Behind the AI battle that NVIDIA, OpenAI, and the cloud giants are fighting over, there's a quiet group collecting rent from everyone — companies that buy the land, pull in high-voltage power lines, pour the concrete, and then lease "space + power + cooling" to the tech giants. Many of them are listed as REITs (real-estate investment trusts). This lesson takes you to the most tangible layer of AI — and why, in this era, the thing you rent out is no longer square meters but "megawatts," and why tenants book years ahead, before the building is even finished.
Contains
Theme index· base 100 · USD total return
No index history for this theme yet.
News & notes movingColocation & Hyperscale REITs
Colocation & Hyperscale REITs▲3impact 4
Nvidia, Google and Emerald AI Launch AI Energy Management Alliance
Nvidia, Google and Emerald AI have launched the AI Energy Management Alliance, or AEMA, a coalition that dynamically manages the electricity use of data centers in response to grid conditions. Emerald AI founder and CEO Varun Sivaram said the alliance's founding members are joined by a cohort of 20 launch partners, including the AI lab Anthropic, the semiconductor firm Analog Devices, and the energy companies AES, NRG, Constellation, RWE and National Grid. Sivaram said Emerald AI, which was founded under two years ago, is building with Nvidia and Digital Realty the world's first from-the-ground-up power-flexible AI data center, a 100 megawatt facility in Manassas, Virginia, that comes online later this year. He said Google, one of the founding members, has already done a gigawatt of demand response for its data centers, while Emerald and Nvidia have completed six demonstrations around the world, in London, Phoenix and Virginia. Sivaram said the alliance is talking to the FERC commissioners, state regulators and the administration about a grand bargain in which flexible AI data centers act as good citizens to grids and communities in return for faster and larger connections to the power grid.
CleanSpark Data Center Bond Draws $10 Billion in Orders for $2.28 Billion Sale
CleanSpark Inc.'s debut junk-bond offering for a data center tied to Meta Platforms Inc. drew orders of about $10 billion, more than four times the deal's size, according to people familiar with the matter. The five-year notes were set at almost $2.28 billion and launched at 98.5 cents on the dollar to yield 8.25%, roughly 1.75 percentage points above the average for BB rated firms tracked by Bloomberg. The fundraising, led by Morgan Stanley, will help finance construction of a data center in Sandersville, Georgia, that has been fully leased to Anviran LLC, a Meta subsidiary, under a $6.6 billion, 20-year contract. The facility is expected to begin operations in the fourth quarter of 2027, with Meta guaranteeing rent and operating expenses. Data center developers have now sold more than $3 billion of high-yield bonds so far this year, most backed by long-term leases with hyperscalers including Oracle Corp. and Amazon.com Inc.
SoftBank's SB Energy IPO Faces Investor Test as OpenAI Timeline Clouds $50 Billion Listing
SoftBank's push to list SB Energy at a valuation of roughly $50 billion is facing a tougher investor test as uncertainty around OpenAI's IPO timeline raises questions about how much of the data-center developer's future growth depends on a single customer. SB Energy is expected to seek between $5 billion and $7 billion in an IPO within the coming weeks. The company has yet to bring a single data center online, but already carries roughly $439 billion of contracted backlog tied largely to 8.8 gigawatts of future capacity, with roughly $357 billion of that contracted revenue not expected to be recognized until 2034 or later. SB Energy generated only $138.7 million of revenue during the first half of this year, primarily from legacy solar operations, and posted a $551.6 million operating loss. OpenAI is the primary tenant for SB Energy's largest pipeline projects, while Nvidia has provided a $105 billion guarantee for the initial phase of an 8-gigawatt Ohio data-center campus, and both OpenAI and Nvidia are expected to own stakes in the publicly listed company. SB Energy estimates it needs more than $170 billion in capital spending to build its pipeline, and analysts cited in the report estimate it may need another $7 billion of equity beyond IPO proceeds, in addition to substantial debt, just to maintain its typical 10% equity contribution to projects.
Meta Raises 2026 Capex Guidance to $130 Billion to $145 Billion
Meta Platforms raised its full-year 2026 capital expenditure guidance to a range of $130 billion to $145 billion, including principal payments on finance leases, narrowed from a prior $125 billion to $145 billion range in its Q2 2026 report on July 29, 2026. The forward guidance nearly doubles Meta's full-year 2025 capex of $72.215 billion, and Q2 capital expenditures alone reached $31.1 billion, driven by servers, data centers, and network infrastructure. To fund the build, Meta ended Q2 with $90.3 billion in cash and marketable securities and $83.7 billion in debt, and announced a strategic venture with BlackRock to develop a one gigawatt data center in El Paso, Texas. CFO Susan Li said Meta is demand constrained today, and CEO Mark Zuckerberg said the company is receiving quite a number of offers at a meaningful premium over what we paid for the compute, framing direct compute sales as one leg of a portfolio that also includes APIs, business agents, productivity tools, and subscriptions. The strain is visible in the quarterly numbers: Q2 free cash flow was $784 million, down 91.31% year over year, and operating margin compressed to 31% from 43%, even as Q2 revenue reached $60.801 billion, up 27.96% year over year and above the $60.286 billion consensus, with advertising revenue of $59.4 billion, up 27%.
Amazon AWS Revenue Hits $42.23B, Up 37% in Fastest Growth in 18 Quarters
Amazon Web Services posted $42.23 billion in revenue, growing 37% year over year, the segment's fastest growth in 18 quarters, with a 39.4% operating margin and a $496 billion contracted backlog. Amazon's chips and AI businesses each eclipsed run rates of more than $25 billion in the second quarter, both growing at triple-digit percentages year over year, while 98% of Amazon's top 1,000 EC2 customers use Graviton and Anthropic and OpenAI have made multi-year, multi-gigawatt commitments to Trainium. Q2 operating income landed at $27.46 billion, up 43.2% year over year, and advertising is a $70 billion-plus trailing-twelve-month business growing 26%. Capex reached $54.21 billion in a single quarter, up 68.4% year over year, and free cash flow swung to negative $7.6 billion on a trailing-twelve-month basis, though most AI capacity is contracted for at least five-year terms. Since Amazon reported Q2 on July 30, 2026, AMZN moved from $230.08 to $251.19, while SPY went from $747.03 to $762.70 and QQQ went from $687.99 to $716.92. Andy Jassy said AWS could become a few hundred billion dollar revenue business and now believes it will be at least double that and very possibly a trillion dollar annual revenue business in time.
CleanSpark Signs $6.6 Billion AI Data Center Lease in Georgia
CleanSpark has signed a 20-year, $6.6 billion triple-net lease for its Sandersville, Georgia, site, a deal the company expects to generate $330 million in annual revenue. Chairman and CEO Matt Schultz said the site includes a fully energized 250-megawatt substation, that CleanSpark secured an additional 122 acres with community support, and that the tenant received short-term exclusivity for up to 885 additional megawatts in Texas, with the first data hall targeted for delivery in December 2027. The announcement came during an H.C. Wainwright panel where mining and AI infrastructure executives described a broad pivot from bitcoin operations toward AI colocation, cloud services and GPU-as-a-service. Core Scientific's Russell Cann said direct-liquid-cooled construction costs have climbed from roughly $4.5 million per megawatt for early projects to about $10 million to $10.5 million per megawatt for facilities turned on this year and $12 million to $13 million per megawatt for projects expected to start operating in 2027. Cann also said Core Scientific recently announced a roughly 500-megawatt agreement with AMD, with AMD holding rights to the next 2,000 megawatts across sites in Hunt County and Pecos, Texas, and Muskogee, Oklahoma, under a 15-year triple-net lease at $125 per megawatt or a modified gross structure at $145 per megawatt. Executives pushed back on AI bubble concerns, with Cann noting roughly 5 gigawatts of rack space is delivered annually against approximately 15 gigawatts of annual chip demand, while cautioning that many power requests may be duplicative or unsupported by actual transmission, substations or fuel supply.
PLANET Says Data Center in EEC Is a Hot Commodity as Thai and Foreign Investors Line Up, Plans to Expand Total Power Capacity to 6.8 MW
Planet Communications Asia, or PLANET, disclosed that its data center in the Eastern Economic Corridor, or EEC, is drawing strong interest from both domestic and foreign investors, with operators continuously approaching the company to negotiate service use and joint investment. The momentum is driven by rapidly expanding demand for data, cloud, and AI technology in the region, along with government support through changes to laws, regulations, and investment incentive measures in the EEC. PLANET currently provides co-location data center services under the name STP Planet DC in Ban Chang district, Rayong province, supporting a maximum IT load of 600 kW per data hall, and plans to accommodate an expansion of total gross load of up to 6.8 MW to meet demand and support the future growth of enterprise customers and large cloud service providers.
San Jose residents push back on AI data center construction, demand transparency and environmental impact review
In San Jose, the largest city in California's Silicon Valley, residents and environmental groups are mounting an opposition campaign against plans to build artificial intelligence data centers. San Jose, with a population of about 1 million, is California's third-largest city. Last year it reached an agreement with utility PG&E to strengthen power supply for data centers and other facilities, and it estimates that each data center opened generates 3 million to 6 million dollars a year in tax revenue for services such as police, fire, libraries, roads and parks. Elina Yin, who leads the residents' group I Love San Jose, says companies should prove how they give back to the community, and is calling for stricter public review before construction is approved, greater transparency on energy and water use, and studies of health and environmental impacts. Mayor Matt Mahan told Reuters that the solution is not to stop construction but to carry it out responsibly, and that residents' questions are entirely reasonable. The California State Assembly and Senate recently passed several data center-related bills, including measures to ensure large electricity consumers share the costs of the transmission grid and requirements to increase transparency around energy and water consumption, and Governor Newsom must decide this month whether to sign or veto them.
Crusoe said Thursday it raised $3.9 billion in a Series F round that pushes its valuation to $30.9 billion. The round was co-led by Atreides Management, Mubadala Capital, and Valor Equity Partners, with Founders Fund, GIC, Nvidia, Qatar Investment Authority, Radical Ventures, and TPG also participating. Crusoe also announced three new board members: Cloudflare CFO Thomas Seifert, Bill Stein, partner and CIO at Primary Digital Infrastructure, and Redwood Materials founder and CEO JB Straubel, who also sits on Tesla's board. The fresh capital will help finance existing data center projects, including a large site in Abilene, Texas, used by OpenAI, as well as smaller, modular AI factories called Spark that can be transported by truck and connected to large power sources almost anywhere. The raise comes 10 months after Crusoe raised $1.38 billion at a $10 billion valuation last October, and follows a reported $13 billion, five-year cloud contract to supply quantitative trading firm Jane Street with GPUs and AI infrastructure.
AXG and Digital Realty Partner on Hosted Private Infrastructure as a Service Across More Than 30 Countries
AXG and Digital Realty announced a partnership to help global system integrators and managed service providers deliver fully hosted, lifecycle-managed private infrastructure as a service across more than 30 countries. The joint offering combines Digital Realty's global data center platform, PlatformDigital, and its ServiceFabric interconnection and orchestration solution with AXG's multi-technology infrastructure-as-a-service model, covering data center space, power, cooling, connectivity, interconnection, and sovereign-ready colocation. Under the arrangement, GSIs and MSPs retain ownership of the customer relationship, solution design, managed services, orchestration, and the application layer, while AXG and Digital Realty provide the underlying hosting, commercial, and operational infrastructure. The offering targets four areas of enterprise and government demand: private AI, edge as a service, cloud repatriation, and neo-cloud capacity, and is available today across 31 markets spanning North America, EMEA, APAC, and LATAM. Digital Realty's Jules Johnston, Senior Vice President of Channels and Alliances, said the combination creates a powerful new route to market, while AXG CEO Bo White said the partnership closes the gap between dedicated infrastructure economics and cloud-style simplicity.
Hyperscale Data Invests Over $70M in Michigan AI Data Center
Hyperscale Data said it has invested more than $70M in Alliance Cloud Services and its Michigan AI data center, with operations under a previously announced master services agreement expected to begin in November 2026. The investment has funded a substantial portion of the capital needed to prepare the facility for an initial 20 MW AI compute deployment to a California-based neocloud provider, and the company said it has acquired a substantial amount of the equipment needed to bring the contracted capacity online. The agreement covers an initial 20 megawatts of critical AI compute capacity for 10 years, with two five-year extension options, and if maintained for the full 20-year term is expected to generate more than $1.2B in revenue. The customer can expand the deployment to 52 MW, which could generate more than $3B in revenue over 20 years, while the Michigan facility has about 340 MW of potential capacity, leaving about 270 MW for future customers after a full 52 MW deployment. The stock price scaled about 4.4% on Thursday during pre-market trade.
US House passes Ratepayer Protection Act in 417-3 vote
The US House of Representatives overwhelmingly passed the Ratepayer Protection Act by a vote of 417 to 3, aiming to shield the public from electricity bill increases that could stem from the expansion of data centers driven by the artificial intelligence race. The bill now heads to the Senate for further consideration. Republican Representative Gabe Evans of Colorado introduced the bill. Its core provision requires technology companies to bear the cost of new energy infrastructure built to serve data centers, rather than shifting that burden onto consumers. It encourages state utility regulators to set standards requiring high-volume electricity customers, including large data centers, to cover the costs of adding new generating capacity. US House Speaker Mike Johnson said the Ratepayer Protection Act will help ensure American households do not have to shoulder the costs of expanding AI infrastructure or face higher energy costs. The International Energy Agency estimated in 2025 that electricity consumption by data centers in the United States could more than double its 2024 level by the end of this decade, while other assessments indicate that data center electricity use could account for more than 10% of US power demand by 2030.
Oracle Tops Q1 Estimates as Cloud Revenue Jumps 62% to $11.6 Billion
Oracle beat Wall Street's revenue and earnings expectations for the first quarter of fiscal 2027, with total quarterly revenues rising 30% year over year to $19.3 billion and total cloud revenues climbing 62% to $11.6 billion. Within that cloud total, cloud infrastructure revenues more than doubled, up 121%, while cloud applications revenues grew 10%. Remaining performance obligations jumped by $209 billion year over year to $664 billion, and operating cash flow reached a record $23.1 billion, up 184%, even as free cash flow stayed negative amid a data center buildout that added 850 megawatts of new capacity in the quarter. Management guided full-year fiscal 2027 capital expenditures toward $90 billion to $95 billion, and the Zacks Consensus Estimate for fiscal 2027 earnings stands at $8.10, implying 6.16% year-over-year growth. Oracle shares have lost 27.9% year to date, underperforming the Zacks Computer and Technology sector's 17.4% gain, and the stock trades at a forward 12-month P/E of 15.8X versus the Zacks Computer - Software industry average of 23.16X; ORCL currently carries a Zacks Rank #3 (Hold).
Cipher Mining Jumps 9% on Conditional ERCOT Status for 3.2 GW of Texas Capacity
Cipher Mining shares rose 9% to $16.42 in early Wednesday trading after the Electric Reliability Council of Texas disclosed conditional grid capacity covering 3.2 gigawatts across the company's Texas projects through the state grid operator's Batch Zero process, rather than through a company release. Peer names moved in sympathy on the same ERCOT read, with Core Scientific up 4% to $16.76 and IREN up 2% to $42.57, after all three sold off a day earlier as the artificial intelligence buildout trade came apart. The wider sector barely participated, with the Global X Data Center & Digital Infrastructure ETF up 1% to $27.47 and the SPDR S&P 500 ETF Trust up 0.35% to $760, isolating Cipher's surge as a grid-access story rather than a broad market move. ERCOT created the Batch Zero process earlier this year to work through a queue of large-load interconnection requests dominated by data centers, and a base load classification covers power expected to be consumed continuously; the classifications are conditional and remain subject to verification and further audits, marking a position in a process rather than a completed grid connection. CoinShares, in its most recent quarterly Bitcoin mining report, estimated that artificial intelligence workloads generate annualized profit of roughly $1.5 million per megawatt against $0.5 million per megawatt for Bitcoin mining, and said Cipher expected mining to become immaterial to its business and was winding that operation down.
Blockfusion Signs 15-Year Anchor Lease With CoreWeave at Niagara Falls AI Campus
Blockfusion USA, Inc. announced that its wholly owned subsidiary, North East Data, LLC, has entered into a definitive anchor lease with CoreWeave, Inc. for capacity at Blockfusion's Niagara Falls, New York campus, alongside a companion expansion agreement. The two agreements convert the non-binding letter of intent announced on June 30, 2026, into definitive agreements, and the lease carries a 15-year initial term with two five-year renewal options. Blockfusion has transformed a decommissioned power plant into a low-carbon data center powered predominantly by hydroelectric power and is now adapting the campus for high-density, liquid-cooled AI infrastructure, benefiting from an existing grid interconnection, clean energy and redundant fiber infrastructure. Chief Executive Officer Alex Martini-Lo Manto called the agreement a defining milestone for Blockfusion and strong validation of the Niagara Falls campus, while Mayor Robert Restaino said the commitment creates opportunities for residents and businesses. The agreements also mark a step in Blockfusion's proposed business combination with Blue Acquisition Corp., which is expected to result in Blockfusion Digital Infrastructure, Inc. becoming a publicly traded company listed on the Nasdaq Stock Market under the ticker symbol BDI, with additional information to be included in an amendment to the registration statement on Form S-4 filed with the U.S. Securities and Exchange Commission.
Fed Expected to Hike Rates as AWS Loses Gulf Data Access
The Federal Open Market Committee is widely expected to raise its benchmark rate to 3.75%-4.00% from 3.50%-3.75% at its September meeting, with a strong August jobs report and hotter-than-expected core CPI supporting the move. Amazon Web Services said it cannot restore access to resources and data hosted exclusively in its Bahrain region after Iranian drone strikes damaged infrastructure, affecting multiple Availability Zones and exceeding what its regional and multi-AZ services were designed to withstand, while one data-hosting zone in the United Arab Emirates also remains affected. SoftBank Group's five-year credit default swaps hovered near a three-year high at around 383.2 basis points, near their highest level since 2023, as investors assessed its funding plans for OpenAI, to which SoftBank has committed $64.6B. OpenAI has held early discussions with large investors about a new capital raise that could value the company at about $1.2T before an IPO, talks initiated by investors that could fund acquisitions and give the company flexibility to delay its IPO by one or two quarters. The Trump administration has appealed a federal court ruling by Judge Beryl Howell blocking the EPA from sending California's vehicle-emissions rules to Congress for potential repeal, a dispute involving stricter standards for cars, trucks, and lawn and garden equipment.
Meta Taps Dina Powell McCormick to Lead AI Infrastructure Push
Meta Platforms Inc. has elevated Dina Powell McCormick, a former Trump administration official and Goldman Sachs Group Inc. executive, to president, vice chairman and co-leader of Meta Compute, the company's effort to build and monetize the infrastructure fueling its artificial intelligence ambitions. Powell McCormick resigned from Meta's board about a week after Chief Executive Officer Mark Zuckerberg asked her to take the internal role during a December board retreat at his Kauai compound, and she assumed the position in January. Meta is expecting to spend more than $600 billion on its AI buildout before the end of the decade, not including the many billions more it is leveraging from Wall Street. In one benchmark deal, Meta and BlackRock Inc. agreed to build a $14 billion data center in El Paso, Texas, with BlackRock raising more than $12 billion in debt to fund the facility and holding an 80% interest in the joint venture while Meta retains 20%. Meta used a similar joint-venture structure with Blue Owl Capital Inc. to help finance its Louisiana data center, dubbed Hyperion, and NextEra Energy Inc. announced Wednesday it will hire 1,000 people out of Meta's America's Workforce Academy, an initiative Powell McCormick stood up, with Meta investing $115 million in the academy in its first year. Powell McCormick's remit extends to countering public backlash to AI and data centers, weighing in on legal and reputational fights including Meta's settlement of up to $18 billion with state attorneys general over children's safety on Instagram and Facebook, and helping lead a $1 billion fund to invest in communities where Meta builds.
Anthropic Signs Australia's First Data Center Deal, 2.16 Gigawatts Near Brisbane
U.S. artificial intelligence company Anthropic has signed a lease for its first data center in Australia. According to two people familiar with the matter, the facility covered by the agreement has a planned capacity of 2.16 gigawatts, a scale comparable to the largest facilities currently operating anywhere in the world. The site is farmland about 250 kilometers from Brisbane, with operations scheduled to begin in 2027. The facility is being planned by Singapore-based developer Zera DC, which will procure power through renewable energy purchase agreements and will also bear the cost of connecting to the transmission grid, so existing customers will not face any cost burden. The agreement is conditional on approval from the Foreign Investment Review Board.
CoinShares: Miners Won't Return from AI Even as BTC Recovers
On September 15, CoinShares published its second-quarter 2026 Bitcoin mining report, expressing the view that even if the Bitcoin price recovers, the shift of listed miners toward AI infrastructure will not reverse. According to the report, in the second quarter the listed mining sector as a whole fell below its cash-based break-even point, with the average cash production cost per BTC at about 75,500 dollars, while BTC had fallen to 58,400 dollars by the end of the quarter. By the firm's estimates, AI brings these companies about 1.5 million dollars in annual profit per megawatt, while mining yields only about 500,000 dollars. Author Luke Nolan cited examples including Core Scientific paying 41.9 million dollars to cancel contracts for next-generation mining rigs. New data center construction itself has also become more difficult, with at least 225 construction halts or regulatory interventions seen across 30 U.S. states. Although BTC has recovered to around 77,000 dollars and many miners are once again above break-even, the firm believes it is unlikely that miners who have committed power and infrastructure to AI through long-term leases and the like will return to mining.
Bernstein Flags CoreWeave as Biggest Loser in Potential AI Slowdown
Bernstein named CoreWeave as the single stock most exposed to a potential AI slowdown, sending its shares down 6.75% to close at $82.98 on Sept. 14. The warning followed an essay by Anthropic CEO Dario Amodei titled "We Must Pace the Frontier," in which he argued AI companies need to slow how they improve model capabilities, a call backed within hours by OpenAI CEO Sam Altman and Elon Musk. Bernstein analyst Madison Rezaei said about 25% of CoreWeave's active U.S. power sits in Tier 3 and Tier 4 rural markets built for training, with another 74% of its contracted but not yet active power in those same markets, leaving it vulnerable if training demand slows. Bernstein rates CoreWeave Underperform with a $74 price target, while the stock's average Wall Street price target sits near $139. By contrast, Bernstein rates Equinix, Digital Realty and Csquare Outperform, with price targets of $1,270, $226 and $27 respectively, noting each keeps more than 90% of its U.S. capacity in metro markets. The sell-off spread across AI infrastructure, with Nebius down 8%, Applied Digital down 6%, DigitalOcean down 5%, IREN down 4%, Lumentum down 8%, Ciena and Coherent each down 7%, Arista Networks down 5%, Marvell and Arm each down 8%, AMD and Intel down 6%, and Broadcom down 4%.
Anthropic Signs $13.7B Compute Deal With RUM Group
Anthropic signed a six-year, $13.7B compute agreement with Trump-linked RUM Group, the company best known for its Rumble video-sharing platform. Under the deal, Anthropic will lease AI compute capacity from RUM's Maysville, Georgia data center, which is under construction with 120 MW of secured grid power and potential expansion to 180 MW, and Anthropic gains warrants to buy roughly 51 million shares of RUM. The contract value is more than 300 times RUM's latest quarterly revenue of $40.4M. RUM shares soared more than 15% on Monday, and the session was set to be RUM's highest volume day in history. The agreement followed a viral social media post by former OpenAI and Anthropic researcher Jacob Coxon warning about AI safety, which drew 170 million impressions in five days and helped lift cybersecurity stocks including Palo Alto Networks, Fortinet, Okta and CrowdStrike.
Lagarde Warns Europe Must Build Its Own AI or Face Foreign Leverage
European Central Bank President Christine Lagarde warned in Vienna on Monday that Europe cannot rely solely on imported artificial intelligence from the United States and must develop its own capacity or risk giving foreign technology suppliers leverage over every corner of its economy. Lagarde said Europe's first priority must be expanding its own computing infrastructure, noting the continent already lacks sufficient data center capacity to meet existing demand and that on current trends the gap is projected to grow more than sixfold within a decade, with closing it potentially costing as much as €600 billion over ten years, chips included. She said the E.U.'s new gigafactories are a start but fill only a fraction of the shortfall, and that Europe must also develop AI models competent enough for the majority of use cases and hosted on European systems, pointing to French firm Mistral and a company in San Sebastián that adapted a Chinese open model to top independent benchmarks among European models. Lagarde said euro area households hold roughly €440 billion in U.S. technology firms while U.S. hyperscalers issued more than $100 billion in bonds last year, and cited ECB estimates that rapid AI adoption could raise productivity by as much as 4% over a decade. She argued the buildout requires patient equity rather than debt alone, noting European households save roughly €1.4 trillion a year but that markets to channel that capital into homegrown technology do not yet exist at the necessary scale. The remarks came alongside prior Brussels steps, including European Commission legislation to restrict sensitive government cloud contracts to providers meeting European sovereignty standards through the Cloud and AI Development Act, and a roughly €30 billion E.U. tender for up to seven AI gigafactories pairing up to €10 billion in public funding with at least €20 billion in expected private investment, with construction targeted to begin in 2027.
UAE Reviews 5-Gigawatt AI Data Center Plan as War Threats Escalate
The United Arab Emirates is reviewing its strategy for a 5-gigawatt AI data center project, an infrastructure undertaking valued at 1 billion dollars, after conflict with Iran spilled over and undermined security stability in the Persian Gulf region. The project, designed to be centralized on a 26-square-kilometer site in Abu Dhabi and intended as one of the largest technology infrastructure projects outside the United States, must urgently shift from centralization to risk diversification. The UAE government and major capital groups such as G42 are driving this strategic adjustment. Key triggers include attacks on Amazon Web Services cloud and data center infrastructure earlier this year, as well as Iran's military revealing coordinates and warning that the Stargate UAE project could become a strategic target. New protective measures under consideration include distributing data centers across multiple locations nationwide, embedding some structures underground, using natural mountain interiors, installing air defense systems and electronic jamming systems, and employing special reinforced-concrete construction materials resistant to explosive force. Meanwhile, the first phase of the project, Stargate UAE, valued at 30 billion dollars, continues under cooperation with partners from the United States, in order to preserve an advantage in accessing Nvidia's advanced AI chip technology in exchange for reducing technology ties with other major powers under global strategic conditions.
Big Tech's $220 Billion AI Bond Spree Is Warping Credit Markets, Alphabet and Meta Show
Alphabet, Amazon, Meta, Microsoft and Oracle issued roughly $220 billion of bonds over the prior year to fund data-center expansion, a wave of supply large enough to distort the credit market around it, Reuters reported on September 10. Similar credits are no longer always trading at similar spreads as repeated giant deals flood the market, and even high-quality borrowers can see spreads widen when investors must absorb too much similar paper. Alphabet has the cash generation, cloud growth and advertising franchise to access investment-grade markets on favorable terms, while Meta pairs the same balance-sheet advantage with a highly profitable advertising engine funding its buildout. Institutional positioning moved in opposite directions: Insider Monkey's database showed 275 hedge funds holding Alphabet in the second quarter, up from 265 in the first, with Berkshire Hathaway increasing its Class A position about 45%, while Meta fell to 254 holders from 262 and Newlands Management trimmed its stake to 9,664,414 shares. Alphabet Class A had 77,698,668 shares sold short on August 31, just 0.72% of float, with 3.48 days to cover. The credit distortion is not a solvency warning for Alphabet or Meta but a price signal: when even elite borrowers repeatedly tap bond markets for AI infrastructure, investors demand compensation for duration and supply.
GULF plans 5-year investment of 140 billion baht to push data centre capacity to 2,000 megawatts
Gulf Development Public Company Limited, or GULF, has announced a major investment plan over the next five years worth as much as 140 billion baht to expand its data centre and digital infrastructure businesses and capture the growth of AI technology. The company aims to raise its data centre capacity to as much as 2,000 megawatts, up from roughly 200 megawatts at present. Chief Executive Officer Sarath Ratanavadi said the private sector is ready to take part in developing power projects of all types under the new Power Development Plan, or PDP, including solar, wind, battery energy storage and natural gas-fired power plants, as well as studying small modular reactor technology. The PDP targets power generation capacity of 50,000 megawatts, and the company sees opportunities in the free electricity business, such as selling green power to large electricity users. On the data centre side, GULF agrees with the government's move to set clear definitions and regulations, noting that demand in Thailand is very high from users of various social media platforms. The company has also joined hands with Singtel Group to take part in developing the Vietnam-Thailand-Singapore Cable System, a submarine cable network linking Vietnam, Thailand and Singapore on an open access basis, with a project value in the tens of billions of baht and service scheduled to begin in 2030.
Texas Audits 474GW Data Center Queue as Meta and Digital Realty Back Abbott's Standards
Texas Governor Greg Abbott ordered an audit on August 3 of projects moving through ERCOT's interconnection process, a queue containing more than 474 gigawatts of requests, over five times the state's record peak demand, and ERCOT began issuing verification requests on September 9. Meta said on August 10 that it would pay the full costs associated with its data centers' energy and water use rather than shift them to consumers, while Digital Realty welcomed the Texas framework on August 7. For Meta, a company generating tens of billions of dollars of cash can absorb infrastructure requirements that may make marginal projects harder to finance, though more disclosure, grid work, water obligations and project verification can push capacity further into the future and raise the all-in cost of each megawatt. Digital Realty benefits if its operating history, utility relationships and existing powered sites become harder to replicate, since scarcity can support rents and the value of developed capacity, but projects stuck behind transmission upgrades or compliance reviews can delay leasing and weaken development returns. Insider Monkey's database showed 254 hedge funds holding Meta in Q2 2026, down from 262 in Q1, while 73 funds held Digital Realty, up from 46, with D. E. Shaw increasing its position 43% to 1,542,316 shares; short interest in Digital Realty stood at 8.3 million shares on August 31, equal to 2.27% of float, with 3.97 days to cover.
Buildots Raises $130M to Expand AI Platform Across $16T Construction Industry
Buildots announced a $130 million funding round led by O.G. Venture Partners, bringing its total capital raised to $297 million. The round included participation from Lightspeed Venture Partners, Intel Capital, Mohari Ventures, Human Capital, Qumra Capital, Avigdor Willenz, Viola Growth and Poalim Equity. The Tel Aviv-based company said the funding extends a multi-year streak of 3x annual revenue growth and will scale deployments of its AI construction platform across the global data center buildout. Buildots is currently deployed by more than 100 large-scale firms, including owners Digital Realty and Intel, hyperscalers, and contractors STO Building Group, JE Dunn, Mortenson, Bouygues and HOCHTIEF. The company said seven-figure, portfolio-wide, multi-year agreements are now the norm rather than the exception, and the round adds two new investors to its cap table: Human Capital and Mohari Ventures.
Hyperscale Data Sets $750M Minimum Sale Threshold for Michigan AI Data Center
Hyperscale Data has set $750M as the minimum valuation at which it would seriously consider selling its Michigan AI data center campus. Management estimates the Michigan Campus could be worth between $750M and $1.25B, based on its master services agreement, existing infrastructure and power access, expansion potential, and valuations of comparable AI infrastructure companies. The company is also weighing other options to maximize shareholder value, including a potential sale of the Michigan Campus, an IPO of wholly owned subsidiary Sentinum, selling a minority stake in Sentinum to fund expansion, or continuing to own and develop the campus. No definitive decision has been made on any transaction.
KKPS says AI drives FDI into Thailand, picks DELTA, GULF and WHA to play the data center theme
Supapong Iamkong-aek, an analyst at Kiatnakin Phatra Securities, or KKPS, said investment in artificial intelligence is entering a new cycle, with the focus shifting from developing the technology to investing in the infrastructure needed to support AI's expansion, including data centers, power systems, electronic components and connectivity. This is an opportunity for Thailand to attract a new wave of foreign direct investment. Although only a limited number of technology and electronics companies are listed on the Thai stock exchange, off-market businesses tied to the downstream side of AI, such as Analog Devices and Seagate, have production bases in Thailand and have seen a significant rise in investment promotion applications to the Board of Investment, or BOI. On data center investment in Thailand, the value of BOI promotion applications in the first quarter exceeded 700 billion baht, but slowed markedly in the second quarter as some foreign investors waited for clarity on data center regulations in Thailand. Thailand currently has fewer than 1 GW of data centers actually in operation, while Malaysia has about 7 GW. Given the steady flow of BOI applications, Thailand's data center capacity could rise to 3 to 5 GW within the next three to five years, and combined with the new Power Development Plan, Thailand is preparing as much as about 9 GW of power capacity to support data centers. On electronics stocks, the analyst said Delta Electronics (Thailand), or DELTA, remains the standout, given its large market capitalization and its direct link to the AI theme, with a price-to-earnings ratio as high as about 100 times, compared with around 20 to 30 times for Nvidia. Among power plant operators, Gulf Development, or GULF, is the top pick, while in the industrial estate group, WHA Corporation, or WHA, is the top pick because it has a larger share of revenue and land area tied to data centers than AMATA.
Data Center IPO Wave Accelerates as Nearly a Dozen Firms Seek Public Markets
Nearly a dozen data center companies have gone public, filed for an IPO or been reported to be exploring one over the past year, with the pace accelerating over the past three months. Recent listings include Blackstone Digital Infrastructure Trust, Brookfield-backed colocation provider Csquare and Texas startup Fermi, while Switch and Singapore-based DayOne have confidentially filed, SoftBank-backed SB Energy has publicly filed and AI data center specialist Nscale is reportedly preparing a listing. Major digital infrastructure players Vantage Data Centers, CyrusOne, Blue Owl, DataBank and EdgeCore are all reportedly at least exploring IPOs, a reversal from 2021 and 2022 when take-private deals including Blackstone's $10B acquisition of QTS left just two listed data center firms. Industry leaders say no single force is driving the shift, attributing it instead to the trillions of dollars required for AI infrastructure and a growing diversity of public equity strategies. Blue Owl is reportedly planning a REIT seeded with around $6.5B of its own stabilized data center assets, while SB Energy is reportedly seeking a valuation of more than $50B despite still developing its first campus.
Anthropic Locks In $517 Billion Compute Commitment Ahead of IPO
Anthropic has lined up a $517 billion compute commitment ahead of its confidential IPO, according to a September 6, 2026 report from The Information, up from the roughly $180 billion in server-leasing spend the company previously told investors it expected through 2029. The figure aggregates contracted and reported deals stretching mostly across the next decade rather than a single contract, and actual outlays will hinge on how much capacity is delivered and utilized. Amazon and Alphabet together account for roughly 11 GW and more than $300 billion over about 10 years, Microsoft committed at least $30 billion for roughly 1 GW of Azure capacity, and the SpaceX/Colossus deal runs up to about $45 billion, with neoclouds including CoreWeave, Lambda, Nscale, and Fluidstack also inside the total. AMD captured a 2-gigawatt MI450 Series deal in Helios racks, complementing rather than replacing Anthropic's NVIDIA footprint. NVIDIA reported Q2 FY2027 revenue of $96.22 billion, up 105.85% year over year, with Data Center at $89.02 billion, up 117%, and guided Q3 to $108.0 billion plus or minus 2%, while CFO Colette Kress said the company expects revenue growth of approximately 70% in fiscal 2028.
Nvidia Partners With Australian Cloud Providers on 2 GW AI Buildout by 2027
Nvidia announced on September 9 strategic partnerships with Australian Nvidia Cloud Partners and AI infrastructure providers, working toward an AI infrastructure buildout of up to 2 gigawatts by 2027. The initiative will expand land, power, and data center shell capacity designed to host multiple generations of Nvidia's DSX AI factory infrastructure, with Nvidia supplying its DSX platform, accelerated computing, networking, software, and ecosystem support. The Australian buildout is one part of Nvidia's broader global AI factory expansion, which also includes a 140-megawatts AI factory in Japan expected to use Rubin GPUs and Vera CPUs, plus new AI supercomputers across multiple European countries. Power constraints remain the key risk, as a 2 GW buildout requires enormous reliable electricity and could face delays in grid connections, transmission, land development, construction, or regulatory approvals, alongside utilization risk if Australian demand for AI compute grows slower than expected. According to Insider Monkey data, 285 hedge funds held Nvidia shares at the end of the second quarter versus 275 in the previous quarter, with Fisher Asset Management raising its position by 3% to approximately $18.19 billion and AQR Capital Management increasing its stake by 18% to approximately $7.48 billion, while short interest stood at roughly 1.23% of shares as of August 14.
Real estate giants ramp up data center development; Mitsui Fudosan to invest over 600 billion yen cumulatively
Major real estate companies, including comprehensive developers, are stepping up data center development. According to the Ministry of Internal Affairs and Communications, the domestic related market is expected to exceed 5 trillion yen in 2028, and further growth is anticipated on the back of the artificial intelligence boom. Daiwa House Industry plans to build a total of 14 data centers in Inzai City, Chiba Prefecture, breaking ground on the fifth in late September. It has also developed a modular small-scale data center in which some components are manufactured at a factory and assembled on site, and the first such unit was completed in April in Okuma Town, Fukushima Prefecture. Mitsui Fudosan plans cumulative investment of more than 600 billion yen in data center-related projects by fiscal 2035, with development underway in four areas: Sagamihara City, Hino City in Tokyo, the Hokusetsu region of Osaka Prefecture, and the Tama region of Tokyo. In the Hino City project, the building's height was gradually lowered from an initial 80 meters to 63.5 meters in consideration of sunlight. New entrants are also appearing one after another. Mitsubishi Estate set up a data center business office in April and has already invested in multiple projects, and a data center in Ishikari City, Hokkaido, in which Tokyu Land Corporation participated in development for the first time, began operating in August using 100 percent renewable energy.
Musk Says SpaceX Will Build Exclusively on Nvidia Chips
Elon Musk said SpaceX has decided to build exclusively on Nvidia hardware, citing the Vera Rubin architecture as the best AI computer, during the company's second-quarter earnings call. The call was Space Exploration Technologies' first earnings call as a public company. SpaceX is using clusters of Nvidia GPUs to design orbital data centers, integrate with Starlink operations and Starship mission planning, and its near-term buzz centers on Starmind, an optimized satellite payload built on Nvidia's Vera Rubin NVL72 system. SpaceX has turned its Nvidia procurements into a fast-growing AI infrastructure business, leasing large clusters of Nvidia compute rather than using the chips solely to train its home-grown generative model, Grok. Anthropic entered a cloud services agreement with SpaceX worth $1.25 billion a month through 2029 for access to compute capacity across SpaceX's supercomputers Colossus and Colossus II, which feature roughly 325,000 Nvidia GPUs, while Google Cloud committed $920 million a month for access to roughly 110,000 Nvidia GPUs, CPUs, memory, and related networking kits through the middle of 2029, and Reflection AI signed a multiyear deal worth up to $6.3 billion to access Nvidia GB300 chips at a reported $150 million per month. Nvidia's data center revenue reached $89 billion in the second quarter, up 117% year over year, compared with AMD's $6.7 billion in data center sales, an increase of 107% year over year.
Dell Books $6.1 Billion in AI Server GPU Orders in Fiscal Q2
Dell Technologies CFO David Kennedy told investors at Citi's TMT conference that the company booked $6.1 billion of AI server GPU orders in its fiscal second quarter, matching the amount booked across the prior three quarters, and $13.2 billion over the last 12 months. Kennedy said pipeline activity is multiples of backlog, spanning neocloud providers, sovereign customers and enterprises, with enterprise demand growing fastest by rate. Growth is broadening beyond AI GPUs: traditional server revenue rose 122% in the second quarter, with every geography, customer segment and vertical growing at triple-digit rates, while storage grew 26% and Dell guided for double-digit storage growth for the full year, expecting to add $2.5 billion in storage revenue and calling the coming quarter potentially a record for Dell storage. Kennedy said Dell expects the supply-demand imbalance to worsen next year, with DRAM and NAND the largest constrained components, prompting customers to discuss infrastructure requirements two, three and four years ahead and to seek multiyear supply-access commitments, though guaranteed supply does not mean guaranteed pricing. Dell's Infrastructure Solutions Group posted a 15% margin rate despite growth in lower-margin AI servers, and the company has guided to operating expenses equal to 8% of revenue, which Kennedy called its most efficient level in Dell's 42-year history, versus 20% six years ago. Dell has more than 6,500 enterprise AI customers, up more than 60% over the past six months, and its share repurchase authorization increased from $1.6 billion to $3.9 billion over 90 days.
Trump Calls AI Data Centers 'Oil of the Next 50 Years' as Google Doubles Capex to $45 Billion
President Trump called AI data centers "the oil of the next 50 years" during an appearance on Laura Ingraham's The Ingraham Angle, arguing the buildout is delivering wealth and investment to American communities, while singling out Google for building a data center in Finland and saying he "wasn't happy about that." The remarks land as Google CEO Sundar Pichai describes a company building the physical layer of that new economy at unprecedented scale, telling analysts on the July 22, 2026 call that Google is "supply constrained" and that "the demand still outpaces that investment." Google's Q2 revenue reached $119.80 billion, up 24.2% year over year, with Google Cloud accelerating to $24.77 billion, up 82%, a Cloud backlog of $514 billion, and EPS of $9.11 against a $3.0427 estimate. Q2 capital expenditures hit $44.92 billion, up 100.14% year over year, driving free cash flow to negative $5.855 billion, while long-term debt jumped from $46.5 billion to $98.2 billion and full-year 2026 capex is guided to $175 billion to $185 billion. The Finland project Trump referenced is Google's single largest AI investment in Europe, a €13 billion ($15.1 billion) commitment across 2027 and 2028 that includes a 22-year nuclear power deal covering up to 50% of Fortum's Loviisa plant output through 2050, projected to add roughly €3.6 billion to Finnish GDP during construction and support 7,000 jobs annually once operational. Trump's claim that community backlash against data centers is a Chinese PR campaign glosses over hard numbers: U.S. electricity prices rose 6.1% year over year, roughly 61% faster than CPI's 3.8%, and AI data-center draw is projected to rise 14-fold by 2028 to 12% of U.S. electricity consumption.
Apple Unveils $1,999 Foldable iPhone Duo as Nvidia, Oracle Dominate Tech Week
Apple took the wraps off its new iPhone lineup, including the foldable iPhone Duo starting at $1,999 for the 256 GB option and running up to 2 TB, alongside the iPhone 18 Pro at $1,199 and iPhone 18 Pro Max at $1,299. Nvidia CEO Jensen Huang reiterated he is confident in growing revenue 70% next year, noting the upcoming Vera Rubin system is priced at roughly $40,000 versus roughly $25,000 for Blackwell and $18,000 for Hopper, while the Justice Department is investigating whether Nvidia structured its $20B licensing deal with AI chip startup Groq to avoid antitrust scrutiny. Oracle reported fiscal first-quarter results and guidance that topped Wall Street estimates by a significant margin, with cloud revenue at $11.6B, up 62% year-over-year, infrastructure revenue soaring 121% to $7.4B, and application revenue rising 10% to $4.2B. Adobe also reported second-quarter results and guidance above estimates, though analysts noted a slowdown in annual recurring revenue growth amid its freemium AI strategy. In other news, Qualcomm and Amazon announced a partnership to build next-generation AI data center infrastructure, and Microsoft plans to more than triple its global data center capacity to more than 38 gigawatts by 2032, up from about 12 gigawatts currently.
Emerald AI Raises $150 Million for NVIDIA and Digital Realty Power-Flexible AI Factory
Emerald AI has raised $150 million in financing to support commercial expansion of flexible AI infrastructure developed with NVIDIA and Digital Realty, with NVIDIA named among the investors in the August 25 announcement. The capital backs a nearly 100-megawatt research factory in Manassas, Virginia, that the companies now expect online later this year, a later schedule than the first half of 2026 targeted in an October 2025 announcement. The approach coordinates computing workloads with onsite energy resources so a facility can adjust its electricity draw when the grid is under strain, using NVIDIA's software and infrastructure controls alongside Emerald's orchestration technology. Emerald already reports commercial deployment at an entire California data center, though the announcement provides no quantified earnings contribution for Digital Realty and no new order value for NVIDIA. August 14 short interest stood at 1.23% of NVIDIA's float and 2.07% of Digital Realty's.
Equinix Partners With Nvidia on AI Inference Exchange as Market Cap Hits $100 Billion
Equinix has expanded its partnership with Nvidia and launched the Equinix Inference Exchange, giving enterprise customers a flexible way to run AI models using Nvidia's Enterprise Reference Architectures and Together AI's open-source inference platform through Equinix's global data centers and connectivity. The colocation data center company, which dates back to the 1998 dot-com era, operates more than 280 data centers across 77 metros with over 10,500 interconnected customers, and its facilities are optimized for Nvidia's B300 Blackwell Ultra GPUs, with some liquid-cooled sites supporting newer Vera Rubin chips. Equinix shares are up over 33% this year, lifting its market capitalization to $100 billion as the most valuable data center REIT, ahead of Digital Realty's $68 billion. In the second quarter, revenue grew 16% to $2.625 billion, AFFO per share rose 19% to $11.78, adjusted EBITDA margin reached a record 53%, and management delivered its largest guidance increase in company history. The company nearly doubled its capital expenditure guidance to $5 billion to $7 billion annually from $3 billion to $4 billion, and it expects the new Nvidia and Together AI offering to launch commercially in the first quarter of 2027.
Berkshire CEO Abel Warns of Growing Community Pushback Against Data Centers
Berkshire Hathaway CEO Greg Abel told CNBC's Becky Quick on September 2 that there is a lot more pushback in communities across the U.S. against new data center construction. Abel said Berkshire's role in the AI buildout centers on supplying power through Berkshire Hathaway Energy rather than building the facilities themselves, and that the company will only serve large computing customers on the condition that existing customers' rates are not affected. The U.S. currently hosts roughly 4,700 data centers, and New York became the first state to impose a one-year moratorium on large new data center construction, with at least 11 states having similar movements underway. Electricity bills near major data center hubs have climbed as much as 267% over five years, and data centers now account for 4% to 5% of all U.S. electricity consumption. Mizuho analyst Vikram Malhotra flagged data centers as a likely defining political issue heading into the midterms, while Abel said he views grid capacity and permitting timelines, not chips or capital, as the real constraint on the AI buildout, noting data centers already make up about 8% of Berkshire Energy's load in Iowa.