY, AND THAT THE COMPANY EXPECTS TO SPEND “MEANINGFULLY MORE” ON CAPEX IN 2027 x.com
AlvaSupportsAmazon reports a $578 billion AWS backlog growing at triple-digit rates and expects to spend meaningfully more on capex in 2027, supporting continued accelerator and data-center demand.Fireside Alpha
Business, Technology and Investing Interview Digest
Amazon CEO Andy Jassy on the chip business: ... “With Trainium demand so strong, customers continue encouraging us to make it available outside of AWS. We intend to do so late in 2027.” x.com
AlvaChallengesJassy says Trainium demand is strong enough that customers want it outside AWS, with external availability planned for late 2027, strengthening the case for a larger non-Nvidia accelerator share.John Coogan
podcast
[SemiAnalysis] OpenAI's quarter is one eight gigawatt multi phase commitment in Ohio with NVIDIA backstopping 4.25 gigawatts. ... We think there's a one gigawatt TPU commitment, and OpenAI is looking for sites. podcasts.apple.com
AlvaMixedSemiAnalysis reports an eight-GW, multi-phase Ohio commitment with Nvidia backstopping 4.25 GW, alongside a possible one-GW TPU commitment, supporting substantial demand while increasing the non-Nvidia share.Ross Pamphilon
podcast
Every two years, you'll get a new release of the GPU, but they are more energy efficient, but they use a lot more energy, probably double the energy each cycle as well. podtrac.com
AlvaChallengesPamphilon says each two-year GPU generation is more energy efficient but may use roughly twice as much energy, challenging the thesis that Rubin upgrades materially reduce new power needs.Reuters
Global News Agency
AWS segment sales grew 39% year-over-year, its fastest pace in nearly five years tradingview.com
AlvaSupportsAWS growth accelerated to 39% year over year, while Amazon continues to expect about $220 billion of 2026 capex, supporting demand and spending by a major accelerator buyer.Dow Jones
Real-Time Financial News Wire
On October 29, 2026, a subsidiary of Solaris Energy Infrastructure, Inc. (the "Company") entered into an agreement to provide approximately 900 megawatts of power generation, associated balance of plant equipment and operations and maintenance services to a leading, investment-grade, diversified industrial and technology company that is an existing customer of the Company. ... Deployments under the agreement are expected to begin in the first half of 2027 and scale through 2028. tradingview.com
AlvaSupportsSolaris has contracted approximately 900 MW of generation for an existing investment-grade customer, with deployments beginning in the first half of 2027 and scaling through 2028.
Hedgeye
Independent Investment Research and Financial Media Company
Data center construction starts hit a record 11.6 gigawatts x.com
AlvaSupportsRecord construction starts of 11.6 GW strengthen the site-capacity pipeline for accelerator deployments; these are projects starting construction, rather than energized capacity.