Techin' Geek

The creators of Starcloud, a startup aiming to deploy a fleet of GPUs in space, have unveiled a fresh proposal for potential backers: a mission to dispatch a small craft on an 80,000-year expedition to Alpha Centauri.

Should they proceed, it would mark the first human-constructed artifact specifically directed towards another star, making it the sixth to exit our solar system.

Named Fermi Explorer, the non-profit organization initiating this endeavor surfaced today to seek collaboration, funding, and guidance. The founding team—Starcloud’s Philip Johnston, Adi Oltean, and Ezra Feilden, together with program director Garret Jameson—has devised a strategy to utilize established space technologies to propel a craft featuring a 10 cm square payload, weighing 1 kilogram, into the realm of interstellar space.

The founders assert that they are launching this venture in the name of human exploration; to accomplish a historic milestone; and to probe the boundaries of the Fermi paradox, which seeks to clarify our lack of findings of intelligent extraterrestrial life and lends the mission its name.

The mission aims for a budget of approximately $15 million, targeting a launch in 2029. The non-profit is reaching out to suppliers for the construction of the craft, financial backing, and scientific and artistic payloads for delivery, similar to the sensors and Golden Record aboard NASA’s Voyager spacecraft, the first human vessel to venture beyond our solar system.

Most proposals for interstellar exploration falter due to the enormous distances involved and the relatively sluggish methods available to traverse them. Voyager, which was launched in 1977, is currently situated roughly 26 billion km from Earth; Alpha Centauri, the closest star to our solar system, is positioned around 41 trillion km away.

The previous attempt from Silicon Valley to tackle this challenge, a non-profit called Breakthrough Starshot, founded in 2016 by Yuri Milner and supported by Mark Zuckerberg, envisioned a multibillion-dollar initiative employing a groundbreaking laser-based propulsion system to reach Alpha Centauri within two to three decades. This effort came to a halt in 2025, yet the aspiration appears to endure.

Fermi Explorer adopts a contrasting approach: instead of advancing technological boundaries, the objective is to leverage existing capabilities to accomplish something.

“I too became enthusiastic and subsequently let down!” Johnston remarked regarding Starshot. “That’s why we are pursuing this mission. We are determined to actually send something off in three years… the significant distinction between most other [interstellar] missions and Fermi Explorer is that others attempt to achieve it within a human lifetime, necessitating costly billion-dollar propulsion R&D projects. We have intentionally selected 80,000 years as the minimum feasible mission.”

The strategy involves launching the spacecraft via a rocket rideshare, followed by utilizing efficient, low-powered electric propulsion to slingshot the vehicle into a sequence of orbits that will ultimately set it on course for Alpha Centauri. The spacecraft is expected to function for around 12 years prior to commencing a final coast phase towards the other solar system.

“If a dinosaur bone can endure for 200 million years, then a metal box can certainly withstand the anticipated radiation environment for 50,000 years,” Johnston stated. “We do not intend to power the spacecraft at the conclusion of the 80,000 years. Our hope is that a future human civilization will discover it.”

Though battery swapping has faced challenges in gaining traction globally, Canadian auto parts leader Magna International is optimistic about its potential in India, where millions of two- and three-wheeled vehicles along with a rapidly expanding delivery sector provide a unique economic landscape.

Magna is betting on Yuma Energy, a Bengaluru-based company that runs a battery-swapping infrastructure for electric two- and three-wheeled vehicles. Emerging from Indian mobility startup Yulu in early 2023, Yuma has executed over 60 million swaps and currently has around 100,000 batteries in circulation throughout its network.

Magna is set to invest an additional $35 million in Yuma, raising its ownership from the 51% stake acquired during the initial formation of the joint venture, according to Yuma’s managing director Muthu Subramanian in a recent interview. Consequently, Yulu’s 49% ownership will be reduced. Subramanian chose not to reveal the new ownership proportions.

Yulu and Yuma represent Magna’s exclusive investments in Indian startups, as confirmed to TechCrunch. In 2022, Magna allocated a total of $77 million to the two enterprises, with $25 million directed to Yulu and $52 million to the battery-swapping joint venture.

Investing in India’s gig economy

Magna’s recent investment largely depends on the growth of India’s gig economy.

Delivery drivers can lose both time and income while charging their electric vehicles. Subramanian estimates that currently, only about 10% to 15% of vehicles utilized by gig workers in India are electric, indicating significant opportunities for companies like Yuma as more drivers transition from gasoline vehicles.

“Given the high operational hours of Indian gig workers, using an EV becomes highly economical,” Subramanian shared with TechCrunch. “Maintaining uptime is crucial.”

Yuma is focused on those high-usage riders, asserting that swapping batteries is more feasible than fast charging. According to Subramanian, a battery can be exchanged in less than two minutes, while even a quick 20- or 30-minute charge takes a rider off the road and demands additional space and power for multiple vehicles.

However, creating that convenience is costly, as Yuma needs to maintain its battery supply and swapping infrastructure in anticipation of an adequate number of riders. “This is a capital-intensive industry, and the unit economics will materialize once at scale,” Subramanian stated.

Yuma has yet to become profitable, although some of its older swapping stations are already EBITDA-positive, Subramanian pointed out. The company runs over 400 stations with more than 2,500 charging units and concluded the financial year in March 2026 with approximately ₹1 billion (around $10.5 million) in revenue. The goal is to achieve EBITDA break-even in the next two quarters, as per Subramanian’s remarks to TechCrunch.

Achieving this will still necessitate Yuma to anticipate demand. The company intends to allocate a significant portion of Magna’s investment to enhance its swapping infrastructure and double its battery fleet from about 100,000 in the next 12 to 18 months.

Yulu remains the primary contributor to Yuma’s 60 million total swaps, though this reliance is beginning to diminish, as roughly 15% to 20% of swaps in the latest quarter came from clients other than Yulu, Subramanian indicated to TechCrunch.

Aside from Yulu, Yuma now caters to over five fleets and has integrated its batteries with more than 10 different vehicle models, including those from Kinetic Green, Motovolt, BGauss, and Quantum Energy. Subramanian anticipates that non-Yulu clients will represent about 25% of its swaps within the next two years.

Earlier this month, Yulu secured $93 million to expand its electric two-wheeler fleet. Consequently, Yuma will need to grow its network to keep up with its largest client and support newer fleets utilizing its batteries.

Yuma conducts its battery-swapping operations across 18 Indian cities, including Bengaluru, Hyderabad, Mumbai, and the Delhi area, along with Jaipur, Lucknow, Indore, Coimbatore, Kochi, and Kolkata. With the new funding, the company aims to branch out into Chennai and Pune in the upcoming quarters and to add more stations in cities where it currently operates.

India will remain Yuma’s primary focus for at least the next 12 to 18 months. However, Subramanian informed TechCrunch that the company intends to take its model abroad as part of its long-term strategy. Markets in Southeast Asia like Vietnam and Thailand, along with certain regions in Africa, could be viable due to their substantial two-wheeler demographics, he mentioned, although Yuma has not yet initiated discussions about entering those markets.

Distinct from operators who merely manage swapping frameworks, Yuma develops and manufactures its own battery packs and charging units. The firm produces battery packs at its plant in Chennai and charging units in Bengaluru, allowing it to control both the hardware and the network that manages them.

Notable solo investor Lachy Groom has invested in a Bengaluru startup with an ambitious goal: to keep an aircraft airborne for over a year by harnessing energy from ocean winds.

Alteon, established by 20-year-old Samay Sanghvi, announced on Tuesday that it secured $2.5 million in a pre-seed funding round led by Groom, with contributions from Together Fund, to create autonomous aircraft modeled after dynamic soaring, a technique used by albatrosses to utilize wind energy. Groom made the decision to invest within the first half hour of their initial meeting, Sanghvi shared with TechCrunch.

Traditional aircraft must carry the necessary fuel or battery power for flight. Alteon aims to overcome this limitation by creating small, fixed-wing autonomous aircraft capable of harvesting energy from wind shear above the ocean through dynamic soaring—a technique where an aircraft repeatedly transitions between air layers moving at different speeds.

“Once you construct airplanes that can remain airborne for over a year, the possibilities are endless,” Sanghvi told TechCrunch. Alteon intends to initially employ the aircraft for maritime surveillance, providing governments with real-time oversight of activities in their waters.

The current strategy is to develop an aircraft with an approximate wingspan of three meters that will fly near the surface of the ocean, climb, and maneuver through faster-moving air, repeating the process to harness energy from the wind. Ultimately, Alteon plans to utilize its propellers as turbines to convert some of this energy into electricity to recharge its onboard batteries, Sanghvi indicated.

However, the startup has yet to demonstrate that its aircraft can actually maintain flight using energy acquired through dynamic soaring. It has recently completed a test of its autonomous flight system over the Bay of Bengal, where the aircraft autonomously executed seven O-shaped cycles at speeds exceeding 62 miles per hour, flying within one meter of the water’s surface.

The next significant milestone for Alteon will be what Sanghvi refers to as “energy-neutral dynamic soaring.” This would enable the aircraft to fly continuously with its propulsion turned off, drawing sufficient energy from the wind to stay aloft.

Dr. Gabriel Bousquet, an aerospace and robotics engineer based in Silicon Valley, who studied dynamic soaring during his PhD at MIT, described Alteon’s low-altitude flight over water as a “promising initial outcome.” However, he emphasized that the greater challenge will be demonstrating that the aircraft can reliably harvest enough energy from real-world winds to remain airborne for extended durations.

Flying low enough to safely capture that energy is particularly challenging, Bousquet noted to TechCrunch, as the aircraft will face turbulence, waves, spray, rain, and varying light conditions while constantly detecting and responding to a dynamic ocean surface.

Dr. Bharath Swaminathan, who received his PhD from IIT Madras with a focus on the stability of dynamic soaring, remarked that the fundamental physics is well recognized and commended Alteon’s initiative. Keeping an aircraft airborne for several days utilizing dynamic soaring would be “a significant step and a major accomplishment,” he told TechCrunch.

Swaminathan, however, added that while large-scale wind conditions may be predictable, local wind shear and turbulence can fluctuate greatly, complicating an aircraft’s ability to continuously harvest energy from the wind. Some of these challenges, he suggested, may only manifest through practical flight testing.

Groom acknowledged the technical risks involved in this venture. “Ambitious projects are inherently risky,” he told TechCrunch. “For me, it came down to trusting that Samay and the Alteon team are equipped to resolve these issues.”

Sanghvi began developing what would later become Alteon straight after high school in 2023, learning to construct aircraft by building—and crashing—radio-controlled models before creating early prototypes. He officially established Alteon in 2025 and obtained initial support from Emergent Ventures and 1517.

Alteon now boasts a team of 20 in Bengaluru and operates out of a 10,000-square-foot facility. The startup is producing four to five aircraft weekly for testing and has conducted over 200 test flights in the preceding 30 days, Sanghvi noted.

In its lawsuit against OpenAI, Apple has presented what it describes as “shocking evidence” to support its claims that former employees misappropriated trade secrets for the benefit of OpenAI. These details surfaced after the legal representative of former Apple employee Chang Liu — who is currently employed at OpenAI — submitted Liu’s prior Apple work laptop for examination earlier this month.

Apple now asserts that Liu utilized a confidential Apple circuit diagram in his role at OpenAI, in addition to a tool that shares a name with an internal Apple engineering application. The company alleges that OpenAI was “well-aware” of Liu’s access to Apple information, and that Liu enlisted the help of OpenAI colleague Yu-Ting Peng to assist in destroying evidence in June upon discovering that Apple was investigating him.

“The MacBook represents the very limited information Defendants have provided to date (and only after weeks of delay), and demonstrates that Apple is not engaged in ‘fishing expeditions’ but rather that its trade secrets are being utilized and evidence is being obliterated,” the filing states.

Although this new evidence has been redacted from public sight, prior submissions from Apple have featured text messages from Liu — which he adorned with “crying laughing” emojis — indicating he was aware of still having access to Apple documents.

OpenAI has previously defended Liu by asserting that he accessed Apple files only after his employment ended to assist former colleagues who sought his help. “Apple now attempts to deflect the blame to ‘residual access,’ but they fail to mention that this is a common issue at Apple, resulting from their inadequate management of system access when employees depart,” OpenAI stated in a blog post earlier this month.

However, Apple contends that Liu maintained access because he “exploited a rare, previously unknown authentication vulnerability.”

TechCrunch has reached out to OpenAI for comments regarding Apple’s latest allegations.

Apple is pursuing a preliminary injunction — a court directive intending to prevent OpenAI from developing hardware based on Apple’s technology during the ongoing case — as well as expedited discovery, a sped-up procedure for collecting evidence, since the company claims that additional former employees may also be implicated.

According to Apple’s initial submission, over 400 former Apple employees are now employed at OpenAI.

A prominent pro-AI super political action committee is behind a newly formed group that intends to target voters in critical battleground states with advertisements promoting data centers, numerous sources are reporting.

This new entity, named Build American AI, aims to persuade residents about the benefits of having data centers in their local areas. It claims affiliation with Leading the Future (LTF), a super PAC significantly financed by venture capitalists Marc Andreessen and Ben Horowitz, along with OpenAI President Greg Brockman.

Build American AI informed Bloomberg that it will concentrate its advertising efforts in Kansas, Ohio, and Wisconsin, where data center-related issues have surfaced as a key topic in statewide races. It noted that its ad budget amounts to several million dollars.

LTF started with over $50 million in financing and has primarily allocated those funds to support particular politicians, according to Politico. It backs candidates who promote AI, while simultaneously working against those who oppose the technology, sources have disclosed to Politico.

LTF has emerged as a prominent, and not necessarily well-liked, player in AI-driven election politics, resulting in OpenAI publicly separating itself from the group in a blog entry in June.

“There have been inquiries about Leading the Future (LTF), which has obtained backing from our President and co-founder, Greg Brockman, and his spouse Anna,” OpenAI stated. “OpenAI does not guide LTF’s activities or have insight into their operations.”

OpenAI also criticized some strategies employed by LTF, stating: “Organizations advocating for AI should clearly articulate their policy positions, be transparent about whom they represent, and refrain from tactics like astroturfing that conceal the real decisions facing lawmakers and the public.” (Astroturfing refers to creating a paid, organized campaign that appears to be spontaneous grassroots support.)

While LTF has captured attention in Silicon Valley, it is not the sole super PAC advancing an AI agenda backed by prominent Valley figures. Anthropic, for instance, is a recognized supporter of Public First Action, a group that endorses AI regulation.

In the meantime, Marc Andreessen and Ben Horowitz have been “investing in politics like no other,” as The New York Times noted. According to the outlet’s estimates, their venture firm, Andreessen Horowitz, stands as the largest donor to the midterm elections so far, outspending notable donors such as Elon Musk and George Soros. The firm has already distributed over $115 million in disclosed federal contributions. Beyond LTF and AI initiatives, it is also strongly supporting pro-crypto movements, particularly through the crypto super PAC Fairshake.

Amazon is encountering a fresh lawsuit from the Federal Trade Commission (FTC) alongside 22 states, alleging that the company has been covertly charging businesses higher fees for advertising on its platform.

Filed on Monday, the lawsuit asserts that Amazon has been quietly raising the costs for advertisers participating in its online ad auctions for over seven years. The complaint indicates that this purported practice impacted more than 1 million brands and sellers, potentially generating tens of billions in extra revenue for Amazon.

The 22 states cooperating with the FTC include Alaska, Arizona, California, Colorado, Florida, Idaho, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Nebraska, New Jersey, New York, North Carolina, Oklahoma, Pennsylvania, Rhode Island, South Carolina, Vermont, and Washington.

The lawsuit focuses on Amazon’s Sponsored Products ads, Sponsored Brands ads, and Display ads that appear alongside its search results. According to the FTC, Amazon informed over 500,000 small and medium-sized enterprises that it operated a “second-price” auction in which the winning advertiser would only pay one cent more than the next highest bid, instead of the entirety of their own bid. Businesses operated under the assumption that they would pay just slightly above the runner-up, leading them to bid high, trusting that the system would manage their true costs.

However, the FTC claims that beginning in 2019, Amazon executed a “secretive” change without notifying advertisers. It incorporated a concealed surcharge that Amazon referred to internally as a “soft reserve price,” and employed what one internal document described as an “invented auction participant” — essentially a fictitious bidder — to inflate prices beyond what genuine competition would have yielded.

The complaint contends this constituted a shill bid: rather than the price arising from a legitimate competing advertiser, Amazon was artificially generating a higher figure for advertisers to surpass. Consequently, the FTC states that Amazon charged Sponsored Products advertisers their full winning bid nearly 80% of the time — effectively converting what was promoted as a second-price auction into a first-price auction.

The FTC argues that Amazon initiated this change to boost its advertising revenue and concealed it to prevent advertisers from lowering their bids, which would negatively impact that revenue.

Last year, the company achieved over $68 billion in advertising revenue. 

In a blog entry, Amazon portrayed the FTC’s lawsuit as “misguided,” maintaining that the complaint “fundamentally misinterprets how advertisers function.”

The company further stated that its auctions assess billions of bids across various placements and formats, resulting in price fluctuations, and affirmed that advertisers are “correctly” informed regarding the pricing system.

The Pentagon has introduced versions of OpenAI’s ChatGPT and xAI’s Grok, providing 3 million civilian and military personnel with access to generative AI tools designed to meet “warfighter needs.”

ChatGPT Mil and Grok for Government — specialized variations of the well-known consumer generative AI tools — have now become part of GenAI.mil, a centralized, secure portal established last year.

GenAI.mil, which initially offered Google Gemini upon its launch, is aimed at facilitating Department of Defense employees to access commercial cutting-edge AI models while preventing the transmission of sensitive government information through typical consumer platforms. Specifically, the official military version is exempt from the data collection that’s challenging to avoid in consumer technology products.

Furthermore, the Department of Defense has indicated that the generative AI portal has garnered significant user interest. According to the department, GenAI.mil has onboarded over 1.7 million unique users out of the department’s 3 million personnel.

These new additions are part of the Pentagon’s larger initiative to leverage AI tools to enhance productivity and provide the Department of Defense with a competitive advantage while maintaining security. They also underscore the absence of Anthropic’s Claude model and the Pentagon’s attempts to collaborate with other companies after its disagreement with the frontier lab. The Trump administration labeled Anthropic a supply-chain risk — a classification it is currently contesting in court — after the company declined to permit the Pentagon unrestricted access to its AI tools and insisted on specific safety measures.

ChatGPT Mil, developed through the company’s OpenAI for Government initiative, will provide an experience akin to that of commercial ChatGPT users. According to the Defense Department, the tool will emphasize chat, files, projects, and custom GPTs. It will also assist with document-heavy, routine unclassified tasks such as administrative duties, logistics, planning, and policy. Additional features will be integrated over time.

The Pentagon has described Grok’s functionalities and potential applications in significantly broader and more militaristic terms.

“Starshield AI’s Grok for Government will deliver immediate productivity enhancements for the warfighter, improved knowledge continuity, and more secure and efficient collaboration,” as stated in the Defense Department’s press release. (SpaceX’s Starshield AI represents a secure satellite network that utilizes the company’s existing Starlink technology.)

The department further stated that Grok would enable its military “to carry out missions more swiftly and with enhanced precision across various operational scenarios, ranging from market research analysis for acquisition specialists to supply-chain management for logisticians.”

The Pentagon has forged partnerships with several other tech firms to bolster its AI capabilities, including Amazon Web Services, Microsoft, Nvidia, and Reflection AI.