Mobile Battery Storage Gets a 56% Capacity Boost as AI Data Centers Outrun Grid Capacity

The rapid expansion of artificial intelligence workloads is pushing data center power demand far beyond what the U.S. electricity grid can deliver in the near term. Hyperscalers planning dozens of new facilities face interconnection queues that stretch years, substation upgrades that take even longer, and permitting timelines that are the longest of all. The result is a growing split in the industry: one conversation focuses on building more grid infrastructure — necessary but slow — while the other focuses on what operators can do in the meantime. That second conversation is where capital has been flowing.

Recent market moves illustrate the trend. Bloom Energy jumped roughly 12% in a single session after a $1.7 billion project backed by Industrial Development Funding and Oaktree committed its fuel cells to power a Nebius AI data center campus behind the meter. Fluence Energy climbed over 10% in early July on accelerating storage demand, carrying a record backlog of about $5.6 billion and master supply agreements with hyperscalers. Investors have signaled they are willing to pay a premium for any solution that delivers electrons without waiting for utility construction.

One of the least glamorous but most practical answers is putting the battery on a truck. NOMAD Power Solutions, Inc. (NASDAQ: NMAD), a company that until three weeks ago was an oncology drug developer named LIXTE Biotechnology Holdings, has pivoted to industrial power equipment through its wholly owned subsidiary Nomad Transportable Power Systems. The subsidiary builds the Voyager series: utility-grade battery energy storage systems integrated into semi-trailers that can be transported to a site and operational in under an hour — essentially a delivery rather than a construction project.

On July 8, NOMAD announced the third generation of its Voyager fleet. The Voyager Eagle and Voyager Falcon saw storage capacity rise from 1.3 MWh to 2.025 MWh, a 56% increase. The smaller Voyager Hawk went from 664 kWh to 1.0 MWh, a 51% gain. Critically, the footprint, rated power output (999 kW for the Eagle, 500 kW for the Falcon and Hawk), voltage (480V), and deployment time (under one hour) all remained unchanged. The result: runtime at full load for the Eagle now reaches roughly 2.0 hours (up from 1.3), the Falcon extends to about 4.0 hours (up from 2.6), and the Hawk doubles its useful window to about 2.0 hours. The same pad position, same interconnection, same permit — but considerably more energy per unit.

For data center operators, the difference between covering a peak and covering an entire event is not academic. It determines whether a mobile unit serves as a stopgap or becomes a genuine piece of infrastructure. The upgrade came from integrating Octillion Power Systems' prismatic lithium iron phosphate pack architecture, liquid cooling, and dedicated HVAC into the same platform, with units designed to meet NFPA 855 fire safety codes, UL1973 certification, and LFP systems undergoing UL9540A testing. The company frames its approach as improving the platform as battery chemistry advances, rather than locking customers into one generation of hardware.

NOMAD’s addressable market is growing quickly. Battery energy storage for data centers is projected to expand from roughly $4.96 billion in 2026 to about $18.79 billion by 2036, a compound annual growth rate near 14%, driven by the same AI and hyperscale workloads that are outpacing grid capacity. Established industry players are converging on similar insights. Fluence Energy launched its Smartstack 10 MWh system this year, expanding capacity while keeping the same electrical architecture and deployment model — precisely the logic NOMAD applied to Voyager at a smaller scale. Fluence has roughly 50 GWh deployed across nearly 300 projects in 48 markets, a record backlog near $5.6 billion, and second-quarter fiscal 2026 revenue of $464.9 million. Vertiv, which supplies power distribution, uninterruptible power supplies, and thermal management to data centers, has also pushed into liquid cooling for high-density AI racks and opened new manufacturing capacity in Malaysia in July.

NOMAD’s corporate transformation is dramatic. Until July 2, the company was LIXTE Biotechnology Holdings, an oncology drug developer trading under the ticker LIXT. It completed a merger with NOMAD Transportable Power Systems, changed its name, and began trading as NMAD on July 6. The company continues to maintain legacy life sciences assets but is executing its primary strategic focus through NOMAD Power Solutions. On July 15, it announced strengthened executive appointments at the subsidiary level. The market capitalization, currently around $105 million to $120 million, reflects a company investors have not yet fully evaluated. Shares have traded between roughly $2.36 and $8.38 over the past year.

Competition is intense. Bloom Energy generates roughly $2.4 billion in revenue selling solid-oxide fuel cells that produce on-site electricity without a grid connection, with a multi-decade backlog tied to partners such as Oracle, Nebius and Brookfield. At the other end, Generac generates around $3.6 billion in U.S. revenue from backup power solutions, including diesel and natural gas generators that still dominate bridge power for data centers. NOMAD’s pitch for transportable storage is that it replaces diesel without fuel logistics, emissions or noise — and the third-generation Voyager’s runtime extension targets applications where diesel still wins.

What comes next for NOMAD is measurable: whether higher-capacity Voyager units convert into orders, whether legacy life sciences assets are sold cleanly, whether the subsidiary leadership build-out translates into commercial execution, and what the upcoming August earnings report shows. The risks are those of a microcap in transformation: a new name, a new industry, a recent merger, legacy assets to divest, and competition with companies far larger in capital, customers and operating history. Still, the underlying insight holds: the AI buildout is throttled by electricity, not silicon, and the constraint will not resolve on the grid’s timeline. Every megawatt-hour that can arrive on a truck and be running within an hour is a megawatt-hour that does not have to wait three years for an interconnection study. NOMAD just made each of those trucks carry more than half again as much energy without changing anything else about them.