Data centers and industrial facilities spend millions on uninterruptible power supply (UPS) batteries that sit dormant 99% of the time, waiting for an outage. To unlock this stranded capacity, Singapore-based energy tech startup [Amplicity](https://amplicity.energy) has closed a $1.01 million seed round backed by [Loyal VC](https://loyal.vc) to scale its battery optimization platform. The funding will support the company's expansion across the Asia-Pacific region and a planned entry into Japan in 2027.

The economics of spaceflight are splitting into two distinct markets: heavy-lift rockets that get mass to low Earth orbit, and specialized "space tugs" that handle the last-mile delivery to deeper space. As NASA accelerates its lunar infrastructure plans, it is increasingly relying on this new commercial logistics layer to move hardware across the cislunar domain.

The generative video sector is facing a critical bottleneck: most foundation models render entire clips in a single pass, limiting their ability to generate long, seamless sequences. To solve this, [Sand.ai](https://sand.ai) has secured a $200 million funding round, pushing its valuation past the $1 billion unicorn threshold. The capital infusion signals strong investor appetite for alternative video architectures, even as major players like [OpenAI](https://openai.com) and [Runway](https://runwayml.com) dominate the headlines.

The physical footprint of the artificial intelligence boom is colliding with the limits of the terrestrial power grid. In response, Portland-based ocean-tech startup [Panthalassa](https://panthalassa.com/) is taking AI infrastructure off the grid entirely. The company is setting up a pilot manufacturing facility in Vancouver, Washington, in partnership with Thompson Metal Fab, to build its Ocean-3 floating data centers, according to local reports on Saturday.

As [SpaceX](https://www.spacex.com) begins to sunset its workhorse Falcon 9 in favor of Starship, satellite operators are facing a looming shortage of medium-lift launch capacity. [Stoke Space](https://www.stokespace.com) is positioning itself to fill this void with a fully reusable architecture designed for rapid turnaround, but achieving airplane-like flight cadences requires specialized ground infrastructure that traditional launch pads cannot support.

The biggest bottleneck in physical AI is no longer compute power or algorithmic architecture—it is the sheer lack of training data. While large language models can scrape the entire internet for text, robotics companies are forced to rely on expensive, slow, and often ineffective real-world video collection. To bridge this gap, Paris and San Francisco-based startup Physicl has secured $10 million in seed funding led by Inovia Capital to generate high-fidelity synthetic data for robotic training.

Europe's space sector is facing a severe bottleneck. With legacy programs like Ariane struggling to meet the continent's projected demand of 62 launches by 2034, satellite operators are increasingly desperate for reliable rides to orbit. Capitalizing on this capacity gap, Colorado-based launch startup [LEAP](https://leapspace.one) has signed a Memorandum of Understanding with Highlands and Islands Enterprise (HIE) to establish recurring launch operations at Sutherland Spaceport in northern Scotland.

The commercial space industry has long operated on a clear division of labor: satellite companies build the payloads, and rocket companies provide the ride. That dynamic is shifting as direct-to-cell satellite operator [AST SpaceMobile](https://ast-science.com) announced plans to build its own 20-ton reusable rocket, dubbed Blacklift. To power the new launch vehicle, the company has struck a $1.3 billion agreement with Texas-based [Firefly Aerospace](https://fireflyspace.com) to acquire 36 rocket engines and the intellectual property rights to manufacture them.

The enterprise AI market is rapidly shifting from selling standardized software seats to deploying autonomous digital workers. While generic foundation models excel at drafting emails or summarizing documents, they often falter in highly regulated environments that require deterministic logic and strict governance. To bridge this gap, Austin-based [Future Secure AI](https://futuresecure.ai) has secured $80 million in new funding, reaching a $1.2 billion valuation.

The logistics industry is increasingly turning to autonomous systems to solve last-mile delivery bottlenecks. [Zipline](https://www.flyzipline.com) has raised $1.5 billion in a round led by [Paradigm](https://www.paradigm.xyz), WCM Investment Management, and Layer Global. The fresh capital brings the company's post-money valuation to $21.5 billion, a massive jump from its $7.6 billion valuation in January.

The rapid proliferation of AI assistants is creating a fragmented new landscape for digital commerce. As consumers increasingly rely on chatbots and autonomous agents to discover and purchase products, merchants face the impossible task of building custom integrations for every emerging AI platform. This shift threatens to lock traditional retailers out of the next generation of digital storefronts.

While [Blue Origin](https://www.blueorigin.com) is aggressively expanding the capabilities of its New Glenn rocket with a new third stage and a low-cost booster engine, customers will have to wait until 2027 to see it fly. The dual reality of ambitious hardware development and operational delays highlights the growing pains of scaling heavy-lift commercial spaceflight.

The traditional divide between space hardware suppliers and satellite network operators is rapidly blurring. Austin-based [CesiumAstro](https://www.cesiumastro.com) has raised a massive $730 million Series D round led by Janus Henderson Investors, propelling its valuation past $4.5 billion. The funding marks a strategic pivot for the company, which has historically supplied active phased array antennas and software-defined radios to other spacecraft manufacturers.

The software development landscape is rapidly shifting from AI copilots that autocomplete code to autonomous systems capable of orchestrating entire projects. [Crawn](https://crawn.ai/), previously known as AutonomLayer, enters this space with a platform that divides the software lifecycle among six specialized AI teams: Research, Product, Design, Marketing, Security, and Engineering. Rather than relying on a single large language model to ingest a massive context window and output a finished application, Crawn breaks development into bounded, verifiable steps.

As the commercial space industry races to deploy mega-constellations in low Earth orbit, the logistical nightmare of managing thousands of satellites from the ground is becoming a critical bottleneck. To solve this, Los Angeles-based space data center startup Orbital has tapped Y Combinator-backed [Orca Aerospace](https://orcaaerospace.com) in a deal worth up to $20 million to deploy autonomous navigation software across its planned satellite fleet.

For nearly two decades, [Dropbox](https://www.dropbox.com) built its business on helping humans store and share files. Now, it wants to provide the same foundational memory layer for artificial intelligence. The company announced Thursday it will acquire Sunnyvale-based [Tigris Data](https://tigrisdata.com) for $350 million in cash. The deal, expected to close in the fourth quarter, represents Dropbox's largest acquisition to date, surpassing its $230 million purchase of HelloSign in 2019.