DD
MM
YYYY

PAGES

DD
MM
YYYY

spot_img

PAGES

Home Blog Page 4

Groq Raises $350m at $3.5bn Valuation as AI Chip Startup Pivots to Nvidia-Powered Cloud

0

Groq has raised $350 million in new funding at a $3.5 billion valuation as the artificial intelligence startup accelerates its transformation from an AI chip developer into a cloud and data-center provider built around Nvidia’s computing hardware.

The funding round was led by investment firm Disruptive, with planned participation from Nvidia, according to the company. The new valuation is roughly half the $6.9 billion valuation Groq reached in September, before Nvidia hired Groq founder and CEO Jonathan Ross and other senior employees as part of a licensing agreement.

Groq said the lower valuation should not be viewed as a conventional down round. A company spokesperson told TechCrunch that the financing establishes a new valuation for the “post-Nvidia-licensing-deal version of Groq.”

The change marks a major strategic shift for Groq.

The company originally sought to compete with Nvidia by developing its own AI processors, known as language processing units, or LPUs. The chips were designed primarily for inference, the computing process involved in running trained AI models and generating responses in real time.

But Nvidia’s recruitment of Ross and other senior Groq employees fundamentally altered the company’s trajectory. After losing much of its key leadership and technical talent, Groq moved away from being a standalone chipmaker and began building an AI cloud business using Nvidia’s GPUs.

The result is an unusual relationship in which a company that once sought to challenge Nvidia now operates as one of its customers.

Groq raised $650 million in June to begin financing the transition. The company plans to expand its data-center capacity from 54 megawatts to more than 200 megawatts by 2027. It currently operates 13 data centers across North America, Europe, the Middle East and Asia-Pacific and says its infrastructure serves more than 6 million developers, enterprises and AI-focused companies.

The latest financing will be used to expand access to Nvidia accelerated computing for customers requiring medium and large clusters for AI training and inference.

“We are building Groq into the world’s leading AI inference cloud,” said Alex Davis, Groq’s chairman and CEO of Disruptive. “Inference will without a doubt become the largest and most critical layer of AI infrastructure.”

The rapid adoption of generative AI has created enormous demand not only for chips but also for the data centers, electricity, networking equipment and cloud services needed to operate them. Companies known as neoclouds have emerged to provide specialized access to high-performance computing without requiring customers to build their own large-scale infrastructure.

Inference is becoming particularly important as companies move AI systems from experimentation into everyday applications. Every chatbot response, AI-generated image, coding task, and automated workflow requires computing resources when the model is being used.

That creates a potentially enormous market for specialized infrastructure providers.

But the business model comes with significant financial risks.

Neocloud companies must spend heavily on GPUs and data-center capacity before generating revenue from those assets. The hardware can also depreciate rapidly as newer generations of processors become available, creating pressure to maintain high utilization rates and secure long-term customer commitments.

CoreWeave illustrates both the opportunity and the risks.

The AI cloud provider has reported strong revenue growth and secured major contracts with companies including Meta and Anthropic. Investors, however, have remained concerned about its substantial capital expenditures, dependence on debt financing, and exposure to rapidly depreciating computing hardware.

The central question is whether strong demand for AI computing will generate enough cash flow to justify the enormous investments required to build and maintain the infrastructure.

Groq’s financial performance remains private, making it difficult for investors to assess the economics of its new strategy. Its rapid increase in data-center capacity, however, shows that the company is betting heavily on sustained demand for AI inference.

The financing also places Groq squarely within Nvidia’s expanding infrastructure ecosystem.

Nvidia is increasingly doing more than selling GPUs. The chipmaker has invested in several companies building AI cloud capacity, while those same companies purchase Nvidia’s processors to operate their infrastructure.

CoreWeave, Lambda and Nebius are among the neocloud providers using Nvidia GPUs, creating a business model in which Nvidia can benefit both from supplying the hardware and, in some cases, investing in the companies purchasing it.

Groq now occupies a similar position.

Its original ambition was to compete with Nvidia at the chip level. Its new strategy instead depends on Nvidia’s dominance in AI accelerators. That does not necessarily make Groq’s business less ambitious. It only changes where the company is competing. Rather than trying to build an alternative to Nvidia’s hardware ecosystem, Groq is seeking to compete for customers who need access to that hardware, particularly customers focused on AI inference.

The shift also demonstrates how difficult it has become for smaller AI hardware companies to compete independently as Nvidia’s ecosystem expands. Building competitive processors requires enormous investments in semiconductor design, software, and manufacturing, while customers increasingly value compatibility with established AI development platforms.

For Groq, becoming an Nvidia-powered AI cloud could provide a faster route to scale than continuing to develop its own chips.

The $3.5 billion valuation, however, shows that investors are also assigning a different value to the company than they did before the Nvidia licensing deal. The challenge for Groq will be proving that its new business can generate sufficient revenue and margins to justify the capital required to expand its infrastructure.

The company is effectively betting that AI inference will become one of the largest computing markets in the technology industry. If demand continues to accelerate, Groq is expected to benefit from its existing developer base, global data-center footprint, and relationship with Nvidia. However, neocloud providers could face intense pricing pressure and weaker returns on expensive computing assets if infrastructure supply grows faster than customer demand.

XRP Whales Go Quiet as Grayscale Pulls Three Altcoin ETF Filings

0

The cryptocurrency market is entering another period of uncertainty as two developments highlight the changing appetite for digital assets: XRP whale inflows to Binance have fallen to their lowest level since 2021.

Grayscale has withdrawn exchange-traded fund registrations for Cardano, Polkadot and Hedera. Together, the developments point to a market becoming increasingly selective about where capital is deployed.

Recent on-chain data shows that the three-month average of XRP whale deposits to Binance has dropped to approximately $61 million, a level not seen since 2021. The figure represents a substantial decline from more than $450 million recorded last year.

While falling whale deposits can initially appear bearish because large holders are becoming less active, the data can also indicate reduced selling pressure because fewer XRP tokens are being transferred toward an exchange.

That distinction is important. Exchange inflows from large holders are frequently monitored because they can precede selling activity. If whales move significant amounts of XRP onto exchanges, traders may interpret it as preparation for distribution.

The current decline therefore does not necessarily mean that large investors have abandoned XRP. Instead, it suggests that whales may be waiting for clearer market conditions before making major moves.

Some evidence points toward continued accumulation. Recent reports indicate that large holders accumulated more than 72 million XRP in a single day.

While the number of wallets holding at least one million XRP increased over a three-month period. This creates an interesting divergence: exchange inflows are falling, but certain measures of large-holder accumulation remain constructive.

The second development involves Grayscale’s decision to withdraw three altcoin ETF registrations. The asset manager withdrew filings associated with Cardano, Polkadot and Hedera on August 7, reportedly within minutes of one another.

The withdrawals came shortly before Cardano became eligible under the relevant SEC seasoning framework, adding significance to the timing. Grayscale’s decision does not necessarily represent a rejection of those cryptocurrencies.

Instead, it may reflect the difficult economics and uncertain demand surrounding smaller altcoin exchange-traded products. Bitcoin and Ethereum have established deep institutional markets.

While the investment case for smaller tokens depends heavily on liquidity, investor demand, regulatory conditions and the ability to attract sufficient assets under management.

The development demonstrates that regulatory progress alone does not guarantee an ETF launch. Even when the regulatory pathway becomes more accommodating, issuers still have to determine whether a product can achieve sustainable commercial scale.

For XRP, the situation is particularly notable because the token has continued to attract institutional attention through its own ETF ecosystem. Recent reporting indicates that XRP-related ETFs continued receiving inflows even as Grayscale withdrew the three competing altcoin filings.

Both developments point toward a more selective crypto market. Capital is no longer automatically flowing into every major altcoin simply because regulatory barriers are falling. Investors appear increasingly focused on liquidity, institutional demand and sustainable market structure.

For XRP, the five-year low in whale exchange inflows could reduce immediate selling pressure, but it is not by itself a bullish signal. The next decisive factor will be whether subdued whale activity is followed by renewed demand and stronger price momentum.

Meanwhile, Grayscale’s ETF withdrawals suggest that the next phase of institutional crypto adoption may favor a smaller group of assets capable of demonstrating durable demand rather than simply winning regulatory eligibility.

YouTube to Count Views From the Moment Videos Start, Redefining How Creators Measure Reach

0
A picture shows a You Tube logo on December 4, 2012 during LeWeb Paris 2012 in Saint-Denis near Paris. Le Web is Europe's largest tech conference, bringing together the entrepreneurs, leaders and influencers who shape the future of the internet. AFP PHOTO ERIC PIERMONT (Photo credit should read ERIC PIERMONT/AFP/Getty Images)

YouTube is changing the way it counts video views, allowing a view to be recorded as soon as a video begins playing or a user enters a live broadcast, a move that could significantly increase publicly displayed view counts across the platform.

The Google-owned video platform announced Monday that the new system will take effect on August 24. YouTube said the change is intended to create a more consistent view-counting system across different formats and give creators a clearer measure of their overall exposure.

“Historically, we’ve used multiple view counting systems across different formats,” YouTube said in a post on its Community forum. “However, we’ve heard that creators want to eliminate this metric confusion and accurately understand their true exposure, which is why we’re making this update.”

YouTube has not publicly disclosed the precise threshold it previously used to register a standard video view, although it has generally been understood that a viewer needed to watch at least 30 seconds for a view to be counted.

Under the new system, simply starting playback will count as a view. The same principle will apply when users enter a live broadcast.

The change brings YouTube’s main video platform closer to the way competing short-form platforms such as TikTok and Instagram count views, where playback generally triggers a view.

YouTube introduced a similar system for Shorts last year. Extending the approach to long-form videos and live streams could substantially increase view totals, particularly for content that attracts large numbers of users who leave shortly after playback begins.

That could alter the meaning of one of the most visible metrics on YouTube.

A video that previously recorded a lower number of views because many users abandoned it quickly could now accumulate substantially more views simply because more people started watching. As a result, the public view counter may become less useful as a standalone indicator of audience engagement or content quality.

YouTube is retaining its previous measurement under a new name: “Engaged views.” Creators will continue to see the metric through YouTube Analytics, allowing them to distinguish between people who merely initiated playback and those who continued watching.

The distinction could become important for advertisers and creators assessing the effectiveness of video content. A high view count under the new system will not necessarily mean that viewers spent significant time watching a video. For advertisers, the change could also make comparisons between campaigns more dependent on deeper engagement metrics rather than headline view totals.

YouTube said the new counting system will not affect creator earnings or eligibility for monetization through the YouTube Partner Program. In other words, the change is primarily a measurement adjustment rather than a direct change to how creators are paid.

The timing comes as YouTube is simultaneously tightening other aspects of its creator economy.

The company announced last week that new creators will face higher thresholds to begin earning money from advertising and subscriptions starting next year. Under the new requirements described in the announcement, creators will need at least 8,000 qualified watch hours during the previous 12 months or 20 million qualified Shorts views during the previous 90 days.

The current requirements are 1,000 subscribers plus 4,000 watch hours over 12 months, or 1,000 subscribers plus 10 million Shorts views over 90 days. Creators also need to maintain the required Shorts-view threshold over a 90-day period to earn money.

The proposed increase has generated backlash among users and creators who say that the higher thresholds will make it more difficult for smaller channels to enter and remain in YouTube’s monetization ecosystem.

Together, the changes point to a broader shift in how YouTube is managing its creator platform. The company is making its headline view metric more inclusive while placing greater emphasis on “Engaged views” inside its analytics tools. At the same time, it is raising the bar for creators seeking access to monetization.

The new view system is expected to provide a simpler and more consistent metric across videos, Shorts and live content for YouTube. For creators, however, the immediate consequence may be a widening gap between the number displayed publicly and the amount of attention their content actually receives.

That makes metrics such as watch time, retention, and engaged views more important for evaluating whether a video’s larger view count represents genuine audience interest or simply more people starting playback.

Reddit Tests “Video Reddit” as It Turns Viral Posts Into Audio and Video Content

0

Reddit is testing a new way to turn its most popular conversations into audio and video, as the social media company seeks to expand how users consume its text-heavy content and capture engagement that is increasingly shifting toward short-form and spoken media.

Starting Monday, Reddit will test an initial version of what it calls a “video Reddit” experience across select communities. The experiment will include both narrated video and audio versions of posts as the company evaluates which formats users prefer and whether either can be scaled across the platform.

The initiative was first disclosed during Reddit’s second-quarter earnings call in July, when CEO Steve Huffman told analysts that users were already consuming Reddit content in similar formats on other platforms.

Popular Reddit stories are frequently repackaged on TikTok and Meta’s Reels, where creators use text-to-speech narration or read posts aloud. Those videos are often paired with the original text displayed on screen or unrelated background footage, including video-game playthroughs and cooking clips.

“There is an emerging content type elsewhere on the internet of, basically, podcasts where people read Reddit content,” Huffman said during the earnings call.

“I think this version of, like, listened-to or spoken Reddit can be really engaging, as well,” he added.

Reddit’s move brings that behavior inside its own platform rather than leaving third-party creators and social networks to capture the audience and engagement generated by Reddit’s content. The initial rollout is deliberately limited. Reddit said the tests are intended to help it understand how audio and video formats could be useful to users and whether they can be presented in a way that remains consistent with the platform’s culture.

Users will be able to select “read” or “play” for posts where the feature is available. The experiment will initially cover only selected English-language posts and will be available through Reddit’s iOS and Android applications.

Importantly, the new formats will not replace Reddit’s traditional text-based posts. Users will continue to be able to read the original post and participate in its comments in the usual way. Instead, Reddit is positioning audio and video as additional ways to consume conversations. A user could listen to a narrated post while exercising, walking, or doing errands, while a video version could turn a written discussion into a more conventional audiovisual experience.

Image credit: TikTok screenshot

Reddit Wants to Capture Engagement Happening Elsewhere

The experiment addresses a longstanding problem for Reddit: some of the platform’s most popular content already travels well beyond Reddit. Stories from Reddit communities have become raw material for creators on short-form video platforms, particularly when posts involve unusual personal experiences, relationship disputes, workplace conflicts, mysteries or other highly shareable narratives.

Those creators can package Reddit conversations into videos and capture views, advertising revenue, and audience attention without requiring users to return to Reddit.

By developing its own audio and video formats, Reddit could potentially retain more of that consumption within its ecosystem. The strategy also fits a broader shift in how people consume written information. Users listen to articles, posts, and discussions rather than reading them directly, particularly on mobile devices and while performing other activities.

For Reddit, the challenge is making that transition without stripping away the community interaction that differentiates the platform from conventional content publishers. The company’s decision to retain the original posts and comments suggests it does not intend to turn Reddit into a purely video-driven social network. Instead, it is testing whether audiovisual formats can serve as another entry point into Reddit’s existing communities.

Reddit Has Experimented With Video Before

The company has a history of trying to expand beyond its traditional text-and-image format. Reddit previously introduced native video hosting and experimented with versions of a TikTok-style video feed. More recently, it introduced video in comments, which Reddit says now accounts for more than 10% of video posts on the platform.

The latest initiative represents a somewhat different approach because the company is using Reddit’s existing text-based discussions as the foundation for audiovisual content. That could allow Reddit to create a large amount of video and audio programming without depending entirely on users to produce original video.

It also gives the company a way to test which types of Reddit conversations translate effectively into spoken or visual formats. Not every Reddit discussion is likely to work as a narrated story, and the company’s early tests could help determine whether users prefer certain communities, topics, or post structures.

Copyright And Authenticity Remain Important Questions

The experiment also raises questions about how Reddit handles the transformation of user-generated content. Reddit’s value comes largely from conversations created by its users, and converting those conversations into professionally packaged audio or video introduces questions about attribution, consent, and the boundaries between platform distribution and creator ownership.

Reddit is also attempting to solve an authenticity problem. Much of the existing third-party content based on Reddit uses automated narration and unrelated background footage, formats that can make genuine user discussions feel manufactured for engagement.

The company says its early experiments will examine whether it can create audiovisual versions that still feel authentic to Reddit.

That may be critical to the strategy’s success. Reddit’s strongest advantage is not simply the volume of text on its platform but the communities and conversations built around specific interests. If turning those discussions into videos makes them feel detached from their original communities, Reddit could lose some of the characteristics that make the content valuable in the first place.

For now, the company is treating the project as an experiment rather than a major redesign. If users embrace narrated and video versions of Reddit posts, however, the test could give the company a new way to increase time spent on the platform, capture consumption that currently occurs on competing social networks, and extend the reach of its enormous archive of user-generated conversations.

Nvidia to Guarantee Up to $105 Billion for OpenAI Data Center in Ohio

0

Nvidia has agreed to provide a guarantee of up to $105 billion to support OpenAI’s lease of a massive data center campus in Ohio being developed by SoftBank-owned SB Energy, deepening the chipmaker’s role in financing the infrastructure required to sustain the artificial intelligence boom.

Nvidia said Monday it will also invest $1.5 billion in SB Energy, adding to a growing portfolio of infrastructure investments designed to expand the computing capacity available to customers using its chips.

The agreement is seen as the latest indication that Nvidia is moving beyond its traditional role as a supplier of AI processors and increasingly helping finance the data centers, power systems and other infrastructure needed to deploy them. It also raises fresh questions about the financial relationships developing across the AI industry, as chipmakers, cloud providers, model developers and infrastructure companies increasingly invest in one another.

The Ohio project, in Pike County, will have as much as 8 gigawatts of computing capacity, making it one of the largest AI infrastructure developments announced in the United States. The first 800 megawatts is expected to come online in 2028, while OpenAI will lease the facility for 20 years.

Nvidia will be the site’s exclusive chip supplier.

Nvidia CEO Jensen Huang defended the financing arrangement, rejecting suggestions that it represents circular financing in which AI companies effectively use money from their suppliers and investors to purchase the suppliers’ products.

“We are securing long-lived infrastructure for Nvidia compute so OpenAI can deploy the most productive AI factories that can be upgraded repeatedly with each new generation delivering more intelligence and better economics,” Huang said.

Nvidia said its guarantee will cover only a portion of the project’s lease and power payments, as well as a commitment to maintain a minimum value for the site. It will not guarantee the entire cost of construction or all of OpenAI’s obligations.

Under the proposed structure, OpenAI would remain responsible for rent. If OpenAI defaults, Nvidia would cover the difference between the guaranteed minimum value and whatever SB Energy can recover by leasing or selling the facility.

The final financing structure has not yet been determined, according to people familiar with the matter cited by Reuters. The project is expected to combine equity and debt. Equity could include money raised through a potential SB Energy initial public offering as well as direct investment from SoftBank.

Once the equity component is established, the remaining financing could come from project-finance loans and potentially public debt such as bonds, the people said.

Nvidia’s Infrastructure Bet

The deal represents an unusually large commitment by Nvidia to infrastructure that will ultimately create additional demand for its own processors.

Nvidia has pursued this strategy as AI developers race to secure scarce computing capacity. The company is effectively helping ensure that the physical infrastructure needed to deploy successive generations of its chips is built ahead of demand.

Huang said the Ohio facility could generate as much as $200 billion in Nvidia revenue over time. Including a potential 3.75-gigawatt expansion, Nvidia expects to sell OpenAI a total of 16 gigawatts of computing capacity and could generate as much as $600 billion in revenue from OpenAI through 2030.

That creates a powerful commercial incentive for Nvidia to help solve one of the biggest constraints facing the AI industry: the availability of suitable data centers with enough electricity and grid capacity.

Nvidia’s latest commitment follows a separate initiative announced last week in which the company partnered with six major financial institutions, including BlackRock, to establish financing platforms targeting more than $500 billion in third-party capital for AI infrastructure.

The scale of those commitments has intensified scrutiny over whether the industry’s rapid expansion is being supported by complex financial arrangements among companies that ultimately depend on one another.

“Investors are right to be worried about what seems to be a never-ending loop of AI deals but realistically the field of players isn’t all that vast and there was always going to be a degree of circular financing,” said Danni Hewson, head of financial analysis at AJ Bell.

“The biggest test is whether these investments ultimately generate decent returns for all those laying out cash and that’s something that can only be figured out further down the line,” she said.

Power Emerges As The Next AI Bottleneck

The Ohio project also reveals a challenge that is becoming as important to AI development as chips themselves: electricity. The proposed campus will initially have 4.25 gigawatts of capacity, eventually rising to as much as 8 gigawatts. One gigawatt of computing power is roughly equivalent to the electricity consumption of about 750,000 U.S. homes on average.

To support the development, SoftBank and SB Energy plan to build at least 10 gigawatts of new power generation and spend $4.2 billion on regional grid infrastructure through a partnership with AEP Ohio.

The scale of that investment highlights why AI infrastructure is now being planned around access to electricity rather than simply the availability of land. The U.S. power grid is already facing constraints in some regions, while data center projects have encountered opposition from communities concerned about electricity prices, water consumption, and the broader impact of large-scale facilities on local infrastructure.

The Ohio project is expected to create about 35,000 construction jobs through 2032 and roughly 2,500 permanent operating positions, according to OpenAI.

People familiar with the project said the risk of significant construction delays from local opposition is relatively low because Ohio officials have supported the development because of its expected economic benefits. OpenAI and SoftBank have also committed $80 million to community projects.

The project involves federal land and has participation from the U.S. Departments of Commerce and Energy.

A Test of The AI Infrastructure Economy

The Ohio development is part of a broader race to build what the AI industry calls “AI factories”: large computing facilities designed to train and run increasingly sophisticated models.

While securing these sites can help Nvidia create a predictable market for successive generations of its processors, it is expected to grant OpenAI essential long-term access to computing capacity as it scales its models and AI services.

But the economics of the arrangement will ultimately depend on whether demand for AI computing grows quickly enough to justify the enormous capital expenditure required to build and operate these facilities.

The 20-year lease provides OpenAI with long-term access to capacity, while Nvidia’s guarantee shifts part of the financial risk of the project onto the chipmaker. That creates an unusual alignment of interests. Nvidia benefits when OpenAI expands its computing footprint because the company is the exclusive chip supplier to the Ohio site. OpenAI benefits by securing infrastructure for future model development. SoftBank and SB Energy gain a major long-term customer for a large infrastructure project.

The arrangement therefore ties together the fortunes of three major participants in the AI infrastructure buildout.