The AI you chat with every day doesn't float in a cloud — it sits inside concrete buildings the size of several football fields, drawing as much power as a whole city. This lesson takes you to AI's 'physical body': the race to build gigawatt-scale data centers that's become the largest construction in human history — and why 'electricity,' not chips, has turned out to be the real bottleneck.
AI build-out broadens globally as debt and power strains mount
▲
Global expansion: Japan, India, Finland, Wisconsin JERA and Dell launched a $15B AI data center at a Japanese power plant, CoreWeave entered India with AdaniConneX (240MW, expandable to 480MW), Oracle subscribed to 125-250MW of nuclear power for its $15B Wisconsin campus, and Google's 507MW Finnish nuclear deal made data centers Finland's top power buyer. The build-out is spreading worldwide, locking in long-term demand for chips, power and construction.
Shows the build-out broadening beyond the US, a key new force for the theme.
Record capital and earnings confirm demand Private infrastructure fundraising hit a record $250.7B in 2025, with data centers drawing $45.7B. Comfort Systems' EPS jumped 92% on record backlog, HPE won its first $1.2B Helios AI server order from Vultr, Penguin Solutions guided to 40% growth, and Applied Digital's revenue surged 322%. UBS found AI investment creates an outsized 1.46x multiplier on US business spending. These show demand is translating into real revenue and fresh capital.
Demonstrates that capital and earnings are confirming the build-out, a core positive force.
Debt surge and financing strains raise costs Tech borrowing hit $500B in nine months, pushing 10-year Treasury yields above 5.30%, the highest since 2002, and Goldman expects $1.2T in 2027. Amazon is offloading $8B of Nvidia chips to outside investors to lighten its balance sheet, and Nvidia carries $279B in supply commitments. Rising debt costs and off-balance-sheet risk are a real counterweight that could slow the build-out.
Highlights the growing financing strain that could throttle the build-out, a key negative force.
Political and regulatory risks threaten projects JPMorgan warned a 2026 Democratic sweep could pressure AI data center stocks, which are 45% of the S&P 500. Goldman noted Texas halted new data center permits and 64% of voters oppose local projects, though it kept its 2027 US capacity forecast near 90GW. Oracle issued a force majeure notice on its New Mexico campus over permitting delays. Politics and regulation are emerging as real headwinds.
Shows political and regulatory pushback that could slow the build-out, a new negative force.
Amazon Weighs $8 Billion Outside Financing for Nvidia Chips
Amazon.com is reportedly considering moving roughly $8 billion of Nvidia Grace Blackwell chips into a special-purpose vehicle financed by outside investors and then leasing the hardware back, a structure that would make part of its AI buildout more asset-light as capital spending is expected to reach $220 billion this year. AWS revenue rose 37% to $42.2 billion in the second quarter, its fastest growth in more than four years, and contract backlog reached $496 billion, while CEO Andy Jassy said the company still lacks enough capacity to meet demand. Trailing-12-month free cash flow swung to negative $7.6 billion from positive $18.2 billion a year earlier as infrastructure spending accelerated. Amazon trades at 23.64 times forward earnings, and Tigress Financial recently raised its target to $385 from $315, arguing earnings from the investment cycle are approaching an inflection point where they grow faster than operating capital. The proposal also raises the question of residual chip value, since Grace Blackwell will eventually be superseded by Vera Rubin while Amazon assumes semiconductor generations remain useful for at least five years.
Artificial Intelligence › AI Compute & Accelerator Silicon ▲Capital
Cloud & Digital Infrastructure › Hyperscale Cloud (IaaS / PaaS) Capital
Artificial Intelligence › AI Compute Cloud & Neoclouds Capital
Artificial Intelligence › AI Data Center & Build-out Capital
AMZN · Capital · Positive Amazon weighs moving ~$8B of Nvidia chips into an outside-financed SPV and leasing them back, making its AI buildout more asset-light as capex heads to $220B.
AMZN · Demand · Positive AWS revenue rose 37% to $42.2B, its fastest growth in over four years, with $496B backlog and Jassy saying capacity still can't meet demand.
NVDA · Demand · Positive Amazon is financing and deploying roughly $8B of Nvidia Grace Blackwell chips, a concrete order for Nvidia's AI hardware.
Tigress Financial Partners, LLC · Capital · Positive Tigress Financial raised its Amazon price target to $385 from $315, arguing earnings from the investment cycle are nearing an inflection point.
GE Vernova AI Data Center Orders Top $5 Billion in First Half of 2026
GE Vernova reports that AI data center related orders exceeded US$5b in the first half of 2026, a figure management says is already more than double the data center order total recorded for all of 2025. The US based electrical group supplies equipment and services to generate, move, coordinate, convert, and store power across the US, Europe, Asia, the Middle East, and Africa, positioning it directly in the path of rising data center electricity needs. Industry analysts suggest power equipment suppliers such as GE Vernova may benefit from AI demand regardless of which developers lead the field. The company's broader story extends beyond this single development, and the projects carry exposure to large, lumpy infrastructure contracts and the possibility of delays, deferrals, or weaker pricing if customer budgets tighten. The clearest way to judge whether this read holds is to watch GE Vernova's disclosed data center backlog conversion and margins across 2027, especially how much of these AI related orders translate into on time revenue and service contracts without a visible rise in project cancellations or profit shortfalls.
Artificial Intelligence › AI Data Center & Build-out ▲Demand
GEV · Demand · Positive AI data center related orders exceeded $5B in H1 2026, more than double all of 2025, signaling strong end-customer demand for its power equipment.
AI bubble warnings mount as Dalio, Burry and Altman flag risks
A growing roster of investors, strategists and industry leaders are warning that the artificial-intelligence boom may be nearing a breaking point, with concerns shifting from the debt and interest rates funding massive data centers to the technology's potential for catastrophic harm. Ray Dalio said the bubble is nearing the point where it may burst because rising interest rates and increased borrowing to fund the AI infrastructure buildout create a need for cash, while strategist Joachim Klement said his core conviction is that the AI bubble will burst in 2027 or 2028. Michael Burry has taken short positions against Nvidia, Oracle, Palantir, Nebius, Micron and CoreWeave, arguing that enormous capital spending might not translate into durable returns, and Galaxy Digital CEO Michael Novogratz called it the biggest bubble of our lifetime while advising investors to ride the wave. OpenAI CEO Sam Altman said investors as a whole are overexcited about AI, and Sequoia's David Cahn put the annual gap between AI infrastructure spending and ecosystem revenue at nearly $600 billion, saying the bubble is reaching a tipping point. Anthropic CEO Dario Amodei called for slowing the pace of AI model improvements, with OpenAI's Sam Altman and SpaceX founder Elon Musk agreeing, after Anthropic insider Evan Hubinger put his personal estimate of the probability that AI could kill all humans at more than 10% within the next decade.
Artificial Intelligence › AI Compute & Accelerator Silicon ▼Capital
Artificial Intelligence › AI Compute Cloud & Neoclouds ▼Capital
CRWV · Capital · Negative Michael Burry has taken a short position against CoreWeave, arguing enormous AI capital spending may not translate into durable returns.
MU · Capital · Negative Michael Burry has taken a short position against Micron, arguing enormous AI capital spending may not translate into durable returns.
NBIS · Capital · Negative Michael Burry has taken a short position against Nebius, arguing enormous AI capital spending may not translate into durable returns.
NVDA · Capital · Negative Michael Burry has taken a short position against Nvidia, arguing enormous AI capital spending may not translate into durable returns.
ORCL · Capital · Negative Michael Burry has taken a short position against Oracle, arguing enormous AI capital spending may not translate into durable returns.
PLTR · Capital · Negative Michael Burry has taken a short position against Palantir, betting its AI-driven valuation will fall as the bubble bursts.
Texas Power Limits Cloud Prologis Hutto Data Center Project
Prologis's Texas data center venture with Skybox Datacenters ran into uncertainty in late September 2026 after new statewide limits on large power grid hookups raised questions about the Hutto project's timing, scale and potential rental income. The restrictions introduce project-specific uncertainty for the Hutto data center but do not materially change Prologis's near-term earnings catalyst, which still rests on lease-up, rent resets and integrating development profitably. A cluster of UK Takeover Code disclosures showed TIAA, Charles Schwab Investment Management and Dimensional Fund Advisors each reporting more than 1 percent stakes in Prologis, as the company raises equity and pursues the potential US$18.8 billion to US$19 billion SEGRO acquisition. Prologis's narrative projects $10.4 billion revenue and $3.6 billion earnings by 2029, requiring 2.5% yearly revenue growth but a decrease of about $0.6 billion in earnings from $4.2 billion today. Three Simply Wall St Community fair value estimates for Prologis span roughly US$126.95 to US$158.23, with the published forecast implying a $158.23 fair value, a 22% upside to its current price.
Artificial Intelligence › AI Data Center & Build-out ▼Regulation
PLD · Regulation · Neutral New statewide limits on large power grid hookups create project-specific uncertainty for Prologis's Hutto data center timing, scale and rental income.
PLD · Capital · Neutral Prologis is raising equity and pursuing a potential US$18.8-19B SEGRO acquisition, with fair-value estimates implying 22% upside.
Skybox Datacenters · Regulation · Neutral Skybox Datacenters' Hutto project with Prologis faces uncertainty from new statewide limits on large power grid hookups.
SGRO.LSE · Capital · Neutral Named only as the target of Prologis's potential US$18.8-19B acquisition; no standalone development reported.
AWS CEO Matt Garman Rejects AI Bubble Fears as Ray Dalio Warns Boom Near Breaking Point
Matt Garman, CEO of Amazon.com Inc.'s AWS, said concerns over an AI spending bubble are overstated, telling a16z's podcast with Raghu Raghuram on Thursday that Amazon does not rely on one customer for most AWS capacity and that "we feel really good about the spend we're making now." Garman said demand is concentrated in production workloads such as computing, storage and AI inference, and that a diversified approach means not all capacity is bundled up in one customer, comparing the market to venture capital where failed startups can be offset by successful investments. His comments come as AWS highlights efforts to make AI more affordable, and as Amazon is reportedly in talks to transfer about $8 billion worth of Nvidia Corp.'s Grace Blackwell AI chips to an investor-backed special-purpose vehicle that could raise funds through debt issuance while Amazon leases the chips back. Amazon CEO Andy Jassy raised the company's 2026 capital expenditure forecast by $20 billion to $220 billion in July, citing rising memory chip prices and strong demand for AI and AWS infrastructure. Separately, billionaire investor Ray Dalio warned that rising AI-related borrowing, higher interest rates and pressure to turn paper wealth into cash could bring the AI bubble closer to bursting, calling the boom a "classic bubble" and comparing the current AI rally to the late 1920s.
Artificial Intelligence › AI Compute Cloud & Neoclouds Demand
Artificial Intelligence › AI Compute & Accelerator Silicon ▲Demand
Artificial Intelligence › AI Data Center & Build-out ▲Demand
AMZN · Capital · Positive AWS CEO Garman defends Amazon's AI capex as sound and Jassy raised 2026 capex forecast by $20B to $220B, signaling continued heavy infrastructure investment.
NVDA · Demand · Positive Amazon is reportedly in talks to transfer about $8 billion worth of Nvidia Grace Blackwell AI chips to a special-purpose vehicle, indicating large Nvidia chip orders.
AI debt issuance halves to $23bn as investors turn wary
Global debt issuance tied to artificial intelligence fell to $23bn in September, nearly half the amount raised the previous month, according to data compiled by Morgan Stanley covering both public bonds and private placements. New financing has declined steadily since the market peaked in June, when companies raised a record $113bn for infrastructure including data centres and chips, and in the US investment-grade market, where blue-chip tech groups borrowed around $306bn between January and August, AI-related bond issuance ground to a halt last month. Morgan Stanley attributed the decline primarily to the volume of debt already raised earlier in the year, but investors are also increasingly concerned about whether such heavy investment can pay off and about growing political opposition to data centre construction. An estimated $466bn of AI-linked debt has been raised by companies in 2026 so far, up from $101bn for the same period last year, and the shift in sentiment is set to test upcoming deals, including a $60bn financing package Wall Street banks are assembling for Broadcom and Anthropic and SpaceX's talks to raise $40bn to purchase Nvidia chips. About $18bn of loans tied to Oracle's New Mexico data centre campus, known as Project Jupiter, were privately quoted at around 85 cents on the dollar in recent days after Oracle issued a force majeure notice when the site struggled to gain access to electricity.
Artificial Intelligence › AI Data Center & Build-out ▼Capital
Artificial Intelligence › AI Compute & Accelerator Silicon Capital
ORCL · Capital · Negative Loans tied to Oracle's New Mexico data centre campus were quoted at around 85 cents on the dollar after Oracle issued a force majeure notice over electricity access.
AVGO · Capital · Neutral Broadcom is part of a $60bn financing package being assembled by Wall Street banks, a deal set to be tested by the shift in AI debt sentiment.
Anthropic · Capital · Neutral Anthropic is part of a $60bn financing package being assembled by Wall Street banks, a deal set to be tested by the shift in AI debt sentiment.
NVDA · Capital · Neutral SpaceX's talks to raise $40bn to purchase Nvidia chips are cited as an upcoming deal that the weaker AI debt sentiment could test.
SPCX · Capital · Neutral SpaceX's talks to raise $40bn to purchase Nvidia chips are cited as an upcoming deal set to be tested by the shift in AI debt sentiment.
Quanta Services Sees Data Center and Tech Customers at 18% of 2026 Revenue
Quanta Services said earlier this month that it expects data center and technology customers to contribute about 18% of its 2026 revenue, up from roughly 10% a year earlier. The disclosure reinforces the company's AI-driven power theme and suggests its infrastructure business mix is tilting further toward large technology and AI-related power needs, potentially making those clients a bigger driver of future contract activity and backlog composition. The most relevant recent announcement remains Quanta's July 2026 guidance raise to US$39.3b to US$39.7b of revenue and US$1.74b to US$1.82b of net income, an outlook that already assumes strong execution on record backlog and contribution from acquisitions. Quanta's valuation has been flagged as stretched, and insiders have sold about US$123m of stock. The company's narrative projects $54.6 billion in revenue and $2.7 billion in earnings by 2029, requiring 18.4% yearly revenue growth and roughly a $1.4 billion earnings increase from $1.3 billion today, while some of the most optimistic analysts already assumed Quanta could reach about US$64.6b of revenue and US$3.2b of earnings by 2029.
Artificial Intelligence › Build-out, Construction & Engineering ▲Demand
Artificial Intelligence › AI Data Center & Build-out ▲Demand
PWR · Demand · Positive Data center and tech customers expected to contribute ~18% of 2026 revenue, up from ~10%, signaling growing end-customer demand for its infrastructure services.
PWR · Capital · Positive July 2026 guidance raise to $39.3-39.7b revenue and $1.74-1.82b net income on record backlog and acquisitions.
New Era Energy & Digital Fair Value Raised to US$11.67 on Data Center Progress
Analysts have lifted the implied fair value of New Era Energy & Digital to about US$11.67 from US$10.50, roughly an 11% increase, as Wall Street weighs progress at the company's Texas Critical Data Centers against sector and regulatory risk. Roth Capital initiated coverage with a Buy rating and a US$10 price target, saying the share price does not fully capture potential commercialization of Texas Critical Data Centers Phases I and II. Northland raised its target to US$12 from US$10 with an Outperform rating, citing progress on the Phase 1 PPA, strong major tenant interest, and management guidance around capacity delivery in 2H27, while B. Riley lifted its target to US$13 from US$10 and kept a Buy rating. B. Riley also noted that the digital mining and HPC peer group has fallen 34% since June 22, compared with 6.2% for the Russell 2000, reflecting regulatory uncertainty, higher yields, and slower leasing activity. The updated model raised the revenue growth assumption to 626.36% from 344.38%, lifted the profit margin forecast to 20.24% from 17.82%, cut the future P/E multiple to 26.88x from 75.20x, and adjusted the discount rate to 7.236% from 7.108%.
Artificial Intelligence › Colocation & Hyperscale REITs Capital
NUAI · Capital · Positive Analysts raised New Era Energy & Digital's fair value to ~US$11.67 and multiple firms lifted price targets on Texas Critical Data Centers progress.
Roth Capital Partners · Capital · Positive Roth Capital initiated coverage on New Era Energy & Digital with a Buy rating and US$10 price target.
Over 70 computing power leasing concept stocks; regulators crack down on hype-chasing and cross-sector drift
Computing power leasing is becoming one of the most crowded tracks in the A-share market. According to data from the China Academy of Information and Communications Technology, the domestic computing power leasing market reached 68 billion yuan in the first quarter of 2026, up about 60 percent year on year, and is expected to exceed 260 billion yuan for the full year. During the same period, domestic AI computing power demand grew 417 percent year on year, while supply grew only 128 percent. According to a Securities Times report in June this year, there are already more than 70 computing power leasing concept stocks, spanning 15 industries, with market capitalizations ranging from 2 billion yuan to 120 billion yuan. A recent investigation by Cailian Press found that among 16 listed companies with computing power contracts exceeding 500 million yuan, 10 had contracts that were terminated after a period of disclosure or had still not begun performance. Among them, Lanyun Technology's 3.707 billion yuan computing power cloud service agreement had still not entered actual performance as of September 8, 2026, while Annil is expected to lose up to 60.4285 million yuan after the termination of its acquisition of a 22 percent stake in Shenzhen Innovation Technology Co., Ltd. Samples of main-business drift are also dense. Qunxing Toys announced the termination of its restructuring, which had lasted nearly eight months, on the evening of October 20. Lianhua Holdings recorded computing power service revenue of 122 million yuan in 2025, accounting for only 3.53 percent of its total operating revenue of 3.452 billion yuan, while the value of computing power leasing contracts terminated early in the same period reached 1.228 billion yuan. On the regulatory front, China Securities Regulatory Commission Chairman Wu Qing made a clear statement at the 2026 Lujiazui Forum, saying that the commission will strictly investigate and punish those who use technology as a pretext to chase hot topics and speculate on concepts. So far this year, the commission has investigated and dealt with seven cases of misleading statements. On July 6, the Shanghai Stock Exchange also upgraded the permission functions of its SSE e-interaction platform, suspending reply permissions for 12 months for companies penalized for hype-chasing.
002575.CS · Capital · Negative Qunxing Toys announced termination of its nearly eight-month restructuring, a failed deal tied to the computing-power leasing hype
002875.CS · Capital · Negative Annil expects to lose up to 60.43 million yuan after terminating its acquisition of a 22% stake in Shenzhen Innovation Technology
IREN's Kent Draper Discusses AI Infrastructure Demand and Physical AI at All-In Summit
IREN Chief Commercial Officer Kent Draper said at the All-In Summit in Los Angeles that the company operates as a vertically integrated AI cloud provider, delivering compute services through company-owned data center infrastructure. Speaking to NYSE Live, Draper said controlling the underlying physical assets lets IREN support customers across a range of AI workloads while expanding capacity to meet growing demand. He identified land, power, labor, supply chains and execution as the central challenges in scaling AI infrastructure, calling the addition of new capacity one of the most important issues facing providers. Draper also pointed to evolving financing markets and growing capital availability for compute and data center expansion, tied to customer commitments and changing financing structures. He cited manufacturing, research and development, drug discovery, enterprise operations and customer-facing technologies as expanding real-world use cases, and highlighted physical AI, including robotics and AI-enabled manufacturing, as an area attracting increased attention.
IREN · Demand · Positive IREN's CCO describes growing customer demand for its vertically integrated AI cloud compute and expanding capacity to meet it.
AZIO AI Signs First Atlas One Hosting Deal Worth Up to $450M
AZIO AI Holdings announced a power purchase and hosting agreement with Power Champion Investment Limited at its Atlas One compute infrastructure campus in South Texas. The deal establishes an initial contracted capacity of 10 megawatts, with expansion rights up to 50 MW, and at that initial capacity the company expects roughly $1.5 million in monthly capacity reservation charges, or $90 million over the five-year initial term, excluding electricity sales and additional services. Full expansion to 50 MW could generate approximately $450 million in capacity reservation charges over the same period. The agreement is the first hosting contract at Atlas One, a 548-acre campus designed for up to 500 MW of planned power capacity, where approximately 11 MW has been secured to date, including 6 MW currently activated. No capacity has yet been energized under the new agreement, expansion beyond the initial 10 MW has not been exercised, and service commencement remains subject to infrastructure development, required U.S. export control approvals and other contractual conditions.
Super Micro Computer Appoints Intel Veteran Shesha Krishnapura to Board
Super Micro Computer has appointed Shesha Krishnapura, a 35 year Intel veteran, as an independent member of its Board, replacing departing director Judy Lin. Krishnapura is known for his work in high performance computing and AI infrastructure, and his operational background in designing and running large, power hungry data centers lines up with the company's push into high density AI racks, liquid cooling, and large AI factory style deployments. The appointment follows recent Super Micro Computer partnerships, including an AI focused collaboration with NetApp on data infrastructure solutions. Super Micro Computer sells modular server and storage hardware used in data centers across the US, Asia, Europe, and other regions, and carries a market cap of $29.5b. The key marker to track is whether the company shows progress on its stretched 149 day cash conversion cycle in upcoming quarterly reports, especially as new AI partnerships and large orders flow through.
Artificial Intelligence › AI Data Center & Build-out ▲Talent
Artificial Intelligence › AI Server OEM & System Integration ▲Talent
SMCI · Capital · Positive Super Micro appoints an Intel HPC/AI data-center veteran to its board, strengthening governance and AI infrastructure expertise.
Oracle Trucks Natural Gas to Data Centers as Pipeline Delays Bite
Oracle is trucking natural gas directly to its data centers to keep construction on schedule, a stopgap measure it is considering for a build in New Mexico where a needed gas pipeline is delayed. The company is already running 30 trucks a day to data centers outside Salt Lake City, according to Bloomberg reporting. Oracle did not respond to requests for comment, but later posted on social media praising the partner helping it carry out the effort. The news added to pressure on Oracle shares, which had already been sliding amid confusion over OpenAI's ARR figure, though the stock was up almost 5% on the day. Analysts said the move signals the delays are more significant than the market expected, since trucking gas is a measure normally reserved for remote mining or temporary industrial operations.
Energy Transition & Power Demand › Natural Gas Value Chain Supply
ORCL · Supply · Negative Pipeline delays force Oracle to truck natural gas to data centers, a costly stopgap signaling significant infrastructure constraints on its buildout.
Senao Networks Launches x86 Enterprise Server Portfolio for AI
Senao Networks Inc. announced the launch of its new x86 enterprise server portfolio, spanning data center and edge deployments with NVIDIA RTX GPU support in selected configurations. The portfolio comprises five platforms: the SR810 and SR710 for high-performance enterprise compute, the SR610 for mainstream enterprise compute, and the SE210 and SE110 for edge intelligence. The systems are designed to support generative AI, AI inference, model fine-tuning, data analytics, visualization, and intelligent edge computing workloads. Dr. Bou Lin, President at Senao Networks Inc., said the portfolio provides flexible computing building blocks that let customers deploy AI according to their own performance, security, and infrastructure requirements, from the data center to the edge. The launch expands Senao Networks' enterprise computing offering beyond its networking, security appliance, and manufacturing business, with product details available on the Senao Computing website.
KONST Raises $30M Series B Led by ADATA Technology
KONST, a Taiwan-based next-generation AI compute operator, has raised $30 million in Series B funding led by ADATA Technology, with strategic participation from MMobility, Pegatron Venture Capital, and several other corporate investors. The company said it will use the proceeds to scale up its AI data center buildout across Asia and accelerate its Token Factory strategy, which treats compute as production equipment and tokens as its output. That strategy rests on three layers: Konstra AI handles compute supply, from site selection and mechanical, electrical, and environmental control systems through GPU cluster tuning and operations; Glows.ai provides the conversion layer, breaking cluster compute into smaller units billed per second; and Horizon AI's (k) ATP Token handles governance and delivery, letting enterprises connect to every authorized model with a single project key and track usage through audit logs. KONST said the round's principal investors come from memory and storage, electrification solutions, and electronics manufacturing, while other participants span energy, semiconductors, and thermal management. Co-founder and chairman Ben Chang said the round is about more than the amount raised, adding that once AI becomes part of operations, computing power becomes essential infrastructure rather than a one-time project expense.
Artificial Intelligence › AI Data Center & Build-out ▲Capital
Artificial Intelligence › AI Compute Cloud & Neoclouds ▲Capital
Artificial Intelligence › AI Tooling, Data & MLOps ▲Capital
Artificial Intelligence › Build-out, Construction & Engineering ▲Capital
3260.TWO · Capital · Positive ADATA Technology led KONST's $30M Series B, a strategic investment in an AI compute operator.
KONST · Capital · Positive KONST raised $30M Series B to scale its AI data center buildout and Token Factory strategy.
Glows.ai · Demand · Positive Glows.ai is KONST's conversion layer providing per-second billed compute, benefiting as KONST scales its AI data center buildout.
AMD Data Center Revenue Doubles to $6.7 Billion as Helios Rack Deployments Ramp
Advanced Micro Devices reported fiscal Q2 2026 Data Center revenue more than doubled, up 107% to $6.7 billion, a segment that now accounts for 58% of company sales. The company said Meta, Microsoft, OpenAI, Oracle, and Anthropic are among those deploying its Helios rack system, which combines Instinct GPUs, EPYC CPUs, and networking, and a recent partnership with Anthropic covers up to 2 gigawatts of MI450 GPUs in Helios racks. Non-GAAP gross margin reached 56% and non-GAAP operating margin hit 27%, helped by an easy comparison against the year-ago quarter's $800 million in charges tied to US export controls on a data center GPU, while non-GAAP earnings per share came in at $1.66 versus $0.48 a year earlier. Management guided fiscal Q3 revenue to about $13 billion, plus or minus $300 million, and expects Data Center sales to accelerate in the second half, though the largest commitments, including Anthropic's, are multiyear deployments rather than revenue already booked. Client revenue rose 23% to $3.1 billion, gaming fell 31% to $779 million on weaker semi-custom sales, capital expenditures reached $808 million, and free cash flow was $1,558 million, down from $2,566 million in fiscal Q1. Investors are paying 85.60 times forward earnings as of October, against a sector multiple of 23.74 and AMD's own five-year average of 39.90, with earnings per share expected to grow 107.2% in 2027.
AMD · Capital · Positive AMD's fiscal Q2 2026 Data Center revenue more than doubled to $6.7 billion, with non-GAAP EPS of $1.66 vs $0.48 and Q3 revenue guided to ~$13 billion.
AMD · Demand · Positive Meta, Microsoft, OpenAI, Oracle, and Anthropic are deploying AMD's Helios rack system, including an Anthropic partnership covering up to 2 gigawatts of MI450 GPUs.
Nvidia Revenue More Than Doubles as Vera Rubin Ramps and Q3 Guide Hits $108 Billion
Nvidia reported a quarter in which revenue more than doubled, with Data Center revenue reaching $89.0 billion in fiscal Q2 2027, up 117% from a year earlier, while gross margin held at 75.0% on both a GAAP and non-GAAP basis. Management guided fiscal Q3 revenue to $108.0 billion, plus or minus 2%, up from the $96.2 billion just reported, and said the next-generation Vera Rubin platform is in full production with racks running at partners including CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius. Nvidia also announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR on financing platforms meant to mobilize over $500 billion of outside capital, subject to definitive agreements, and returned about $26.0 billion through buybacks and dividends in the quarter, backed by a remaining share repurchase authorization of nearly $99 billion. The company guided fiscal Q3 gross margin to 74.0%, plus or minus 50 basis points, a modest step down from the 75.0% it just delivered, and the outlook assumes no data center compute revenue from China. On forward earnings, Nvidia trades at 25.67 times, against a sector multiple of 23.77 and its own five-year average of 42.20, while analysts expect earnings per share to grow 69.66% in 2027.
Artificial Intelligence › AI Compute Cloud & Neoclouds ▲Demand
Artificial Intelligence › AI Data Center & Build-out ▲Demand
NVDA · Capital · Positive Nvidia's revenue more than doubled with Data Center revenue at $89.0B and Q3 guidance of $108.0B, plus $26B returned via buybacks and dividends.
NVDA · Demand · Positive Vera Rubin platform is in full production with racks running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius.
Gorilla Technology Signs First U.S. Data Center LOI, Targets 100 MW Portfolio
Gorilla Technology announced Friday that it signed a non-binding letter of intent to acquire its first U.S. data center, with the acquisition targeted to close by the end of November 2026. The existing facility has 6 MW of current power capacity and potential expansion to 36 MW, while the local utility supports an additional 30 MW expected to be delivered in about 18 months, subject to infrastructure work. The operational facility generates colocation revenue that the company expects to continue after the acquisition, subject to final lease and transfer arrangements. Gorilla targets a U.S. data center portfolio exceeding 100 MW within 12 to 18 months through acquisitions, expansion, and AI computing deployments. The deal remains subject to due diligence and definitive agreements.
GRRR · Capital · Positive Gorilla Technology signed a non-binding LOI to acquire its first U.S. data center, targeting a 100 MW portfolio via acquisitions and AI computing deployments.
XMax to acquire Hexa Creation in AI power infrastructure push
XMax announced on Friday that it agreed to acquire 100% of Hexa Creation, extending its AI strategy into power infrastructure aimed at AI data centers, electric vehicles, energy storage, and ultra-fast charging. Hexa Creation holds exclusive rights to university-owned intellectual property underpinning its high-voltage GaN platform. Yole Group projects the global GaN power device market will grow to $3B by 2030 from $355M in 2024, a 42% CAGR.
Suchat Pushes for CoP Draft to Regulate Data Centres on Environmental Standards
Suchat Chomklin, Minister of Natural Resources and Environment, said after the first meeting in 2026 of the environmental subcommittee under the Data Centre Business Policy Committee that the government is pushing a draft Code of Practice, or CoP, for data centre operators to serve as an environmental regulatory framework and reduce impacts on nearby residents and communities. The draft CoP will set minimum standards that operators must follow, covering electricity use, water management, environmental management, and the storage and management of reserve fuel, in order to raise industry standards and reduce risks from potential impacts on communities and the environment. Suchat said the development of digital infrastructure is a key factor in boosting the country's competitiveness, but the growth of data centres must go hand in hand with environmental responsibility, with the government setting clear standards from the outset. He also stressed the need for clear operating guidelines, continuous communication of information to the public, and integrated work with relevant agencies. After hearing opinions from relevant agencies and experts, the draft CoP will be revised for completeness before being submitted to the Data Centre Business Policy Committee for consideration in the next step.
Michael Burry Warns of Strain in Private Equity and Private Credit Markets
Michael Burry has warned that private-equity and private-credit markets are showing signs of strain, raising concerns about insurer assets and the financing behind the AI data center buildout. In an October 8 post on his Substack newsletter, Burry said he was hearing "creaks and groans" and "some loud snapping and crackling" in private-market credit, pointing to valuation marks and insurers' role in financing AI infrastructure and warning that rising long-term interest rates could add pressure before data center projects generate returns. The investor known for predicting the 2008 financial crisis also questioned the scale of AI spending, arguing that advances in AI compression could reduce computing requirements and leave some infrastructure investment as "sunk cost," while companies' use of smaller, cheaper AI models, including Chinese open-weight models, could further challenge assumptions about future demand for computing capacity. In an earlier post, Burry said, "We are somewhere in there, hard to say where," referring to the market's stage in a potential downturn, and warned that private equity and private credit were "getting torn apart under the surface of the calm" and that insurers were "holding too many bad assets." Notably, Burry held a bearish position in MetLife through long-dated put options, with a portfolio update listing MET puts expiring in 2029.
Firmus pulls $5 billion IPO, a blow to Australian capital markets
Investors say the scrapping of a $5 billion initial public offering by Australian data centre operator Firmus, which is backed by Nvidia, is a blow to the country's capital markets. According to Dealogic data, the IPO would have been the fourth-largest in the world so far this year and the second-largest in the history of the Australian Securities Exchange, behind Telstra's $10 billion listing in 1997. Oscar Oberg, lead portfolio manager at Wilson Asset Management, said the shortage of new listings has been a very frustrating situation for us for a long time. According to Jamie Hanna, deputy head of investments and capital markets at VanEck Australia, the ASX has been steadily losing listed companies to private takeovers, particularly in the infrastructure sector, with the number of listed companies falling to 1,891 as of September 2026 from 2,066 in 2016. Hanna said the failure of these large listings to materialise is a blow to the market.
Nvidia-backed Firmus cancels $5B Australian IPO, citing market volatility
Australian AI data center operator Firmus, which is backed by Nvidia, has withdrawn its planned $5B initial public offering, citing market volatility and conditions. The company said its board determined the terms of the proposed offering did not adequately reflect the strength of its business and long-term growth outlook, and concluded that proceeding was not in the best interests of the company and its shareholders. Firmus had planned to raise $5B in the IPO, pricing shares at A$11 apiece, which would have made it the second-largest new share sale in Australia's history and valued the company at around $30.6B. The company will now pursue capital from the private markets and consider alternative public and private market options. Firmus unveiled a $2B funding round backed by Nvidia, Coatue Management, Blackstone, and Jane Street in August, bringing its total equity raised over the preceding year to more than $3B and its valuation to over $10.5B. Investors began pulling their orders on Wednesday as they lost confidence in the deal, after being told on Tuesday about escrow arrangements that would have allowed more than half the stock to be sold by existing investors from day one.
Artificial Intelligence › AI Data Center & Build-out Capital
Artificial Intelligence › AI Compute Cloud & Neoclouds Capital
Firmus Technologies · Capital · Negative Firmus withdrew its planned $5B Australian IPO, citing market volatility and weak terms, and will now seek private capital instead.
Kasikorn Thai says World Bank urges Thailand to accelerate AI adoption, supporting 8 stocks in industrial estates, power, and electronics
Kasikorn Securities stated that the World Bank recommends Thailand accelerate its use of AI to boost productivity and escape the middle-income trap, after finding that only 12% of Thai businesses use AI, compared with 43% in the United States. The government should urgently develop workforce skills, establish clear regulations, and ensure energy security for data centers. Meanwhile, Thailand's manufacturing base, services sector, and AI hardware supply chain are key strengths. The brokerage views this as structurally positive for the industrial estate, utilities, power plant, digital infrastructure, and electronics sectors, such as WHA, AMATA, WHAUP, GULF, BGRIM, DELTA, KCE, and HANA.
Artificial Intelligence › AI Data Center & Build-out ▲Demand
AMATA.BK · Demand · Positive World Bank-backed AI adoption push is seen as structurally positive for Thailand's industrial estate sector, where Amata operates.
BGRIM.BK · Demand · Positive Brokerage cites the AI/data-center drive and energy-security needs as structurally positive for Thai power plant and utilities names like B.Grimm.
DELTA.BK · Demand · Positive Thailand's AI hardware supply chain and electronics sector are named key strengths benefiting from accelerated AI adoption, supporting Delta Electronics.
GULF.BK · Demand · Positive World Bank call to ensure energy security for data centers is viewed as structurally positive for Thai power and utilities, including Gulf Energy.
HANA.BK · Demand · Positive Thailand's AI hardware supply chain and electronics sector are cited as key strengths, supporting Hana Microelectronics.
KCE.BK · Demand · Positive World Bank-backed AI adoption push is seen as structurally positive for Thailand's electronics/AI hardware supply chain, which includes KCE.
Arthur Hayes Warns AI Crash and Bailout Could Fuel Bitcoin's Best Run
Arthur Hayes, co-founder of BitMEX and crypto investment firm Maelstrom, said excessive AI investment could eventually trigger a crash and a government bailout that sends Bitcoin and other cryptocurrencies higher. Speaking to CNBC at the Gamma Prime Investing Conference in Singapore on Tuesday, Hayes said humanity is wasting multi-trillion dollars building AI data centers, and that the construction surge will first create extremely plentiful, inexpensive computing power before producing overcapacity, a downturn and eventually government-backed support for the industry. He pointed to Nvidia and other memory-chip manufacturers that are already profitable, urging investors to assess whether current valuations adequately reflect expected future earnings, while noting that SpaceX, OpenAI and Anthropic rank among major computing-power users despite not making money. Hayes said pressure could surface in late 2027 or 2028, when new capacity comes online and infrastructure providers seek payment from AI companies, and told CNBC that Bitcoin and other cryptocurrencies could soak up excess liquidity after such a bailout, potentially making them the strongest-performing assets. The comments build on his August prediction that Bitcoin could reach $250,000 as governments increase liquidity to address credit-market stress, rising bond yields and AI's enormous funding requirements.
Bitdeer AI Secures Over 70% Offtake for Malaysia AI Cloud Data Center
Bitdeer Technologies Group said its Bitdeer AI unit has secured over 70% offtake commitments for its upcoming AI Cloud data center in Malaysia. The Johor Bahru site has more than 70% of its roughly 21.7MW AI Cloud capacity spoken for, representing about US$1.7b in expected revenue over 5 years, part of a US$2.9b total expected revenue backlog. Bitdeer AI is procuring NVIDIA GB300 NVL72 systems to equip the facility, aiming for faster deployment and scalable GPU capacity. The key test will be energization and ramp up at A201 in the first quarter of 2027, with GPUs installed and customers drawing on the contracted capacity. For a group historically centered on Bitcoin self mining and hosting, the contracted AI backlog helps shift its mix toward recurring infrastructure-style revenue tied to GPUs and data centers.
Artificial Intelligence › AI Data Center & Build-out ▲Demand
Artificial Intelligence › AI Compute Cloud & Neoclouds ▲Demand
BTDR · Demand · Positive Bitdeer AI secured over 70% offtake commitments for its Malaysia AI Cloud data center, representing ~US$1.7b in expected revenue over 5 years.
NVDA · Demand · Positive Bitdeer AI is procuring NVIDIA GB300 NVL72 systems to equip the Johor Bahru facility, a concrete GPU order.
SpaceX's $40 Billion AI Chip Financing Tests Investor Appetite for AI Debt
Space Exploration Technologies Corp. is reportedly planning a $40 billion financing package to fund purchases of Nvidia Corp chips, a deal expected to close in 2027 that would include about $10 billion of bank loans and $30 billion of investment-grade debt. The plan arrives only months after SpaceX raised $25 billion through its first bond sale, following $85.7 billion in net proceeds from its June IPO, according to its SEC filing. SpaceX reported $38.4 billion of total debt as of June 30, up from $22 billion at the end of 2025, with its June bond deal carrying interest rates from 5.35% to 6.65% and a weighted average rate of 5.855%. The trend extends beyond SpaceX: Oracle Corp said it expected to raise $45 billion to $50 billion during 2026 through debt and equity to expand Oracle Cloud Infrastructure, while CoreWeave closed a $2.6 billion loan facility backed by customer commitments in August. Neuberger Berman estimated that hyperscaler, data-center and semiconductor financing had reached roughly $165 billion before the midpoint of 2026, about $27 billion more than the entire 2025 total, making the pricing and demand for the next wave of AI-related debt the key signal for investors.
Artificial Intelligence › AI Compute & Accelerator Silicon ▲Capital
Artificial Intelligence › AI Data Center & Build-out ▲Capital
Artificial Intelligence › AI Compute Cloud & Neoclouds ▲Capital
SPCX · Capital · Neutral SpaceX is the subject, planning a $40B financing package after prior bond sale and IPO proceeds, but the debt raise is a mixed signal.
NVDA · Demand · Positive SpaceX's $40B financing is intended to fund purchases of Nvidia chips, a concrete end-customer order driver.
CRWV · Capital · Neutral Mentioned only as another example of AI-related debt financing (closed a $2.6B loan facility), no new company-specific development.
ORCL · Capital · Neutral Cited only as context for the broader AI-debt trend, noting its planned $45-50B 2026 raise for OCI.
WHA Expects Q4 2026 to Be the Year's Peak, Supporting Target of 2,500 Rai in Land Sales
Jareeporn Jarukornsakul, Chief Executive Officer of WHA Corporation Public Company Limited, or WHA, disclosed that operating results in the fourth quarter of 2026 will improve markedly from the third quarter of 2026 and will mark the highest growth period of the year, driven by a substantial volume of industrial estate land sales awaiting transfer and revenue recognition, together with support from asset sales into an investment trust. The company plans to sell assets into the WHA Premium Growth Real Estate and Leasehold Investment Trust, or WHART, worth more than 2.5 billion baht, with transfer and revenue recognition expected in the fourth quarter of 2026, and is in the process of filing with the Securities and Exchange Commission, or SEC, to seek approval for additional asset sales into the WHA Industrial Leasehold Real Estate Investment Trust, or WHAIR. For 2026, the company is maintaining its land sales target at approximately 2,500 rai, after selling about 504 rai in the first half of this year, not yet including a large volume of sales awaiting transfer. Meanwhile, the company must temporarily slow the intake of new customers pending clarity on government policy criteria and promotional measures for data center projects, which, if clarified, would help support the investment climate and land demand in WHA's industrial estates in the future. Its main customer base consists of global multinational technology groups with strong financial standing. Related agencies forecast that in 2026 total investment promotion applications will reach as much as 1.8 trillion baht, the highest since the BOI was established, of which 1.36 trillion baht will be foreign investment.
WHA.BK · Capital · Positive Q4 2026 expected to be the year's peak, driven by asset sales into WHART worth over 2.5 billion baht plus SEC filing for additional asset sales into WHAIR.
WHA.BK · Demand · Positive Maintains 2026 land sales target of ~2,500 rai with a large volume of industrial estate land sales awaiting transfer and revenue recognition.
Tianyang Technology's official WeChat account discloses purchase of 20,000 GPUs; order landing remains uncertain
Tianyang Technology announced before market open today via its official WeChat account that it has signed a GPU Procurement Agreement with Company X to purchase 20,000 NVIDIA RTX PRO 5500 Blackwell units. Boosted by the news, the company's closing price that day was 16.16 yuan per share, up 12.38% from the previous trading day. A Cailian Press reporter interviewed multiple industry insiders and learned that the NVIDIA RTX PRO 5500 Blackwell has not yet officially gone on sale, and its official pricing has not been determined. The domestic quote for this GPU may exceed 100,000 yuan per unit. Based on a purchase volume of 20,000 units, the aforementioned agreement could be worth more than 2 billion yuan, but practitioners bluntly said the order has barely begun to take shape, and getting the goods is estimated to happen no earlier than the first quarter of next year. Xia Hailong, a lawyer at Shanghai Shenlun Law Firm, said a preliminary estimate puts the order amount at potentially several billion yuan. If it reaches the financial threshold requiring disclosure, it should be disclosed through statutory channels in accordance with the law, and publishing it only through a WeChat official account is clearly non-compliant. A representative from Tianyang Technology's securities department responded that specific details are inconvenient to disclose, and if the relevant disclosure threshold is met, there will be subsequent disclosure. Previously, Tianyang Technology had already disclosed large-scale computing power projects, including a project cooperation agreement signed in May with Qiming Xinghan and others for an overall computing power procurement scale in the range of 3.5 billion to 4 billion yuan, a joint venture established in June with Capital Online with a capital contribution of 70 million yuan for a 70% stake, and a Computing Power Cloud Service Agreement signed with Yunli Zhihui with a total tax-inclusive amount of 303 million yuan, which was raised to 402 million yuan in July. The semi-annual report shows that as of the end of June this year, the company's fixed assets were approximately 275 million yuan. The two major contracts with Qiming Xinghan and Yunli Zhihui recorded zero sales revenue recognized in the current period and cumulatively. In the first half of the year, non-GAAP net profit was negative 6.9296 million yuan, turning to a loss year-on-year. Net cash flow from operating activities was approximately negative 341 million yuan. Monetary funds were approximately 508 million yuan, and short-term borrowings were approximately 615 million yuan.
Aofei Data Chairman Feng Kang Proposes Buyback of 50 Million to 100 Million Yuan in Shares
Aofei Data announced on October 9 that its actual controller and chairman, Feng Kang, proposed the company use its own funds or self-raised funds to repurchase part of its RMB ordinary shares through centralized bidding. The buyback amount will be no less than 50 million yuan and no more than 100 million yuan, with a buyback period of 12 months from the date the board approves the plan. In the first half of 2026, Aofei Data achieved revenue of 1.469 billion yuan and net profit attributable to the parent of 197 million yuan.
Xingyun Technology expects Q3 net profit to surge over 5x, turns profitable in first three quarters
Xingyun Technology disclosed its earnings forecast on the evening of October 8, expecting attributable net profit of 240 million to 290 million yuan for the first three quarters of 2026, swinging from a loss to a profit year-on-year. Based on this estimate, third-quarter net profit alone is projected at 228 million to 278 million yuan, up 569% to 716% from 34 million yuan in the second quarter. The company said demand in the AI computing infrastructure market remains strong, and its computing business, especially server sales, achieved explosive growth, becoming the core driver of the significant improvement in revenue and net profit and the turnaround to profitability. Headline customers including V Client, VB Client, and VC Client negotiated upward adjustments to lease prices for existing signed orders. The first batch of server clusters delivered and deployed for these three customers officially began leasing and recognized lease revenue from August. During the reporting period, the company and its subsidiaries applied for total credit lines of 21.544 billion yuan from financial and non-financial institutions, and signed financing and credit contracts totaling 6.589 billion yuan. In addition, the company launched its first equity incentive plan for 2026, requiring amortization of incentive expenses of about 58 million yuan. Excluding the impact of this share-based payment amortization, the profitability of its core computing business improved significantly. As of September 21, 2026, the company's announced five-year long-term computing framework orders on hand reached 16.004 billion yuan.
Amazon to Deploy Two Million NVIDIA GPUs in 2027-2028, Invest $1.9 Billion in Driver Pay
Amazon.com Inc. plans to deploy two million additional NVIDIA GPUs during 2027 and 2028 as part of its AI infrastructure buildout, alongside a $1.9 billion investment in its Delivery Service Partner program in 2027 that is mostly directed toward driver wages. That driver-pay commitment, which Amazon says will lift driver pay to a national average of nearly $24 an hour, brings total funding for the eight-year-old program to $21.7 billion. The company also completed a £4.25 billion sterling bond offering on September 14, 2026, across four maturities of 3, 6, 12, and 19 years, broadening its financing sources as AI spending raises capital requirements. In the second quarter, Amazon posted $200.6 billion in total revenues, up 20% from Q2 FY25, with AWS revenue jumping 37% year-over-year to $42.2 billion, while 2026 capital outlays are projected at close to $220 billion and free cash flow has slipped into negative territory. Separately, Amazon is raising base starting wages for qualifying full-time employees in core U.S. operations by $1 to $20 an hour effective September 27, a move it says reflects a commitment of over $1.5 billion.
AMZN · Capital · Positive Amazon plans $220B 2026 capex and a £4.25B bond offering to fund AI infrastructure, alongside $1.9B driver-pay investment and $1 wage hike.
AMZN · Demand · Positive AWS revenue jumped 37% year-over-year to $42.2B, showing strong end-customer demand for its cloud services.
NVDA · Demand · Positive Amazon will deploy two million additional NVIDIA GPUs in 2027-2028, a concrete order for NVIDIA's AI chips.
Applied Digital Q3 Revenue Jumps 322% to $341.9 Million, Beating Estimates
Applied Digital beat Wall Street's revenue expectations in calendar Q3 2026, with sales up 322% year on year to $341.9 million against analyst estimates of $135.3 million. The digital infrastructure provider's non-GAAP loss of $0.01 per share was 96.7% above consensus, while adjusted EBITDA of $64.41 million beat estimates of $39.79 million at an 18.8% margin. Management attributed the quarter to rapid expansion of its AI-focused data center campuses in North Dakota, which delivered 150% more capacity year over year, and to a new long-term power purchase agreement with Base Electron securing 1,200 megawatts of natural gas generation. CEO Wes Cummins said the company expects approximately 250 megawatts of expansion leases to be executed by calendar year-end at materially higher pricing compared to prior leases. ChronoScale, in which Applied Digital holds a majority stake, announced a significant AI compute deployment with Microsoft, and the company entered Finland as a measured step into Europe with initial power available in 2028.
APLD · Capital · Positive Applied Digital beat revenue estimates with sales up 322% to $341.9M and adjusted EBITDA of $64.41M, well above consensus.
APLD · Demand · Positive Rapid expansion of AI-focused data center campuses drove 150% more capacity and ~250MW of expansion leases expected at materially higher pricing.
EKSO · Demand · Positive ChronoScale, majority-owned by Applied Digital, announced a significant AI compute deployment with Microsoft.
Base Electron · Supply · Positive Applied Digital signed a new long-term power purchase agreement with Base Electron securing 1,200 megawatts of natural gas generation.
ADVANC says iPhone sales outstrip supply, invests 50 billion baht a year in technology
Pratthana Leelapanang, Chief Executive Officer of Advanced Info Service Public Company Limited, or ADVANC, said demand for iPhones among upper-tier consumers remains so high that supply cannot keep up, even though imports rose 30-40% from a year earlier, while middle- and lower-tier consumers are tending to hold back on buying new mobile phones. The company is maintaining its operating forecast for 2026, expecting core service revenue to grow 3-5% and EBITDA to grow 2-4%, and has set a target of investing about 50 billion baht a year. That breaks down into roughly 30 billion baht a year to expand its main communications network, about 10 billion baht a year in Data Center business investment through its subsidiaries, and about 10 billion baht a year in spectrum investment, in order to drive the digital economy and Data Center, Sovereign Cloud and Sovereign AI technologies. It expects that by 2027 Intelligence and AI technology will generate as much as about 500 billion baht in added economic value, or more than 2% of GDP. As for its partners, AIS is a subsidiary of GULF and works with Singtel. The GSA project is divided into GSA01, GSA02 and GSA03, with combined power generation capacity of nearly 300 megawatts to support Data Center growth. At present, demand for investment in Thailand waiting to begin construction is no less than 2 gigawatts, and may reach 3 gigawatts, from more than 30 foreign operators.
Nvidia-backed Firmus said to delay Australia IPO, weigh private funding round
Nvidia-backed AI infrastructure startup Firmus Technologies is expected to delay its planned initial public offering in Australia, according to a Bloomberg report citing people familiar with the matter. Firmus is instead in talks with existing investors and others for a private funding round, though order-taking for the IPO closed as scheduled on Thursday morning with no clear indication of the price or deal structure. Deliberations are ongoing and details could still change, the people said, and a spokesperson for Firmus did not immediately respond to a Bloomberg request for comment. Bloomberg reported last month that Firmus was in talks to raise about $10 billion in financing ahead of its planned Australian IPO. Firmus began as a Bitcoin miner in Tasmania in 2019 and now operates seven AI factories across Australia, Singapore, Indonesia, and Malaysia.
Artificial Intelligence › AI Data Center & Build-out Capital
Artificial Intelligence › AI Compute Cloud & Neoclouds Capital
Firmus Technologies · Capital · Negative Firmus is delaying its planned Australian IPO and pivoting to a private funding round, a negative financing development.
Oracle stock falls as debt-funded AI chip buying and OpenAI revenue shortfall weigh
Oracle stock fell on Thursday after a Wall Street Journal report said the enterprise cloud provider is preparing to tap debt markets to fund its AI infrastructure build-out. Oracle is reportedly in talks with Apollo and Goldman Sachs to arrange financing and wants a deal done as soon as this year, with investors likely funding a separate company that buys the chips and leases them to Oracle over time so the company avoids borrowing more money itself. Broadcom and SpaceX were also mentioned by the Journal as pursuing similar deals. Oracle shares are down nearly 30% this year, and its capital spending hit $28.5 billion in the June-to-August quarter, up from $2.3 billion in the same quarter two years earlier. Adding further pressure, OpenAI's annualized revenue is reportedly $20 billion short of prior estimates; OpenAI is one of Oracle's biggest cloud customers, and the contract backlog stood at $664 billion last quarter. Oracle has nearly doubled its long-term debt to more than $160 billion over the past two years, according to Bloomberg, ranking it as the fifth-largest borrower in the US corporate bond market.
ORCL · Capital · Negative Oracle is preparing to tap debt markets and use off-balance-sheet financing to fund its AI chip build-out, with capex ballooning to $28.5B and long-term debt nearly doubling to over $160B.
ORCL · Demand · Negative OpenAI, one of Oracle's biggest cloud customers, is reportedly $20B short of prior annualized revenue estimates, threatening the demand behind Oracle's $664B contract backlog.
Box CEO Levie Says AI Compute Needs Will Be 'Insane' as Nvidia, Apple, Microsoft Top 21% of S&P 500
Box co-founder and CEO Aaron Levie said the compute required for the next stage of AI will be "insane," citing personal agents, agent swarms defending enterprises, code-review agents, and background agents working around the clock. In a post on X, Levie said this will require inference volume growing by orders of magnitude plus the computers, networking, and file systems agents need, adding that the buildout is only in its early stages. His comments came in response to a fellow tech executive calculating the vast computing needs of Meta's new Muse AI agent. The remarks landed during a headline-making week for AI-centric tech names: Nvidia touched a record high, AMD hit a record high, and Marvell Technology lifted its revenue outlook by billions of dollars while striking a bullish note on its long-term total addressable market. Meanwhile, the S&P 500 has never been more concentrated in three stocks, with Nvidia, Apple, and Microsoft representing over 21% of the index, according to data from Creative Planning, far above the 13.4% peak share held by IBM, AT&T, and ExxonMobil in the mid-1980s. Yorkville Ives partner Dan Ives called the AI Revolution the largest technology buildout in history, saying it is no longer a story about one sector but about the capital, power, and policy of the entire economy.
Artificial Intelligence › AI Compute & Accelerator Silicon ▲Demand
Artificial Intelligence › Agentic AI & Autonomous Workflows ▲Demand
Artificial Intelligence › AI Data Center & Build-out ▲Demand
Artificial Intelligence › AI Applications & Copilots ▲Demand
Artificial Intelligence › AI Compute Cloud & Neoclouds ▲Demand
BOX · Demand · Positive Box CEO Levie's remarks that AI compute needs will be 'insane' and inference volume will grow by orders of magnitude imply rising demand for enterprise data/AI infrastructure like Box's.
MRVL · Capital · Positive Marvell lifted its revenue outlook by billions and struck a bullish note on its long-term total addressable market amid the AI buildout.
AMD · Demand · Positive AMD hit a record high amid the AI buildout, with Levie's comments on surging AI compute demand supporting the AI-chip demand narrative.
Analyst Dan Ives Initiates Nvidia and SpaceX With Outperform Ratings
Analyst Dan Ives initiated coverage of Nvidia and SpaceX with Outperform ratings at his new firm, Yorkville Ives, arguing the AI buildout remains in its early innings. Ives set a $300 price target on Nvidia, implying 26% upside, and a $225 target on SpaceX, implying about 34% upside. Nvidia, now valued at more than $5.6 trillion, reported fiscal Q2 2027 data center revenue up 117% year-over-year to $89 billion and total revenue up 106% to $96.2 billion, beating Wall Street's forecast by $4.06 billion, with adjusted earnings of $2.22 per share beating by $0.13. SpaceX, which went public in June in Wall Street's largest ever initial public offering, reported 2Q26 total revenues of $7.8 billion, up 92% year-over-year and $994 million above expectations, while its net loss narrowed to $541 million from $1 billion a year earlier. SpaceX's capex rose to $18.37 billion in 2Q26 from $2.83 billion in 2Q25, with $15.83 billion of that total listed as AI capex. Ives said AI infrastructure spending by the five largest cloud platforms is running on the order of $850 billion this year, describing the industry as being in the third inning of a nine-inning game.
IREN Reports $4 Billion Contracted AI Cloud Revenue as Shares Fall 7.3%
IREN Limited reported fiscal 2026 revenues of $707 million, up 41.1% year over year, as the Bitcoin miner pivots to an AI cloud platform. AI Cloud Services revenues surged to $128.8 million from $16.4 million, while Bitcoin mining contributed $578.2 million, and fourth-quarter AI cloud revenues more than doubled sequentially to $70.5 million. The company announced $4 billion in contracted annualized run-rate revenue for its 2026 capacity, with $1 billion operational as of late August, though ARR represents annualized contracted activity rather than recognized revenues. IREN delivered its first 50-megawatt Horizon deployment to Microsoft in August, with three additional phases scheduled for delivery during the fourth quarter of calendar 2026, and management targets approximately 300 megawatts of cumulative IT capacity in 2026, rising to 800 megawatts in 2027. The company reported a fourth-quarter fiscal 2026 net loss of $684 million, largely driven by noncash impairment charges from retiring Bitcoin mining equipment, and management expects fiscal 2027 capital expenditures of $25-$30 billion. IREN shares have declined approximately 7.3% over the past three months, and the stock carries a Zacks Rank #3 (Hold).
IREN · Capital · Negative IREN posted a $684 million Q4 net loss from noncash impairments and guided to $25-$30 billion in fiscal 2027 capex.
IREN · Demand · Positive IREN reported $4 billion in contracted annualized AI cloud revenue and delivered its first 50MW Horizon deployment to Microsoft.
MSFT · Demand · Positive Microsoft is the customer receiving IREN's first 50MW Horizon deployment, with three additional phases scheduled.
Michael Burry Says Big Tech AI Spending Bets on 'Too Big to Fail' Oligopoly
Investor Michael Burry said Tuesday that big technology companies are spending wildly on data centers because their CEOs assume they will be granted an oligopoly that is too big to fail. In a post on X, Burry argued that this expectation is the only reason for the spending spree. JPMorgan estimates AI capital spending could rise from roughly $700 billion this year to $1 trillion next year, while Goldman Sachs expects the five biggest hyperscalers to issue about $250 billion in bonds this year and $400 billion in 2027. Separately, Space Exploration Technologies Corp. is reportedly seeking $40 billion in loans and debt, led by Apollo Global Management, to buy Nvidia Corp. chips, according to the Financial Times. Amazon.com Inc., Alphabet Inc., and Microsoft Corp. are on track to spend roughly $590 billion to $600 billion on capital expenditures this year, much of it on data centers and AI infrastructure, while Meta Platforms Inc. reportedly described some of its AI data centers to the IRS as pilot models to claim federal research tax credits, cutting its tax bill by $3.9 billion in 2025, according to The New York Times.
SPCX · Capital · Neutral SpaceX is reportedly seeking $40B in loans and debt led by Apollo to fund Nvidia chip purchases, a financing event with unclear net effect.
AMZN · Capital · Neutral Named among hyperscalers on track to spend ~$590-600B on capex this year, much on data centers/AI infrastructure.
GOOG · Capital · Neutral Named among hyperscalers on track to spend ~$590-600B on capex this year, much on data centers/AI infrastructure.
META · Capital · Neutral Reportedly described some AI data centers to the IRS as pilot models to claim federal research tax credits, cutting its 2025 tax bill by $3.9B.
MSFT · Capital · Neutral Named among hyperscalers on track to spend ~$590-600B on capex this year, much on data centers/AI infrastructure.
NVDA · Demand · Positive SpaceX reportedly seeking $40B in loans/debt to buy Nvidia chips, a concrete end-customer order for Nvidia's AI chips.