Techin' Geek

YouTube is broadening its budget-friendly Premium Lite subscription service at $7.99 a month, introducing new features such as downloading videos for offline viewing and allowing videos to play in the background while the screen is off or when using other applications. These features were previously exclusive to users on the full subscription plan priced at $13.99 a month.

The company indicated that these enhancements stemmed from user feedback, as participants in its pilot initiative expressed a desire for these functionalities to enhance the appeal of the subscription.

Launched in March, YouTube Premium Lite provided a cost-effective subscription option aimed at eliminating ads from “most” videos on the platform, encompassing popular categories such as gaming, fashion, beauty, cooking, news, and others. However, advertisements will still appear on music content and music videos. Moreover, Lite plan subscribers will not have access to the ad-free YouTube Music application.

With these new features, ad-free music content will now be the sole incentive to upgrade to the full Premium subscription. It is also expected to enhance the Lite tier’s attractiveness to customers who were reluctant to pay solely for ad-free content but were seeking additional benefits.

The Lite subscription option was initially rolled out in Thailand, Germany, and Australia, before its arrival in the U.S. last year. It is currently available in various other global markets, including Canada, Brazil, the U.K., India, Mexico, and other regions in Europe and Asia.

YouTube’s subscription model has been consistently expanding. When combined with advertising, YouTube’s total revenue hit $60 billion in 2025, based on information provided by its parent company Alphabet earlier this month during its Q4 earnings announcement.

The company also disclosed that YouTube’s advertising revenue grew by 9%, reaching $11.38 billion in the fourth quarter. At the same time, the “subscriptions, platforms and devices” segment saw a revenue growth of 17% to $13.6 billion in Q4, which the company attributed to robust expansion in YouTube subscriptions, particularly for YouTube Music and YouTube Premium.

Alphabet reported over 125 million YouTube Music and YouTube Premium users globally as of March 2025. While the company did not provide an updated statistic in its recent Q4 earnings report, it noted that the total now surpasses 325 million paid subscriptions across consumer services, including YouTube Premium and others like Google One.

On Tuesday, Google revealed that the generative AI music platform ProducerAI will join Google Labs.

Supported by The Chainsmokers, the ProducerAI service enables users to compose music using natural language prompts — such as “create a lofi beat” – to generate tracks. It employs Google DeepMind’s Lyria 3 music-generation model, capable of transforming text and even images into audio outputs.

Last week, Google announced plans to integrate Lyria 3 features into the main Gemini app, but ProducerAI allows users to interact with the AI model more as if it were a “collaboration partner,” according to Elias Roman, Senior Director of Product Management at Google Labs.

“ProducerAI has opened up new ways for me to create,” Roman expressed in a blog entry. “I’ve played with genre fusions, crafted personal birthday songs for friends and family, and produced custom workout playlists for myself and others.”

[embedded content]

Google also disclosed that three-time Grammy winner Wyclef Jean utilized the Lyria 3 model and Google’s Music AI Sandbox for his latest track “Back From Abu Dhabi.”

“This isn’t merely a machine where you push a button a hundred times and you’re finished. It involves a thoughtful curation process where you sift through options and say, ‘I believe that’s something we can work with,’” stated Jeff Chang, Director of Product Management at Google DeepMind, in a video released by the company.

Jean reflected on his curiosity about how a flute might sound in an already recorded track, and his ability to swiftly incorporate a flute sound using Google’s tools.

“What I want everyone to grasp […] is that we now live in an era where human creativity must prevail,” Jean mentioned in the video. “One advantage you have over AI: a soul. One advantage AI has over you: infinite information.”

AI in the music industry

Some musicians have vigorously opposed the integration of AI tools in music creation, as it’s almost certain that generative AI tools were trained on copyrighted materials from artists without permission. Numerous musicians, including big names like Billie Eilish, Katy Perry, and Jon Bon Jovi, endorsed an open letter in 2024 urging tech companies not to compromise human creativity through AI music generation tools.

A group of music publishers has also recently filed a lawsuit against the AI firm Anthropic for $3 billion, alleging that the company illegally downloaded over 20,000 copyrighted songs, inclusive of sheet music, lyrics, and compositions. (Anthropic had previously been ordered by the court to offer a $1.5 billion settlement to authors whose books were used without authorization for AI training.)

Conversely, certain artists have welcomed this technology’s potential to enhance audio quality rather than seek creative support.

Paul McCartney utilized AI-based noise reduction technology — similar to what Zoom or FaceTime use to eliminate unwanted background sounds during video calls — to restore an old, low-quality demo by John Lennon. The resultant “new” Beatles song, “Now and Then,” garnered a Grammy in 2025.

In the meantime, AI music generation tools like Suno have produced synthetic music that is sufficiently realistic to dominate charts on Spotify and Billboard. Telisha Jones, a 31-year-old from Mississippi, transformed her (allegedly organic) poetry into the viral R&B track “How Was I Supposed To Know” using Suno, landing a record deal with Hallwood Media worth an estimated $3 million.

Legal clarity surrounding the use of copyrighted works as training data remains ambiguous — one federal judge, William Alsup, determined last year that training on copyrighted data is permissible, while piracy is not.

The fintech powerhouse Marquis has initiated legal proceedings against its firewall service provider SonicWall, asserting that a previous breach facilitated hackers in stealing sensitive data related to customer firewalls, culminating in a ransomware incident affecting Marquis’ infrastructure.

The legal action, lodged on Monday in the U.S. District Court for the Eastern District of Texas, calls for a jury trial. It asserts that the 2025 compromise at SonicWall “revealed vital security information for Marquis and all clients utilizing SonicWall’s firewall cloud backup service.”

Satin Mirchandani, Marquis’ CEO, informed TechCrunch in a statement that SonicWall purportedly neglected to safeguard its backup service, leading to “significant damage to reputation, operations, and finances” for the company.

The announcement of the lawsuit follows weeks after TechCrunch disclosed that Marquis intended to pursue restitution from SonicWall. The fintech giant, based in Plano, Texas, had communicated to its clientele that it held SonicWall accountable for the breach that allowed hackers to exfiltrate sensitive data about customer firewall configuration files, including its own.

“SonicWall permitted a threat actor to acquire the means to circumvent that protective barrier and gain access to Marquis’s internal network, precisely what SonicWall’s firewall was designed to prevent,” the complaint states.

Firewalls are designed to block unauthorized access to corporate networks, but Marquis contends that the ransomware incident was caused by hackers utilizing information stolen from SonicWall regarding customer firewall configurations, including emergency access codes (referred to as scratch codes) that granted entry to Marquis’ internal network.

Marquis, which enables numerous banks and credit unions to analyze their customers’ data, indicated that the hackers acquired “personally identifiable information related to customers of several of Marquis’s financial institution clients” during the cyberattack.

The compromised information encompasses customer names, birthdates, residential addresses, and financial data, including bank account, debit, and credit card numbers, as well as Social Security numbers of the customers.

A representative from SonicWall has not yet responded to inquiries regarding the lawsuit.

SonicWall first acknowledged a breach in its systems in mid-September, claiming that less than 5% of its customer firewall configuration backup files had been exfiltrated from its storage servers, which are hosted on Amazon’s cloud and managed by SonicWall. In October, the firewall manufacturer admitted that in fact every customer had their firewall backup files compromised in the breach.

In December 2025, Marquis began alerting those affected that its networks had been infiltrated the previous August. SonicWall has not disclosed when hackers first gained entry to its systems.

The exact cause of the breach at SonicWall remains uncertain. Marquis asserts in its complaint that SonicWall altered the code in one of its APIs months prior, in February 2025, which “introduced a vulnerability that could be exploited by threat actors.” Marquis claims this flaw enabled hackers to access customer firewall configuration backup files “without appropriate authentication” by predicting firewall serial numbers.

“Although we swiftly managed to secure our network and client data, our investigation indicated that our exposure to threat actors resulted from SonicWall’s network breach and its failure to inform us that our firewall defenses were possibly compromised,” stated Mirchandani, the CEO of Marquis, in a statement provided to TechCrunch.

Mirchandani mentioned to TechCrunch that SonicWall has yet to supply any non-public insights regarding the root cause of its breach. 

“We anticipate learning more through the legal proceedings,” remarked Mirchandani.

Marquis has not disclosed the number of individuals impacted by its data breach. A record with Texas’ attorney general shows that at least 400,000 individuals across the U.S. are confirmed to be affected by the fintech giant’s breach. 

The total number of affected individuals is expected to grow as additional data breach notifications are submitted to various U.S. attorneys general.

When the edtech firm Preply reached unicorn status earlier this year, its Kyiv team marked the occasion with cake — as is customary, even amid a conflict.

Over the four years following Russia’s large-scale invasion of Ukraine, Ukrainian startups have not only managed to endure: they continue to innovate and expand. Preply, for example, plans to use funds from its latest investment round to recruit around 100 engineers globally — including in Ukraine, where one-third of its engineering workforce is located.

Preply stands as one example among numerous cases. The defense technology sector has drawn significant attention, particularly for the rapid pace at which new innovations are deployed in combat. Yet, the same engineering expertise and creativity are proving to be vital across various fields also aiding in fortifying Ukraine.

Facing a significantly larger aggressor, Ukraine adopts a serious stance towards attempts to evade conscription. Nonetheless, startups may receive special status that protects essential personnel from being drafted if they are recognized as bolstering the nation — and Aspichi is among them, as its founder, Victor Samoilenko, explained to TechCrunch.

Initially established in the U.S. in 2021, the company underwent a total transformation with the onset of the war. It is now primarily known for Luminify, a mixed-reality platform dedicated to mental health care that assists Ukrainians in dealing with trauma from the conflict. The startup works closely with military units as well as several clinics that offer mental health services to the community.

This includes soldiers and veterans, grieving families, and millions who have relocated to western Ukraine or overseas. For those who have always resided in Kyiv, the psychological impact is equally severe.

“Everyone is enduring hardship,” Samoilenko remarked. “My daughter has celebrated several New Year’s and Christmas underground; thus, the emotional fallout is immense.”

During a winter characterized by almost daily strikes on the power grid, Ukraine’s capital has not been spared. Natali Trubnikova, CMO at the Kyiv-based IT consulting firm Gart Solutions, noted that residents have adapted to outages with high-capacity power banks, gasoline cookers, and diesel generators. However, increasing costs necessitate that these resources are used judiciously, often resulting in indoor temperatures being kept at a bare minimum.

These severe conditions have transformed offices into shelters from the cold.

“Our office is equipped with various generators to ensure we have electricity, internet, and warmth, and it remains open 24/7 for any Ukrainian team member to utilize at any time,” Preply CEO Kirill Bigai shared with TechCrunch last month.

For smaller enterprises, coworking spaces have also evolved into sanctuaries. LIFT99 Kyiv Hub, a six-year-old facility that was damaged by a Russian attack last August, has experienced a surge in membership since reopening two months ago, according to sales and partnership manager Lada Samarska’s LinkedIn update.

Despite missile assaults, Kyiv remains distant enough from the frontlines to retain its status as Ukraine’s primary startup hub, though it is not alone in this regard. Lviv, the largest city in western Ukraine, has attracted a considerable number of displaced Ukrainians, including tech professionals. LEM Station, a revamped tram depot transformed into a creative hub, symbolizes its expanding ecosystem, which has also gained from its closeness to the Polish border.

Although the number of international visitors has decreased, Lviv is still accessible by train, and the IT Arena 2025 tech conference welcomed 6,450 participants from over 40 countries. While a segment focused on defense tech, the event also highlighted support for a broader spectrum of startups. Despite ongoing hostilities, VC firms remain active in the nation, including 1991, Flyer One Ventures, and SMRK.

Regardless of the challenges, Ukrainians continue to attend tech conferences abroad. At Techarena in Stockholm earlier this month, members of a Ukrainian delegation expressed fatigue and anticipation for the spring. Yet before parting ways, they began to enumerate Ukraine’s unicorns — a fitting response, even in the face of war.

Meta intends to acquire potentially as much as $100 billion in AMD chips, sufficient to power nearly six gigawatts of data center energy demand, the companies revealed on Tuesday.

As part of the long-term agreement, AMD has granted Meta a performance-linked warrant for up to 160 million shares of AMD common stock — equivalent to around 10% of the company — at a price of $0.01 each, designed to vest in accordance with specified milestones. The complete stock award is contingent upon AMD’s share value, which must reach $600 for Meta to obtain its final portion, according to The Wall Street Journal. AMD’s stock price closed at $196.60 on Monday.

In line with the agreement, Meta will acquire AMD’s MI540 series GPUs and its most recent generation of CPUs. CPUs are increasingly recognized as a key component of the AI inference compute stack due to their efficiency, scalability, and the fact that they do not solely bind companies to Nvidia.

“The CPU market is truly booming,” AMD CEO Lisa Su remarked on Tuesday morning during a briefing for investors. “Demand is substantial. It continues to expand, driven by the deployments of AI infrastructure as inferencing intensifies and agentic AI develops, and our portfolio is exceptionally well-positioned.”

AMD has been progressively making headway as AI companies seek to lessen their dependency on Nvidia, the long-time frontrunner in AI chips that has commanded a premium for its position. Last October, AMD and OpenAI forged a similar agreement that involved equity in exchange for a chip purchase contract.

Meta CEO Mark Zuckerberg stated that the collaboration with AMD is “a crucial advancement” as it diversifies its computing capabilities and progresses toward “personal superintelligence.” Zuckerberg has characterized personal superintelligence as AI systems intended to profoundly understand and enhance individuals’ daily lives.

Meta has committed to investing a minimum of $600 billion in U.S. data centers and AI infrastructure over the upcoming years, including an estimated capital expenditure of $135 billion in 2026. Meta recently announced plans for a $10 billion gas-powered data center campus in Indiana aimed at providing 1 gigawatt of compute power.

The AMD collaboration follows a recent agreement in which Meta expanded its data centers with millions of Nvidia’s latest CPUs and GPUs. The company formerly known as Facebook is also developing its in-house chips but has reportedly encountered delays.

This article has been revised with additional information from AMD CEO Lisa Su.