Susana Schwartz
Did Nvidia's record $57 billion revenue report yesterday burst the AI bubble talk, or is it a temporary euphoria?
Valued at $4.5 trillion, Nvidia is getting big enough that its earnings could determine where the economy goes next. After a long losing streak for indices and a massive selloff last week, there’s suddenly a huge rally in the market based on the eye-popping numbers Nvidia posted yesterday, with $57 billion in Q3 revenue centered largely around its data-center GPUs. Notably, China was excluded in its report because of U.S. export restrictions on advanced AI chips.
Some analysts say the record-breaking numbers are due in part to a larger-than-expected allocation from TSMC, which may be “floating the numbers,” said Seaport Research senior analyst Jay Goldberg during a Bloomberg podcast. Despite some cynicism from the few who still have a “sell” rating on Nvidia, the company beat forecasts about its data center revenues, with Q3 DC revenues of $51.2 billion (with forecasts hovering at about $49.34 billion).
CEO Jensen Huang capped off the announcement by stating Cloud GPUs are “sold out, and compute demand keeps accelerating and compounding across training and inference — each growing exponentially.”
After yesterday’s Q3 report, Coreweave and Nebius gained, as did AMD, Intel, Broadcom, and other semiconductor companies — buoyed either temporarily or longer term by the report. It will be interesting to see how today’s Labor Department jobs report will affect current sentiment about AI infrastructure, as many non-investing people are not as enthusiastic about AI, fearing they will lose their jobs in upcoming months and years, as has started to happen in portions of the tech industry — something I reported in my newsletter a couple of days ago.
Below, see why Hivelocity CEO says bare metal cloud, edge computing, and carrier-neutral connectivity are the drivers to turnkey IaaS solutions for SMBs and enterprises.
Susana Schwartz
Technology Editor
RCRTech
AI Infrastructure Top Stories
Inferencing requires performance of bare metal and flexibility of cloud: Bare metal cloud, edge computing, and carrier-neutral connectivity are the ‘next big thing,’ says Hivelocity CEO Jeremy Pease in an RCRTalks episode with executive editor Kelly Hill that highlights inferencing and ‘getting AI out to end users’ through pre-validated “turnkey” IaaS solutions. By combining the raw performance of bare metal, the flexibility and automation of the cloud, and a carrier-neutral approach, Hivelocity aims to quiet the “noisy neighbor” issues that plague shared virtual machines.
Nokia restructures around ‘network infrastructure’ and ‘mobile infrastructure’: Has Nokia lost its mind? asks RCR Wireless’s James Blackman. Bolstered by Nvidia’s $1 billion investment, Nokia wants to capitalize on the “AI supercycle,” but is the company’s decision to shed the unit in which it is a market leader a risky abandonment of its innovative long-term growth engine, or a brilliant move to tap the AI supercycle?
SKT reorg deepens AI infrastructure push: SK Telecom is making a big push to scale national AI hubs. and Oh Je-Hoon, team leader of the AIDC business strategy team at SK Telecom told RCRTech that the company’s collaboration with OpenAI—currently at the MoU stage—reflects its broader goal of building an “AI Infrastructure Superhighway” across the country. It is putting itself at the center of Korea’s AI infrastructure expansion, with new hyperscale data center projects and a growing network of global partnerships.
AI-Powered Telecom Infrastructure
Supermicro, in collaboration with NVIDIA, delivers AI-powered infrastructure tailored for telcos, enhancing operational efficiency, network management, and customer experiences. Explore now
AI Today: What You Need to Know
The Hague suspends intervention into Nexperia: The Dutch government suspended intervention into chipmaker Nexperia.As we had reported, The Hague took action in September over “serious governance shortcomings” and concerns over the European supply of semiconductors for cars and other electronic goods. In response, Beijing blocked exports of the firm’s chips.
Optics to address AI model training bottlenecks: Lambda has become one of the first to deploy NVIDIA co-packaged optics for AI infrastructure. Traditional networking approaches are not keeping pace, so there’s a shift toward co-packaged optics to address bottlenecks asAI models train on hundreds of thousands of GPUs.
Indian semiconductor industry to get big investment: Within the next one decade, the Indian semiconductor industry is expected to rival US and Chinese chip makers, said Union minister Ashwini Vaishnaw during Bloomberg’s New Economy Forum in Singapore, where he pointed to a $10 billion semiconductor investment into that goal.
Saudi Arabia first part of AMD, Cisco, and Humain joint venture: AMD, Cisco, and HUMAIN will form a joint venture to deliver world-leading AI infrastructure. Project will deliver 1 GW of AI infrastructure by 2030, starting with a 100 MW deployment in the Kingdom of Saudi Arabia.
Semiconductor tariffs are delayed: Trump’s semiconductor tariff plan is likely delayed, with U.S. officials telling Reuters in an exclusive that the president is “treading carefully to avoid reigniting a trade war with China.” Trump had promised 100% tariffs on chips back in August.
