Nvidia news — AI chips, data-center demand, earnings, and export policy — covered without spin.
tech
5
9h ago
4 min read
Nvidia CEO Jensen Huang posted on X Sunday that 'AGI has arrived,' crediting OpenAI's new Astra model and his own chips for the milestone. It's the third time in six months Huang has made some version of this claim, and OpenAI's own leadership can't agree on what the term even means.
tech
6
13h ago
4 min read
Eight months after signing its $20 billion Groq licensing deal on Christmas Eve 2025, Nvidia confirmed on August 24, 2026 that the resulting Groq 3 LPX inference racks are in full production and will go live at cloud provider Nebius before year's end. The hardware promises a 35x efficiency edge over conventional setups, but the deal's unusual structure, a license and hiring spree rather than an outright buyout, still leaves investors guessing at what Nvidia actually bought.
tech
7
23h ago
4 min read
Aivres Systems, once known as Inspur Systems before its Chinese parent landed on the U.S. Entity List in 2023, sold roughly 2,300 of Nvidia's most advanced GB200 chips to an Indonesian telecom carrier last November, according to Crypto Briefing and KuCoin. Aivres itself has never been individually blacklisted, and separately, Hewlett Packard Enterprise has sued the parent company and its subsidiaries over patent infringement.
tech
6
1d ago
3 min read
Nscale is telling prospective IPO investors it has $103 billion in total contracted revenue, double what it claimed months ago, largely on the strength of a $45 billion Anthropic deal. That number sounds massive, but it's a multi-year backlog, not annual revenue, and a source close to the deal admits the figures are illustrative, not guidance.
tech
7
1d ago
4 min read
DeepSeek has ordered 160,000 Huawei Ascend 950DT chips worth roughly $2.64 billion to run its AI models out of a new data center in Ulanqab, Inner Mongolia, according to Bloomberg. The chips handle inference only, meaning every query a user sends the model gets processed on Chinese soil, while DeepSeek still relies on Nvidia hardware to train its models. That split shows both the limits of U.S. export controls and a real question for anyone routing data through DeepSeek's API.
defense
7
2d ago
4 min read
The Wall Street Journal reported September 4 that U.S. negotiators dangled access to Nvidia's Blackwell AI chips for an Armenian data center to help push Armenia toward the August 2025 peace declaration with Azerbaijan. It worked, and it's paying off in Trump-branded infrastructure and investment deals across the South Caucasus. Whether that counts as smart dealmaking or uncomfortable leverage depends on who you ask.
finance
8
2d ago
4 min read
Nvidia's equity portfolio ballooned from $2.2 billion to $99 billion in two years, with nearly $50 billion parked in the same AI labs and cloud companies that buy its chips. CFO Colette Kress calls it fueling a growth 'flywheel.' Michael Burry calls it 'overreaching.' Both can be true, and neither has been tested by a downturn yet.
tech
7
4d ago
4 min read
Nvidia CEO Jensen Huang, OpenAI's Sam Altman and Anthropic's Tom Brown pitched G20 ministers in Chapel Hill, North Carolina on treating AI compute as national infrastructure, while Elon Musk projected AI could add $30 trillion a year to the global economy and warned of an imminent power shortfall. Nvidia is lining up $500 billion in financing for the buildout even as critics call the arrangement circular, and back home Trump's 'backwards and poor' jab at data-center opponents has Republicans and Democrats alike pushing back.
finance
8
4d ago
4 min read
Since its August 26 earnings report showed revenue more than doubling to $96.2 billion, Nvidia has piled on a 70% growth forecast for fiscal 2028, disclosed a 122.8 million-share stake in SpaceX, and reportedly agreed to buy Hugging Face for $12.9 billion. Wall Street is split: some analysts call the guidance shockingly bullish, others call the accompanying $500 billion financing push bananas.
tech
6
5d ago
3 min read
The EU's High-Performance Computing Joint Undertaking signed a €387.8 million ($451 million) contract with French firm Bull to build LUMI-AI in Finland, running on AMD's Instinct MI430X chips instead of NVIDIA hardware. It's a real break from NVIDIA's near-total grip on Europe's AI infrastructure, but calling it 'sovereign' compute is a stretch when the chips still come from an American company.
tech
8
5d ago
4 min read
Nvidia is putting $1.5 billion into SB Energy and guaranteeing up to $105 billion in lease financing for the Ohio AI campus that will run exclusively on Nvidia chips. Draft IPO filings show SB Energy's losses widened to $3.2 billion in the first half of 2026, and its own paperwork flags heavy reliance on OpenAI as a risk, even as Piketon residents and environmental groups weigh in on a project built next to a former uranium enrichment site.
tech
8
6d ago
4 min read
Anthropic locked in a $35 billion cloud deal with Nvidia-backed Lambda for a Texas data center that Nvidia itself holds the lease on, according to Bloomberg and The Wall Street Journal. That brings Anthropic's compute commitments this year past $250 billion, a debt-and-lease-fueled buildout that lets a company without an investment-grade credit rating secure capacity it couldn't get through normal financing.
finance
7
6d ago
4 min read
Nvidia said Monday it bought $3.5 billion of a $3.9 billion MediaTek convertible bond offering, with Alphabet chipping in an undisclosed amount, tying Nvidia's returns to MediaTek's adoption of its NVLink Fusion platform. It's the third major Nvidia financing deal of 2026, and Wall Street analysts and short-sellers are asking whether the AI boom's cash is just circling back on itself.
finance
7
Aug 31
4 min read
SEC filings show Big Tech's blowout Q2 2026 profits were juiced by more than $160 billion in unrealized markups on private AI stakes, led by SpaceX's June IPO and Anthropic's rising valuation. Chipmakers, meanwhile, are the ones actually posting real stock gains this year while Alphabet, Amazon and Microsoft shares lag behind.
tech
5
Aug 30
3 min read
Nvidia's new Arm-based RTX Spark superchip, shown at Computex Taipei, runs 120-billion-parameter AI models locally on laptops costing $3,000 to $4,000, directly targeting the on-device AI market Apple's M-series chips have owned for years. Nvidia hasn't named Apple as its rival, but the specs and pricing tell a different story.
finance
6
Aug 30
5 min read
Caterpillar just posted its first-ever $20 billion-plus quarter, powered by data center demand for generators and turbines, not bulldozers. The stock is up nearly 90% in a year and now trades at a richer multiple than Nvidia, Microsoft or Alphabet, which raises a fair question: is a company that still makes diesel engines really a tech stock now, or is this priced for perfection.
tech
8
Aug 30
4 min read
NVIDIA posted the largest quarterly revenue in semiconductor history on August 26 and delivered its first Vera CPU server to Amazon Web Services the next day. Chip stocks like Micron and Marvell are up 220% and 185% this year while the Magnificent Seven ETF has managed just 4%, and that concentration is now a risk analysts are flagging openly.
tech
7
Aug 29
4 min read
Nvidia's fiscal Q2 earnings, reported Wednesday, August 26, 2026, capped a week where the chipmaker backed a $500 billion GPU financing pact, put up to $105 billion behind an OpenAI data center in Ohio, and opened talks to buy Hugging Face for more than $13 billion. The GPU is still the product, but the moat now runs through orchestration hardware and Nvidia's own balance sheet. Whether that's smart capital deployment or Nvidia financing its own demand is a fair question nobody has settled yet.
tech
8
Aug 29
5 min read
Nvidia has reportedly struck a $12.9 billion deal for Hugging Face, the go-to hub for open-weight AI models, per The Information. It follows Nvidia's $6 billion Poolside deal and Stripe's $7 billion OpenRouter buy, all within weeks. Chinese labs like Alibaba, Moonshot AI and DeepSeek are dominating open-model downloads, and American companies are scrambling to buy their way back into the game rather than build it themselves.
tech
5
Aug 28
4 min read
Nvidia VP Bryan Catanzaro says the company has compressed its open-weight Nemotron model release cycle from six to eight months down to four to six weeks, thanks to synthetic data generation and multi-teacher distillation. The latest release, Nemotron 3.5 Lightning, shipped August 11, 2026. Nvidia gives the models away for free because every download drives demand for the chips those models run on.
energy
8
Aug 28
4 min read
Nvidia reported $96.2 billion in quarterly revenue and $89 billion in data center sales, up 117% year over year, while CEO Jensen Huang acknowledged the industry needs to do better winning over communities worried about power costs and grid strain. The numbers are real. So are the tradeoffs, and the data shows some towns are cashing in while others are getting stuck with the bill.
tech
8
Aug 27
4 min read
OpenAI's first in-house inference chip, Jalapeño, beat Nvidia's GB200 and GB300 on independently verified efficiency benchmarks by up to 1.9x. A day later Nvidia reported $96.2 billion in quarterly revenue, up 106% year over year, and CEO Jensen Huang told investors OpenAI is still committing to roughly 12 gigawatts of Nvidia compute through 2030. Both things happened. Neither cancels out the other.
tech
7
Aug 27
4 min read
Nvidia is rolling out optimizations for Chinese open models like DeepSeek V4 Flash and Alibaba's Qwen 3.8, betting that keeping those models fast on American hardware keeps developers inside the US tech stack. At the same time, Nvidia told the SEC that potential White House restrictions on supporting Chinese open-source models could hit its bottom line. It's a company trying to have it both ways, and Washington hasn't decided which way it's going yet.
tech
7
Aug 27
4 min read
A new Rowhammer attack called GPUThor defeats the ECC protection NVIDIA has been telling customers to enable, hitting four Ampere workstation GPUs used across cloud and AI infrastructure. The exploit code goes public November 15, 2026, giving cloud operators a shrinking window to patch or mitigate before it's out in the wild.
finance
7
Aug 27
3 min read
Since Wednesday's earnings call, Nvidia shares jumped as much as 7.3% in after-hours and pre-market trading, putting the company on track to add roughly $370 billion in market value in a single day on top of its already record-setting numbers. The move is pure market reaction to Jensen Huang's guidance and the newly reported Hugging Face deal, not a new earnings event, and it's happening while gold, bitcoin and so-called dollar debasement trades all rally at the same time.
tech
8
Aug 27
4 min read
Nvidia has agreed to acquire open-source AI hub Hugging Face for $12.9 billion, according to The Information. The deal, still unconfirmed by either company, would hand the world's dominant AI chipmaker control of the developer platform that decides which open models get distributed and how.
tech
8
Aug 27
4 min read
A day after Nvidia's blowout $96.2 billion quarter, analysts on the earnings call spent four separate questions pressing CEO Jensen Huang on how he can guide a full year out for the first time ever. The real number nobody's talking about enough: five customers now account for 70% of Nvidia's accounts receivable, up from 56% a year ago.
tech
8
Aug 27
4 min read
Nvidia beat Wall Street estimates on Wednesday and stunned analysts with guidance for 70% revenue growth in fiscal 2028, well above the 44-45% consensus, sending shares up more than 5% after hours. The catch: memory chip shortages are dragging gross margins down to 71-72% by year end, and the company's own numbers confirm it's still leaning heavily on a shrinking circle of customers who are racing to build chips that compete with it.
economy
8
Aug 27
4 min read
Amazon and Nvidia announced Wednesday a deal to add 2 million more Nvidia GPUs to AWS data centers by 2028, on top of the 1 million already committed five months ago. Nvidia posted $96.2 billion in quarterly revenue, up 106% year over year, proving the AI buildout still has real money behind it, not just hype.
tech
8
Aug 26
4 min read
Nvidia reported fiscal Q2 revenue of $96.2 billion, up 106% year over year, and told investors it now has $279 billion locked into supply and procurement deals, more than double last quarter's $119 billion. CFO Colette Kress said memory costs are running hotter than expected and will keep constraining supply through at least fiscal 2028. The stock rose after hours, but the SEC filing shows a chunk of those "commitments" are cancelable or conditional, not locked-in cash.
tech
6
Aug 26
4 min read
Nvidia used the Hot Chips 2026 conference to detail its Vera CPU, a custom 88-core chip built specifically for AI agents, claiming it beats AMD's EPYC 9655P by up to 22% on Linux kernel compilation. The company also announced a partnership with SpaceXAI to power Grok workloads and paired the launch with an Indian data center order for 9,000 Vera Rubin chips. The benchmarks are Nvidia's own, so independent verification is the next thing to watch.
tech
7
Aug 25
4 min read
OpenAI unveiled benchmark results Tuesday claiming its custom Jalapeño chip beats Nvidia's Blackwell systems on speed and power efficiency for AI inference. The chip is not shipping in any real volume until late 2026 at the earliest, and OpenAI itself says Nvidia remains a core partner, so treat the victory lap with appropriate skepticism.
tech
6
Aug 25
4 min read
Perplexity rolled out Portable Computer, a version of its AI agent that runs entirely on Nvidia hardware with zero cloud token costs. The launch lands the same week reports surfaced that Nvidia is negotiating a fresh equity stake in Perplexity that could value the startup above $30 billion, more than 50% higher than its valuation a year ago.
tech
8
Aug 25
4 min read
Nvidia reports second-quarter earnings after the bell Wednesday, with analysts expecting revenue to nearly double to $92.18 billion. But the real story is what Nvidia has become on the side: a $500 billion financier of its own customers, including a $105 billion Ohio data-center backstop for OpenAI, and Wall Street isn't sure if that's genius or a house of cards.
finance
6
Aug 25
4 min read
Nvidia reports fiscal Q2 earnings Wednesday, August 26, after the close, with options markets pricing a 5.4% move worth roughly $280 billion in either direction. The stock fell for seven straight sessions into the print, but that's noise, not signal, and the real number to watch is next quarter's revenue guidance, not the beat everyone already expects.
tech
7
Aug 24
4 min read
Nvidia announced Monday its Groq 3 LPX inference racks have entered full production, headed to cloud provider Nebius later this year. The stock still fell after the news, part of a losing streak heading into Wednesday's earnings, as investors weigh chip pricing and Nvidia's ballooning web of infrastructure investments against the hype.
tech
7
Aug 24
3 min read
Keelung prosecutors indicted nine people, including a Nvidia Taiwan manager and two Super Micro Taiwan employees, for allegedly forging end-user documents to route 130 Nvidia B300-equipped servers around US export controls. Seventy-four servers made it to China through Indonesia, Japan and Hong Kong before customs caught the rest. Illegally exporting these chips isn't even a crime under Taiwanese law, so prosecutors had to charge breach of trust and forgery instead.
defense
7
Aug 24
4 min read
Ukrainian military intelligence says it pulled an Nvidia Jetson Orin NX module, packaged in March 2025, from the wreckage of a Russian S-71M cruise missile recovered in August 2026. Nvidia says it doesn't sell to Russia and left the market in 2022, meaning the chip almost certainly moved through a resale or transshipment network sanctions haven't closed.
tech
6
Aug 24
4 min read
Nvidia is discussing an equity investment in AI search startup Perplexity that would value it above $30 billion, according to The Information. That's more than 50% above last year's valuation, and Perplexity's revenue has tripled to over $750 million annualized. Nobody's confirmed a signed deal yet, but the numbers show real revenue growth behind the hype for once.
energy
8
Aug 23
4 min read
Nvidia signed deals with six Wall Street giants to mobilize $500 billion in financing so customers can buy GPUs without straining their own balance sheets. Meanwhile Vistra, a merchant power producer with 44,000 megawatts of capacity, is quietly profiting off the same AI boom by selling electricity straight into wholesale markets instead of taking government-capped utility returns.
finance
7
Aug 23
4 min read
Wall Street took a beating last week as chip stocks sold off and bond yields climbed, while U.S.-Canada trade talks collapsed and new 50% tariffs on $20 billion of Canadian goods took effect Saturday. Markets now turn to Nvidia's earnings and Fed Chairman Kevin Warsh's first Jackson Hole speech, both landing this week.
tech
6
Aug 23
4 min read
AMD announced on May 21, 2026 a multi-year investment topping $10 billion in Taiwan's semiconductor supply chain to secure capacity for a new packaging technology called EFB. The company says the technology could help it avoid the bottleneck constraining AI chip production, but whether the bet pays off depends on converting announced deployments from OpenAI, Meta, and Anthropic into shipped hardware before 2029.
tech
8
Aug 22
3 min read
Nvidia has notified its biggest customers that servers running Vera Rubin and Grace Blackwell chips will cost more than 15% more starting with systems shipped in early 2026, according to Bloomberg News. Blame the memory chip shortage. Everyone building AI data centers, and eventually everyone paying for AI services, gets to eat the bill.
finance
6
Aug 22
4 min read
Nvidia reports fiscal Q2 results Wednesday, August 26, hours before the Fed's preferred inflation gauge drops and a day before Fed Chair Kevin Warsh speaks at Jackson Hole. Bond yields are already at multi-decade highs, national debt just crossed $40 trillion, and the market has spent the week bracing for all three to hit at once.
tech
6
Aug 21
4 min read
Nvidia researchers found that a custom software harness took Claude Opus 5 from a 30% score to a 100% score on a tough AI reasoning benchmark, without changing the underlying model at all. The result backs up a broader trend from OpenAI, DeepSeek, and independent researchers: the scaffolding around an AI model, not the model itself, is what actually makes agents work. That's good news for competition and bad news for the idea that whoever builds the biggest model wins.
tech
5
Aug 21
3 min read
Nvidia researchers built a technique that lets AI systems hand off tasks between different-sized models without recomputing the entire conversation from scratch. It runs up to 25 times faster than the current brute-force method while keeping 98% of accuracy, using basic linear math instead of another expensive neural network.
tech
5
Aug 21
4 min read
The Information reported, citing two Nvidia employees, that Nvidia planned small-batch shipments of a China-tailored AI inference chip by year-end. Nvidia flatly denied it, saying no such product exists on its roadmap. The dispute lands a week before Nvidia's August 26 earnings report, with the company already booking zero expected China data-center revenue in its own guidance.
tech
6
Aug 21
4 min read
Chicago journalists, podcasters and voice actors filed nine lawsuits in mid-May 2026 accusing nine tech giants of scraping their voice recordings to train AI models without permission, using Illinois' biometric privacy law. The companies have moved to dismiss, arguing no concrete harm was shown, while the plaintiffs' firm calls it one of the biggest biometric privacy violations ever attempted.
tech
6
Aug 21
4 min read
Brazil announced 2.3 billion reais ($444.2 million) in AI infrastructure spending, handing over half to a China-linked supercomputing project with Huawei and iFlytek, and putting the rest toward a tender Nvidia is expected to win. Lula's government calls it strategic autonomy. It's really a country trying to have it both ways with two superpowers, and taxpayers are footing the phased bill either way.
tech
7
Aug 21
3 min read
Nvidia is paying $6 billion to license AI coding models from Poolside AI and putting in another $1 billion at a $12 billion valuation, according to a letter to investors obtained by Newcomer. 109 Poolside employees have received job offers from Nvidia, the latest example of Big Tech buying its way around a formal acquisition.
tech
6
Aug 20
3 min read
Hudson River Trading inked a multiyear, multibillion-dollar deal with CoreWeave on Thursday, becoming one of the first customers with wide-scale access to Nvidia's new Vera Rubin chips. Wall Street quant shops are now buying AI infrastructure like frontier AI labs, and CoreWeave is cashing in as it tries to shake its dependence on Microsoft.
tech
6
Aug 19
4 min read
Nvidia CEO Jensen Huang told CNBC nobody deploys AI better than Meta, a comment that landed right after Meta shares fell 7% on worries about its $125 billion to $145 billion 2026 AI spending plan. Huang runs the company that sells Meta the chips, so the praise isn't exactly disinterested, but Meta's ad-targeting and engagement numbers back up at least part of the claim.
tech
6
Aug 19
4 min read
Nvidia is quietly connecting companies that want GPU computing power with Nordic data center operators that have the land and electricity to house it, according to sources who spoke to CNBC. It's a smart monopoly play dressed up as customer service, and it's working because Europe's power grid near major cities is maxed out.
tech
8
Aug 19
4 min read
The U.S. bans Nvidia's top AI chips from going to China directly, but Chinese companies are reportedly accessing that same computing power remotely through data centers in Thailand, Malaysia and Japan. Separately, Nvidia's less-powerful H200 chips are now shipping into China after Beijing spent months blocking approved sales. Both stories show export controls are leakier and messier than the policy on paper suggests.
tech
7
Aug 19
4 min read
Beijing has started letting Nvidia's H200 AI chips trickle into China, with ByteDance and Tencent each receiving about 10,000 units, according to the Financial Times. Washington already cleared these companies to buy up to 100,000 chips apiece, so the real bottleneck has been Chinese approval, not American export policy. Beijing is steering much of the hardware to Hong Kong to protect Huawei and other domestic chipmakers from full-blown Nvidia dependence.
tech
8
Aug 17
4 min read
Nvidia disclosed a securities filing guaranteeing up to $105 billion in lease obligations for a massive OpenAI data center in Pike County, Ohio, on top of a $1.5 billion equity stake in developer SB Energy. The chipmaker is betting big that OpenAI can pay its bills for the next 20 years, and if it can't, Nvidia eats the shortfall.
finance
8
Aug 17
5 min read
Nvidia lined up half a trillion dollars from Wall Street to turn GPUs into loanable assets, betting they hold value like real estate. Meanwhile Chinese hedge funds managing over $1.7 billion combined are dumping Nvidia and hyperscaler stock, calling the whole AI trade a super bubble. Both bets can't be right, and nobody's balance sheet has decades of data to prove which one is.
tech
7
Aug 16
4 min read
Since September 2025, Beijing's Cyberspace Administration has barred major Chinese tech firms from buying Nvidia AI chips, pushing them toward Huawei's Ascend processors instead. Nearly a year later, sources at top Chinese AI labs say Nvidia chips remain the default for training their most advanced models, because switching software ecosystems is more expensive than switching hardware. Nvidia has reported zero revenue from H200 chip sales to China even after Washington cleared the chip for export, according to Crypto Briefing, showing the block is coming from Beijing, not the U.S.
tech
7
Aug 15
4 min read
Nvidia is negotiating a $3 billion investment in SoftBank's SB Energy alongside roughly $100 billion in credit support for a massive Ohio data center that would house OpenAI's computing needs, according to The Information. The numbers have already shifted once, down from an earlier $250 billion guarantee figure reported by the Wall Street Journal, which tells you these deals are still being hammered out and nothing is signed.
finance
8
Aug 15
4 min read
Since Apollo warned this week that the AI buildout needs $2 trillion in new debt, bond traders have zeroed in on a specific number: roughly $70 billion in "phantom" liabilities that Nvidia and other AI giants keep off their balance sheets through residual value guarantees. Nvidia's credit default swap spread has more than doubled since late May, and its stock has been whipsawing on the same worries this outlet flagged Thursday about a $2 trillion debt gap.
Showing the 60 most recent of 178. Browse the full archive →
