4 Sources
[1]
SpaceX AI Buildout Shows Cracks Ahead Of Google Deadline - SpaceX (NASDAQ:SPCX)
SpaceX AI Buildout Shows Cracks as Musk Shakes Up Data Center Team Space Exploration Technologies Corp. (NASDAQ:SPCX) has reshuffled its data-center leadership after engineering concerns and reliability problems emerged at facilities in Tennessee and Mississippi, The Information reported Tuesday. The shake-up comes as Elon Musk races to expand SpaceX's AI infrastructure business ahead of a Sep. 30 capacity deadline tied to Google's $920 million-a-month compute deal. Jake Palmer, who led physical infrastructure for SpaceXAI, left in late July alongside several other data-center executives, according to the report. SpaceX veterans from its rocket and Starlink businesses have since taken larger roles in the operation. Musk's Speed Advantage Faces a Reliability Test Some facilities operated for months without backup cooling and power systems, The Information reported. The publication separately reported in May that SpaceXAI's Macrohard facility relied on more than 100 mobile chillers and had recorded uptime well below an internal target of 99.9%. Temporary power and cooling systems contributed to outages that interrupted AI model training, according to the report. Trending Execution pressure has also surfaced elsewhere. Mississippi regulators allowed temporary gas turbines to operate longer than planned after supply-chain problems delayed 41 permanent units. SpaceX Had Already Flagged the Risks Weeks before the latest shake-up, SpaceX added a specific AI infrastructure risk factor to its Aug. 4 quarterly filing. The company said its cloud business depends on "reliable data center operations and timely development," identifying construction delays, workforce turnover, power constraints and equipment shortages as risks that could delay capacity or disrupt service. SpaceX had 1.4 gigawatts of compute capacity at the end of June, up from 0.4 GW a year earlier, and spent about $15.8 billion on AI infrastructure during the second quarter. Google's $920 Million-a-Month Deadline Alphabet Inc. (NASDAQ:GOOGL) unit Google agreed to pay SpaceX $920 million per month at full capacity for access to roughly 110,000 Nvidia Corp. (NASDAQ:NVDA) GPUs. SpaceX must deliver the committed GPUs by Sep. 30. After a one-month grace period, Google can terminate the agreement or accept fewer GPUs and reduce payments proportionately if SpaceX falls short. Anthropic has overtaken SpaceX as prediction market traders' favorite to have the largest IPO of 2026. SpaceX led through much of the summer, but Anthropic surged ahead in August as expectations grew that it could list at a valuation approaching $2 trillion. Traders now put Anthropic at 59% versus SpaceX at 40%, with about $5.8 million traded. There is no indication the problems will affect Google's deal. But they test Musk's claim that SpaceX can bring compute online "better than anyone else so far." Image: Shutterstock Markets OpenAI's $300B Deal Could Make Oracle the Weak Link in the AI Boom Oracle's $300B OpenAI deal could become a weak link in the AI boom as rising debt and huge compute commitments raise contagion risk. 3 min read Read this article Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.
[2]
Elon Musk Says Google and Anthropic Turned to SpaceX For AI Compute Because Power is Tight, Says SPCX Bui
Elon Musk said bringing large artificial intelligence data centers online requires far more than chips, as Alphabet Inc.'s (NASDAQ:GOOG)(NASDAQ:GOOGL) Google, Anthropic and other companies lease computing capacity from Space Exploration Technologies Corp. (NASDAQ:SPCX) amid tightening power supplies across the fast-growing AI industry. Musk Warns AI Power Crunch Is Here "To bring large AI data centers online requires building massive power plants, transformers, liquid cooling loops & chillers, as well as incredibly complex networking (especially for training clusters)," Musk wrote on X Tuesday. The comments followed Musk's video appearance at the G20 Innovation Ministerial Summit in Chapel Hill, North Carolina. Reuters reported that Musk urged leaders outside China to develop new energy sources for data centers as he made the case for faster infrastructure expansion. "Even before next year, there are challenges with power, which is why Google, Anthropic and many other companies are leasing compute from SpaceX," Musk told delegates. "We've been able to bring AI compute online better than anyone else so far." Musk has argued that this vertical-integration approach allows SpaceX to move faster than grid-dependent rivals. "Constructing our own power plants is the only way we were able to do it," he added. Tech SpaceX Goes Ultra-Vertical: Why Elon Musk Is Building Its Own Power Supply for AI SpaceX is developing gas-turbine manufacturing in Texas as Elon Musk targets the power bottleneck slowing AI data-center expansion. 3 min read Read this article SpaceX Turns Power Bottleneck Into Business That strategy has become a major business for SpaceX. Google signed a multiyear cloud agreement in June worth $920 million per month at full scale for access to roughly 110,000 Nvidia GPUs. The deal followed SpaceX's expanded Anthropic pact, under which the AI developer agreed to pay $1.25 billion monthly through May 2029 for computing capacity. SpaceXAI is also building a 1.2-gigawatt permanent power plant near its Greater Memphis supercomputer sites while retiring temporary mobile turbines as permanent capacity comes online. Musk has said this vertical-integration approach lets SpaceX move faster than grid-dependent rivals. AI Power Crunch Extends Beyond SpaceX The electricity crunch extends beyond SpaceX. Lawrence Berkeley National Laboratory estimates U.S. data centers could consume 11.8% of national electricity by 2030, with scenarios ranging from 9.5% to 15.3%. Musk, during his video address, also criticized European Union technology rules, saying the approach "inhibits progress." Tech SPCX CEO Elon Musk Mocks AI-in-Space Critics, Says They 'Claim They're Super Smart' But Don't Know Basic Thermal Limits Musk says SpaceX is already engineering around the toughest constraints facing orbital computing as AI's power needs soar on Earth. 3 min read Read this article Photo courtesy: Shutterstock Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.
[3]
SpaceX just targeted Howmet Aerospace, a key AI supplier: The stock tanked
Every industrial supply chain has a part almost nobody notices, until it becomes the reason nothing else can ship. For the natural-gas turbines racing to power artificial intelligence data centers, that part is the blade sitting inside the hottest section of the machine. On Monday, Aug. 31, investors got a blunt reminder of how much that single part matters. Shares of Howmet Aerospace Inc. (HWM), one of the few companies on Earth that knows how to cast those blades, tumbled more than 8% after Elon Musk said SpaceX plans to make its own, according to CNBC. The stock move looked like a simple competitive threat: a customer becoming a rival. Wall Street's read, delivered within hours, was almost the opposite, and it centers on a shortage most investors have never had reason to track or even know about. What Musk actually said about the turbine-blade bottleneck Musk posted on X (the former Twitter) on Aug. 29, saying that Space Exploration Technologies Corp. (SPCX) and Tesla are each racing to build 100 gigawatts of annual solar capacity. Natural gas will still be needed to fill the gap for years, he added. He then named the real constraint. Casting the blades and vanes inside gas turbines is what slows new production, Musk wrote, and bringing that work in-house at SpaceX could get turbines online up to 18 months at sooner, a change he called a "profound game-changer." SpaceX is building that capability at a foundry in Bastrop, Texas, tied to a planned 20-gigawatt power project for AI data centers, according to Seeking Alpha. The location matters because it signals SpaceX wants full control over one of the industry's tightest chokepoints, not just a cheaper supplier. Elon says, Elon does. How Howmet and SpaceX stock moved apart Howmet fell as much as 7.7% to a two-month low during the session, Seeking Alpha noted, before closing down more than 8%, according to CNBC. The stock had closed at $264.85 the prior Friday, Aug. 28, and it carries a market capitalization near $97 billion. On Aug. 31, shares gapped down to open at $248.25. Space Exploration Technologies Corp. shares moved the other way. Options traders were positioning for further upside in SPCX the same day Howmet sold off. The divergence signals investors initially read this as a wealth transfer from one company to another, rather than a shared response to industry-wide scarcity. That matters for anyone holding Howmet. The company makes precision-cast metal components for jet engines and industrial gas turbines, and its gas-turbine segment has become one of its fastest-growing businesses, with revenue climbing39% in the first quarter after a 25% gain for all of 2025. Vithun Khamsong / Getty Images Wall Street calls the sell-off a buying opportunity Bernstein analyst Douglas Harned pushed back on the panic within hours. He wrote that he sees "little risk to Howmet from the SpaceX announcement" and framed the news as a positive signal instead, according to Seeking Alpha. Bloomberg Intelligence reached a similar conclusion. Analyst Omid Vaziri said SpaceX's plan validates turbine scarcity rather than threatens established suppliers. Incumbents such as Siemens Energy AG and GE Vernova Inc. are already expanding their own casting capacity to meet the same shortage, Investing.com confirmed. Harned's argument centers on scarcity, not sentiment. Turbine-blade demand is outpacing global casting capacity, and that imbalance is why a well-funded buyer like SpaceX would rather build its own supply than wait in line. Bernstein said Howmet's supply agreements with major turbine makers extend into 2030. The firm also flagged six additional capacity expansions expected before year-end, which could lift blade capacity by as much as 38% from early 2025 levels. Bernstein raised its price target on Howmet to $328 from $248 and kept its Outperform rating, treating the Aug. 31 decline as an entry point rather than a warning sign. Citi Research joined the pushback, placing Howmet on a 30-day upside catalyst watch with a $329 price target, according to Seeking Alpha. The bank noted that SpaceX entering the casting space demonstrates extreme market demand and tight supply constraints, forecasting earnings to reach $8.08 per share by 2028. AI's power race has a hidden chokepoint The Howmet story is really a supply story, and it extends well beyond one stock. Only four companies worldwide cast the nickel-superalloy blades and vanes that survive inside a turbine's hottest section, and all four are currently running at capacity, Benzinga reported. Howmet is one of only two in that group that trade publicly. Precision Castparts sits inside Berkshire Hathaway, and newly listed Doncasters is the other. The other rivals remain private, according to Benzinga. That means the Aug. 31 sell-off doubled as a rare moment when investors could actually price the entire bottleneck. That scarcity is colliding with an unprecedented wave of demand. Microsoft signed a 20-year agreement with Chevron in June for a 2.67-gigawatt gas-fired power project in West Texas. It is one of several hyperscaler deals now competing for the same turbine backlog, according to Bloomberg. SpaceX's foundry does not eliminate that scarcity. It just proves how far a well-capitalized company will go to jump the line. Investors should watch whether other AI builders follow the same path, because the next chokepoint story may not involve Howmet at all. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published September 1, 2026 at 8:07 PM.
[4]
SpaceX Goes Ultra-Vertical to Power AI - SpaceX (NASDAQ:SPCX)
SpaceX Goes Ultra-Vertical: Why Elon Musk Is Building Its Own Power Supply for AI Space Exploration Technologies Corp. (NASDAQ:SPCX) has built rockets by controlling more of the manufacturing process than traditional aerospace companies. Now Elon Musk is applying the same playbook to AI's biggest physical bottleneck: electricity, with SpaceX developing its own gas turbine component manufacturing in Texas to avoid a power equipment supply chain that has been stretched for years. SpaceX's Power Play SpaceX is laying the groundwork for a foundry in Bastrop, Texas, to produce the blades and vanes used in large gas turbines, The Information reported. SpaceX has been hiring engineers for the facility, including roles tied to materials, automation, tooling and the construction of a new manufacturing line. Musk subsequently confirmed the strategy on X, saying natural gas will still be needed to "supplement and bootstrap solar for several years." He said bringing blade and vane casting in-house could accelerate gas turbines coming online by as much as 18 months, calling the result a "profound game-changer." The reason SpaceX is going this far is simple: AI data centers need enormous amounts of reliable power, and waiting for new grid connections can take too long. AI's Power Bottleneck The hardest part may not be generating electricity itself, but getting the equipment needed to generate it. The Information reported that only a handful of companies can produce the highly specialized turbine blades required for industrial-scale gas turbines, with those foundries already operating near capacity. GE Vernova Inc's (NYSE:GEV) turbine business, for example, has seen orders extending into 2030. The blades operate at temperatures of roughly 3,000 to 3,600 degrees Fahrenheit and require sophisticated casting techniques, making the manufacturing process difficult to replicate quickly, TechCrunch reported. For SpaceX, that makes vertical integration more than a manufacturing preference. It becomes a way to remove a bottleneck standing between an AI facility and the computing power it needs. Tech Elon Musk Just Flagged AI's Next Bottleneck. Only a Few Companies Can Cast the Fix Elon Musk says only three companies can make the turbine blades limiting AI infrastructure. Here's why investors should know them. 3 min read Read this article From Rockets to AI Infrastructure The bigger shift is where SpaceX is applying that philosophy. The company is developing Terafab, an ambitious semiconductor manufacturing project in Texas, while also expanding its AI infrastructure. The turbine effort effectively connects those ambitions: SpaceX wants to produce AI hardware, but it also needs sufficient electricity to power the infrastructure that supports it. That is an unusually broad form of vertical integration. SpaceX isn't simply buying GPUs or leasing data-center capacity; it is increasingly trying to control the physical systems around the compute, from manufacturing to power. The most important signal is not that SpaceX is entering the turbine business. It is that AI demand is pushing even technology companies into traditionally industrial businesses. As power becomes as important as chips in determining how quickly AI infrastructure can scale, companies that control constrained physical inputs could gain an advantage that software alone cannot provide. Tech Tesla's AI Future Is Starting to Look a Lot Less Electric Tesla's $16.8 billion Terafab will use natural gas and batteries for power, underscoring the energy demands of its AI and robotics ambitions. 3 min read Read this article Photo courtesy: Shutterstock Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.
Share
Copy Link
SpaceX has overhauled its data center leadership following reliability problems at AI facilities in Tennessee and Mississippi. The shake-up comes as Elon Musk pushes to expand AI infrastructure and meet a critical September 30 capacity deadline tied to Google's $920 million-per-month compute deal.
SpaceX AI has reshuffled its data center leadership after engineering concerns and reliability problems surfaced at facilities in Tennessee and Mississippi
1
. Jake Palmer, who led physical infrastructure for SpaceX AI, departed in late July alongside several other data center executives. SpaceX veterans from its rocket and Starlink divisions have since assumed larger roles in the operation. The leadership changes arrive as Elon Musk races to expand AI infrastructure ahead of a September 30 capacity deadline connected to Google's $920 million-per-month compute deal1
.Some SpaceX AI facilities operated for months without backup cooling and power systems
1
. The Macrohard facility relied on more than 100 mobile chillers and recorded uptime well below an internal target of 99.9%. Temporary power and cooling systems contributed to outages that interrupted AI model training. Mississippi regulators permitted temporary gas turbines to operate longer than planned after supply chain problems delayed 41 permanent units. SpaceX acknowledged these AI infrastructure risks in its August 4 quarterly filing, identifying construction delays, workforce turnover, power constraints and equipment shortages as factors that could delay capacity or disrupt service1
.Google agreed to pay SpaceX $920 million per month at full capacity for access to roughly 110,000 Nvidia GPUs
1
. SpaceX must deliver the committed GPUs by September 30. After a one-month grace period, Google can terminate the agreement or accept fewer GPUs and reduce payments proportionately if SpaceX falls short. SpaceX had 1.4 gigawatts of AI compute capacity at the end of June, up from 0.4 gigawatts a year earlier, and spent approximately $15.8 billion on AI infrastructure during the second quarter1
. Anthropic also signed an expanded pact with SpaceX, agreeing to pay $1.25 billion monthly through May 2029 for computing capacity2
.Elon Musk emphasized that bringing large AI data centers online requires far more than chips. "To bring large AI data centers online requires building massive power plants, transformers, liquid cooling loops & chillers, as well as incredibly complex networking," Musk wrote on X
2
. He told G20 Innovation Ministerial Summit delegates that power challenges explain why Google, Anthropic and other companies lease compute from SpaceX. "We've been able to bring AI compute online better than anyone else so far," Musk stated2
. SpaceX is constructing a 1.2-gigawatt permanent power plant near its Greater Memphis supercomputer sites while retiring temporary mobile turbines as permanent capacity comes online.
Source: Benzinga
Related Stories
SpaceX is establishing a foundry in Bastrop, Texas, to produce blades and vanes used in large natural-gas turbines
4
. The company has been hiring engineers for the facility, including roles tied to materials, automation, tooling and construction of a new manufacturing line. Musk confirmed the strategy, saying bringing blade and vane casting in-house could accelerate gas turbines coming online by as much as 18 months, calling it a "profound game-changer"4
. Only a handful of companies can produce the highly specialized turbine blades required for industrial-scale gas turbines, with those foundries already operating near capacity. The blades operate at temperatures of roughly 3,000 to 3,600 degrees Fahrenheit and require sophisticated casting techniques4
.Shares of Howmet Aerospace, one of the few companies that can cast turbine blades, tumbled more than 8% after Elon Musk announced SpaceX plans to manufacture its own
3
. However, Wall Street analysts framed the news as validation of turbine scarcity rather than a competitive threat. Bernstein analyst Douglas Harned raised the price target on Howmet to $328 from $248, treating the decline as an entry point. Citi Research placed Howmet on a 30-day upside catalyst watch with a $329 price target, noting that SpaceX entering the casting space demonstrates extreme market demand and tight AI supply chain bottleneck constraints3
. Lawrence Berkeley National Laboratory estimates U.S. data centers could consume 11.8% of national electricity by 2030, with scenarios ranging from 9.5% to 15.3%2
. SpaceX is also developing Terafab, a semiconductor manufacturing project in Texas, connecting its turbine manufacturing effort to broader ambitions in controlling physical systems around AI compute4
.Summarized by
Navi
[2]
05 Aug 2026•Technology

24 Aug 2026•Technology

22 Apr 2026•Business and Economy

1
Policy and Regulation

2
Technology

3
Technology
