The AI power infrastructure trade is no longer a side bet. Two deals from the first week of May 2026 confirm that the $725 billion AI buildout runs on land, grid access, and energy storage. Both Hut 8 Corp. (NASDAQ:HUT) and Fluence Energy, Inc. (NASDAQ:FLNC) just locked in structural positions inside that demand. However, a third development, reported by The Wall Street Journal on May 12, introduces a long-horizon risk that most coverage of these two stocks has not addressed.



The Lease Structure Is What Investors Should Actually Price



On May 6, Hut 8 signed a 15-year, triple-net, take-or-pay lease at its Beacon Point campus in Nueces County, Texas. The base-term contract value is $9.8 billion. The tenant remains confidential but carries a high-investment-grade credit rating.



The structure matters more than the headline figure. A triple-net lease means the tenant covers all operating costs. A take-or-pay clause means the tenant pays even without using the facility. Together, these terms convert a real estate play into something closer to a contracted utility revenue stream. Expected average annual net operating income is $655 million upon stabilization, per Hut 8's May 6 press release. With three five-year renewal options, the total contract could reach $25.1 billion.



The deal also brought Hut 8's total contracted AI capacity to 597 MW. Aggregate base-term contract value across both campuses now stands at $16.8 billion. The facility runs on Nvidia Inc. (NASDAQ:NVDA)'s DSX reference architecture. Execution partners include American Electric Power Company, Inc. (NASDAQ:AEP), Vertiv Holdings Co (NYSE:VRT), and Jacobs Solutions Inc. (NYSE:J). AEP Texas has executed an interconnection agreement for 1,000 MW of utility capacity. Initial energization arrives in Q1 2027. First data hall delivery follows in Q3 2027.



Hut 8 shares jumped more than 30% on the day. Needham subsequently raised its price target on HUT to $12.



Fluence's Hyperscaler Agreements Signal a Category Shift



Before May 7, Fluence was a battery storage company trying to break into the data center market. After May 7, it is a pre-qualified global supplier to at least two of the world's largest AI infrastructure spenders. That distinction is what investors should focus on, not the quarterly revenue miss.



Here is what actually happened. Two separate hyperscalers each ran structured competitive processes to find an energy storage partner. One process started with 26 vendors. Fluence cleared every round first and signed a global master supply agreement before any competitor, per CEO Julian Nebreda on the May 7 earnings call. The other customer set requirements so specific that most rivals could not meet them. Fluence qualified there too.



That kind of process is not a handshake deal. Hyperscalers run procurement this way because they are committing future data center capacity to a single supplier's technology. Winning both signals that Fluence's battery systems solve a problem that most competitors cannot. That problem is power quality. AI GPU clusters surge and drop power demand in milliseconds. Standard grid connections cannot absorb that volatility cleanly. Fluence's systems, with advanced controls built directly into hardware, can.



The first order under one of these agreements is expected in Q3 fiscal 2026. That is the moment the agreements become revenue. Until then, they represent access, not income. However, access to a hyperscaler's procurement pipeline is itself a structural position. Once a supplier qualifies at this level, switching costs for the customer are high. Fluence now sits inside that relationship.



The record $5.6 billion contracted backlog and the doubled year-to-date order intake support the momentum story. But the hyperscaler agreements are the signal that changes the nature of what Fluence is, from a grid-scale storage vendor into a named supplier for the AI infrastructure build-out.



Why Power Quality Is the Bottleneck Both Companies Are Solving



Nvidia CEO Jensen Huang put the investment case for both companies in plain language in a March 2026 post on the Nvidia blog. He described AI as a five-layer stack: energy, chips, infrastructure, models, and applications. Then he wrote: “Every successful application pulls on every layer beneath it, all the way down to the power plant that keeps it alive.” That is not a marketing line. It is the clearest available description of why energy infrastructure is not peripheral to the AI trade. It is foundational to it.



Hut 8 and Fluence both operate at the base of that stack. Every AI query running on Nvidia hardware traces back to a grid connection and a stable power supply. Without those, every layer above stops. That is the problem both companies are solving, and they are solving it from different angles.



Hut 8 solves it at the site level. Its power-first development model secures utility-scale grid access before a tenant exists. That means it can offer a ready site with locked-in power capacity. Most competitors cannot do that. CEO Asher Genoot made the point directly on the Q1 2026 earnings call: Hut 8 underwrites assets that single-use-case developers overlook, then redirects them toward higher-value uses as demand evolves. Beacon Point started as a crypto mining site. Today it carries a $9.8 billion AI lease.



Fluence solves a different part of the same problem. AI GPU clusters generate extreme, millisecond-level power swings. Those surges damage equipment and reduce performance. Standard grid connections cannot absorb that volatility on their own. Fluence's battery systems include advanced controls built directly into hardware. Those controls smooth the fluctuations before they reach the compute layer. CEO Nebreda confirmed on the May 7 call that quality of power, not just volume of power, is the primary requirement both hyperscalers brought to their procurement processes. Together, these two companies address the two most urgent constraints at Layer 1 of Jensen's stack: getting power to the site, and making that power stable enough for AI-grade compute.



The Orbital Risk That Changes the Long-Duration Case



Both investment theses rest on one assumption: AI compute stays on the ground. The Wall Street Journal reported on May 12 that Alphabet Inc.'s Google (NASDAQ:GOOGL) is in talks with SpaceX to launch orbital data centers. SpaceX filed with the FCC in January to deploy up to one million satellites for this purpose. Furthermore, on May 6, Anthropic confirmed it has expressed interest in partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity, per a joint CNBC report.



Orbital compute directly removes the two constraints that Hut 8 and Fluence exist to solve. Space-based data centers draw on continuous solar power with no grid required. They also eliminate land acquisition, ERCOT interconnection queues, and the ground-level power volatility that makes Fluence's battery systems necessary. If that model scales, both companies lose the structural advantage their current contracts are built on.



However, the economics are not close to working today. SpaceX's standard rideshare launch price runs $7,000 per kilogram, per Tom's Hardware. Google's own Project Suncatcher breakeven math requires launch costs near $200 per kilogram. That gap is enormous. OpenAI CEO Sam Altman said at the India AI Impact Summit in February 2026 that orbital data centers are “not something that's going to matter at scale this decade.” Gartner VP analyst Bill Ray went further, calling the concept “peak insanity” in a Gartner report titled “Orbital Datacenters Won't Serve Terrestrial Needs, so Focus on Earth.”



The near-term thesis for both stocks remains intact. Nevertheless, investors holding positions across a 15-year contract horizon should treat Google's Project Suncatcher prototype launch, targeting 2027 with partner Planet Labs, as a live signal worth tracking. That prototype is the first real-world test of whether the orbital economics can close. If costs compress faster than Altman's timeline, the ground-based infrastructure trade has a ceiling that current valuations do not reflect.



What to Watch



For Hut 8: Q1 2027 energization at Beacon Point is the first hard delivery date. Any delay pushes the $655 million average annual NOI contribution further out. Also watch for Phase 2 leasing at the 1,000 MW campus and any announcements from Hut 8's 7,500 MW broader pipeline.



For Fluence: The Q3 fiscal 2026 first hyperscaler order is the single most important near-term catalyst. A confirmed order converts the MSAs into booked revenue. Watch the Q3 earnings call for any upward revision to full-year guidance of $3.2 billion to $3.6 billion in revenue or $40 million to $60 million in adjusted EBITDA.



For the long-duration risk: Track Google's Project Suncatcher prototype launch, expected in 2027. That is the earliest moment investors will have real data to weigh against Altman's math and Gartner's skepticism.​​​​​​​​​​​​​​​​ If that prototype succeeds and SpaceX’s IPO, reported at a $1.75 trillion valuation, drives further cost compression, the ground-based power infrastructure trade will face a fundamental challenge. For now, neither HUT nor FLNC faces that threat before 2030 under any credible scenario. However, the best time to understand a structural risk is before it prices in, not after.



Benzinga Disclaimer: This article is from an unpaid external contributor. It does not represent Benzinga’s reporting and has not been edited for content or accuracy.

FLNC

BESF

The AI power infrastructure trade is thriving, highlighted by Hut 8's $9.8 billion lease and Fluence's agreements with hyperscalers. Hut 8's lease, a 15-year triple-net contract, ensures stable revenue, while Fluence's new status as a global supplier marks a significant shift in its business model. Both companies address critical power quality issues essential for AI operations, with Hut 8's capacity reaching 597 MW and Fluence securing a record $5.6 billion backlog. These developments underscore the foundational role of energy infrastructure in the AI sector.

benzinga_article

Hut 8 Mining

Fluence Energy

Bastion Energy ETF

Franklin Exponential Data ETF

SPDR S&P Kensho Intelligent Structures ETF

Global X Data Center & Dgtl Infrs ETF

Global X Robotics & Artificial

ISHRS S&P Glb It

Proshares Big Data Refiners ETF

iShares US Digital Infrastructure and Real Estate ETF

Global X Cloud Computing ETF

ARK Autonomous Tech & Robotics ETF

NVIDIA

American Electric Power

Vertiv

Jacobs Solutions

NVIDIA Corporation

The AI Power Infrastructure Trade Has Never Been Stronger, But One Space Race Could Change That