Dell Technologies preps new AI servers with Nvidia’s B100, B200, GB200 SuperChip

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Dell Technologies said it will support Retrieval-Augmented Generation (RAG) with Nvidia systems as it rolled out a set of Dell PowerEdge servers with HGX H200, HGX B100, HGX B200 and GB200 SuperChip.

The servers, which include Dell's first liquid-cooling system, headlined a slate of additions for the IT giant. Dell's most recent earnings call highlighted strength in AI-optimized systems and a strong backlog. Dell, HPE and SuperMicro have all seen stock gains as AI-optimized systems start to sell well. Dell outlined its additions following Nvidia CEO Jensen Huang's keynote. 

In addition, generative AI workloads appear to be moving on-premise due to cost, latency and data security.

• Will generative AI make enterprise data centers cool again?
• Dell Technologies AI optimized server backlog swells to $2.9 billion exiting Q4
• Dell Technologies to exit commercial agreement with VMware
• How AI workloads will reshape data center demand
• Dell Technologies, Hugging Face aim to make on-premises generative AI deployments easier

Varun Chhabra, SVP of Dell's Infrastructure Solutions Group and Telecom units, said: "one of the things that has been coming out loud and clear as we talk to customers is bringing GenAI to the enterprise where customer data is continues to be challenging," said Chhabra, who said data silos, governance, compliance and policies are all trouble spots. "We want turnkey solutions that package up storage, compute, networking, GPUs and a software stack that's easy to understand and consume."

Here's the roundup of what Dell announced in support of Nvidia's GTC launches to cover training to inference.

• PowerEdge XE9680 with 2x performance with Nvidia HGX H200.
• PowerEdge XE9680 also has options for next-generation AI acceleration with air-cooled Nvidia HGX B100 and Dell's first liquid-cooled 8-way GPU with Nvidia HGX B200.
• Dell will support Nvidia's GB200 SuperChip, which will feature real-time inferencing with multi-trillion parameter models, 40x lower total cost of ownership compared to Nvidia 8-way HGX  H100 and 20x better processing performance.
• Systems will support InfiniBand BlueField3 SuperNIC options as well as Spectrum-X Ethernet AI fabric. Dell has supported InfiniBand, but is adding Spectrum-X to the mix.
• PowerEdge R760xa will support Nvidia's Omniverse OVX 3.0 platform.
• Dell's system RAG design will have Nvidia microservices via NEMO and embedding framework in PowerEdge, PowerScale and PowerSwitch gear.
• PowerScale ethernet storage will have Nvidia DGX SuperPOD validation.
• Dell Data Lakehouse, which will have an analytics engine powered by Starburst.

With the Dell Data Lakehouse, Greg Findlen, senior vice president of Dell's infrastructure solutions group (ISG), AI and data management solutions, said the data lakehouse effort is designed to connect clusters on-premises and in the cloud. "We want to make it easy to scale on-premises and control cost," said Findlen. "GenAI bills are getting more expensive, and enterprise want to leverage local processing wherever that is."

Ihab Tarazi, CTO AI and Compute, Dell ISG, said enterprises have experimented with models in the public cloud, but are looking on-premises for some workloads. The Dell data lakehouse effort also plays into the company's partnership with HuggingFace.

To tie together the AI-stack and software, Dell has also launched services.

Nvidia Huang lays out big picture: Blackwell GPU platform, NVLink Switch Chip, software, genAI, simulation, ecosystem

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Nvidia CEO Jensen Huang said "accelerated computing has reached the tipping point" across multiple industries as the company launched new GPUs including a new platform called Blackwell, NVLink Switch Chip, applications and a developer stack that blends virtual simulations, generative AI, robotics and multiple computing fronts.

Huang laid out Nvidia's lofty goal during his Nvidia GTC 2024 keynote. "The industry of using simulation tools to create products, and it's not about driving down the cost of computing. It's about driving up the scale of computing. We would like to be able to simulate the entire product that we do completely in full fidelity completely digitally. Essentially what we call digital twins. We would like to design it, build it simulated operated completely digitally," said Huang.

A big theme of Huang's talk was Nvidia as a software provider and ecosystem that sits in the middle of generative AI and multiple key categories. And yes, to do what Nvidia wants it's going to take much bigger GPUs. "We need much, much bigger GPUs. We recognized this early on. And we realized that the answer is to put a whole bunch of GPUs together. And of course, innovate a whole bunch of things along the way," said Huang. "We're trying to help the world build things. And in order to help the world building states we gotta go first. We build the chips, the systems, networking, all of the software necessary to do this."

Huang laid out more powerful GPUs and systems in a cadence. The upshot is Nvidia wants to network GPUs together so they can operate as one. "In the future, data centers are going to be thought of as an AI factory. An AI factories' goal in life is to generate revenues and intelligence," he said. 

Nvidia’s GTC conference used to be for developers—and still is by the way—but given the GPU giant’s recent run Wall Street and the tech industry was closely watching Huang’s talk for signs of continuing demand, the roadmap ahead and indicators for generative AI growth. Constellation Research CEO Ray Wang noted that GTC is now the Davos of AI.Also see: Will AI Force Centralized Scarcity Or Create Freedom With Decentralized Abundance? | AI is Changing Cloud Workloads, Here's How CIOs Can Prepare

There’s a good reason why analysts were closely evaluating everything Huang said. To date, most of the spoils from the generative AI boom have gone to Nvidia with SuperMicro being an exception. Huang knew about the newfound interest in GTC. He took the stage and joked to the audience that he hoped they realized they weren’t at a concert. Huang warned that folks would here a lot of science and wonky topics.

• Will generative AI make enterprise data centers cool again?
• Nvidia's GPU boom continues, projects Q1 revenue of $24 billion, up more than $20 billion a year ago
• Nvidia's uncanny knack for staying ahead
• Nvidia launches H200 GPU, shipments Q2 2024
• Nvidia sees Q4 sales of $20 billion, up from $6.05 billion a year ago
• Why enterprises will want Nvidia competition soon
• GenAI trickledown economics: Where the enterprise stands today

Nvidia’s Blackwell platform is an AI superchip with 208 billion transistors, second generation transformer engine, 5^th generation Nvlink that scales to 576 GPUs and other features.

The game here is to build up to full data centers—powered by Nvidia of course.

Key items about the Blackwell platform and supporting cast.

• Blackwell Compute note has two Grace CPUs and four Blackwell GPUs.
• 880 petaFLOPs of AI performance.
• 32TB/s of memory bandwidth.
• Liquid cooled MGX design.
• Blackwell is ramping to launch with cloud service providers including AWS, which will leverage CUDA for Sagemaker and Bedrock. Google Cloud will use Blackwell as will Oracle and Microsoft. 

"Blackwell will be the most successful product launch in our history," said Huang. 

Other launches include the GB200 Grace Blackwell Superchip with 864GB of fast memory, 40 petaFLOSs of AI performance. The Blackwell GPU has 20 petaFLOPs of AI performance. Relative to Nvidia’s Hopper, Blackwell has 5x the AI performance and 4x the on-die memory.

To complement those processors, Nvidia has a stack of networking enhancements, accelerations and models to synchronize and have GPUs work together.

Huang said:

“We have to synchronize and update each other. And every so often, we have to reduce the partial products and then rebroadcast out the partial products that sum of the partial products back to everybody else. So, there's a lot of what is called all reduce and all the all in all gathers. It's all part of this area of synchronization and collectives so that we can have GPUs working with each other. Having extraordinarily fast lakes and being able to do mathematics right in the network allows us to essentially amplify even further.”

That amplification will come via the NVLink Switch chip that will have 50 billion and bandwidth to make GPUs connect and operate as one. 

Huang made a few jokes about pricing but did note that the company is focusing on the quality of service and balance cost of tokens. Nvidia outlined inference vs. training and data center throughput. Huang argued that Nvidia’s software stack, led by CUDA, can optimize model inference and training. Blackwell has optimization built in and “the inference capability of Blackwell is off the charts.”

Other items:

• Nvidia said it will also package software by workloads and purpose. Nvidia Inferencing Micro Service (NIMS) will aim to assemble chatbots in an optimized fashion without starting from scratch. Huang said these micro services can hand off to enterprise software platforms such as SAP and ServiceNow and optimize AI applications.
• Nvidia is working with semiconductor design partners to create new processors, digital twins connected to Omniverse. Ansys is reengineering its stack and ecosystem on Nvidia's CUDA. Nvidia will "CUDA accelerate" Synopsys. Cadence is also partnering with Nvidia.
• "We need even larger models. We're going to train it with multimodality data, not just text on the internet, but we're going to train it on texts and images and graphs and charts."
• Huang said he even simulated his keynote. "I hope it's going to turn out as well as I head into my head."

• Nvidia press releases on switches, Blackwell SuperPod, Blackwell, AWS partnership, Earth Climate Digital Twin, partnership with Microsoft and more.

ServiceNow adds more tuck-in acquisitions to build out industry, operational technology offerings

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

ServiceNow continues to go shopping. the company said it has acquired 4Industry, a Netherlands company and partner focused on manufacturing, and Smart Daily Management, a connected worker application from EY.

The most recent deals bolster ServiceNow's current portfolio of operational technology offerings that are focused on manufacturing, energy and transportation and logistics.

In February, ServiceNow said it acquired Atrinet's NetACE technology to better focus on telecommunications companies. In December, ServiceNow acquired UltimateSuite for task mining tools.

• ServiceNow delivers strong Q4, ups 2024 outlook
• ServiceNow CEO McDermott talks business transformation, generative AI, processes
• AWS, ServiceNow ink 5-year collaboration pact
• ServiceNow acquires UltimateSuite, adds task mining
• ServiceNow's latest Now Assist generative AI features highlight its strategy

ServiceNow's broader strategy with these deals is to take apps that are critical to industries and replatform them on its Now platform. For instance, UltimateSuite will be added to its platform. Atrinet NetACE will be replatformed to ServiceNow. And 4Industry, which provides an app with tasks, knowledge base and data for field workers, was already on ServiceNow's platform.

By adding these smaller acquisitions, ServiceNow can integrate them and then launch new products. 4Industry and the Smart Daily Management application will be used to build a new Connected Worker service on the ServiceNow platform in 2025.

Terms of the aforementioned deals weren't disclosed.

Cisco closes Splunk purchase, previews integrations ahead

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Cisco closed its $28 billion purchase of Splunk and outlined what'll be a steady drumbeat of product integrations in the months ahead.

In a blog post outlining what's next for customers, Cisco CEO Chuck Robbins and Splunk Executive Vice President and General Manager Gary Steele said the unified company will be heavily focused on security and observability.

• Cisco acquires Splunk in $28 billion observability, AI and cybersecurity play
• Cisco is buying Splunk for $28B in cash. What does it mean for you as a customer?

• Cisco to cut 5% of workforce, Q3 outlook weak

With the deal closed, Cisco becomes one of the largest enterprise software companies and retools its business mix. Here's how the product integrations will play out:

• Cisco's Talos threat intelligence will be embedded into Splunk's cybersecurity offerings.
• Cisco and Splunk will unify AI assistants for security so there's a common experience across the combined portfolio.
• Splunk's SIEM and SOAR platforms will leverage Cisco's cloud, network and endpoint analytics.
• The companies will combine for a full-stack observability platform to work across clouds. The integration will start with a common experience and workflow optimizations across Cisco and Splunk observability offerings.
• Over time, the combined Cisco and Splunk observability products will include AI-driven root cause enhancements and assistants including Splunk IT Service Intelligence.

Splunk and Cisco will also combine data for networking and AI deployments. More details about the Cisco integration of Splunk will land in June. Cisco Live is June 2-4 and Splunk's .conf24 is June 11-14.

• Constellation ShortList™ Security Information and Event Management (SIEM)
• Constellation ShortList™ AIOps
• Constellation ShortList™ Observability

Multi-cloud computing isn't 'a bunch of separate clouds'

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

This post first appeared in the Constellation Insight newsletter, which features bespoke content weekly and is brought to you by Hitachi Vantara.

Oracle CTO Larry Ellison can play multiple roles: Enterprise technology troll, provocateur, visionary and swashbuckling billionaire who will make big bets even when believers are few and far between.

Given that backdrop, Oracle's earnings conference calls are usually worth a listen. Ellison, along with Oracle CEO Safra Catz, did a victory lap as infrastructure-as-a-service revenue was up 49% in the third quarter. Oracle Cloud Infrastructure (OCI) has GPUs from Nvidia and is only lacking capacity to grow faster. To that end, Oracle is building data centers as fast as it can.

Many of those data centers are being built to run OCI Database Services within Microsoft Azure data centers. Oracle is also building out OCI data centers for countries that want to keep their data and large language models (LLMs) in country.

Toward the end of Oracle's conference call, Ellison dropped some knowledge on how multi-cloud environments should work. Apparently, Oracle Database@Azure can be a model for other hyperscalers. I chuckled at the idea since the only reason the Oracle-Microsoft deal works is because both have mutual enemies in Amazon Web Services and Google Cloud and mutual customers.

Ellison's take on multi-cloud is that Oracle's Autonomous Database would be the reason multiple hyperscalers would partner with the company. Ellison said (emphasis mine):

"We expect the multi-cloud initiative to continue to expand amongst other hyperscalers where we build OCI regions inside of and coexisting with their existing cloud infrastructure. We think the era of walled gardens is coming to an end. What customers really want is the ability to use multiple clouds to talk to one another. It is really called cloud computing. It's not called a bunch of separate clouds. We expect multi-cloud to become the norm and Oracle DB to be available everywhere. We think that will preserve our franchise in database because the autonomous database is a unique piece of technology, and there's nothing like it in the world. No one else is working on anything like that. No one else is even trying to duplicate the autonomous database. We think it will be it will become a very successful product. In every cloud."

A few takeaways from Ellison's comments.

• Walled gardens do need to end, but that doesn't mean they will.
• Interoperability between all the hyperscalers would be swell.
• Interoperability would also help blend private cloud infrastructure, which will play a role in AI workloads. Constellation Research analyst Dion Hinchcliffe has argued that AI workloads will fundamentally change cloud economic models.
• Oracle can prod the likes of AWS and Google Cloud to do Oracle DB deals because the company could always make support more difficult on those clouds.
• Sure, Ellison is talking up Oracle Autonomous Database, but the technology is unique enough to make strange bedfellows.
• Customers may demand more interoperability to make it easier to mix and match hyperscale compute.
• The Oracle Database@Azure model isn't much different than retail store-within-a-store partnerships (Kohl's-Amazon, Kohl's Sephora, Apple in Best Buy etc.).
• Oracle co-location strategy within other cloud data centers has lower capital expenses and gives the company more coverage. Simply put, Oracle Database@Azure, @AWS and @Google Cloud is damn good strategy and joint customers are everywhere.

Today, multi-cloud really just means there are two or three providers operating in silos. For instance, Equifax uses AWS for Oracle mission critical workloads and Google Cloud for its data layer. Despite "all-in" press releases, most enterprises prefer to have a second cloud provider to keep the primary one honest.

Regulators may also push these hyperscale cloud partnerships between archenemies. AWS recently said it will offer free data transfer for enterprises leaving the cloud. That move follows a similar announcement by Google Cloud. Microsoft Azure matched those announcements. You should check out the fine print on all those announcements.

The Federal Trade Commission has launched an inquiry into cloud computing business practices including data egress fees, software licensing and minimum usage contracts.

It's unclear whether regulators and Ellison will spur hyperscale cloud harmony, but stranger things have happened.

Chirag Mehta on the intersection of cybersecurity, design thinking and AI

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

New Constellation Research analyst Chirag Mehta outlined his approach to cybersecurity on DisrupTV. Chirag is the former Chief Product Officer at SaaS vendors Zipline and iCIMS and held various leadership roles at Google and SAP.

Here are a few takeaways from his DisrupTV appearance.

The variables in cybersecurity. Mehta said there are three key parts to cybersecurity. First, data driven signals and what a company can infer from them and respond. The second part is the human story. "We human beings are inherently trustworthy," he said. And third, what does the response look like? "We're all going to get breached. How can I respond to the test? What does that incident response kind of system look like?"

Design thinking and cybersecurity. Mehta's previous stints revolved around developing applications and products at enterprise software companies. He focuses on design thinking to humanize cybersecurity for CXOs. "You need to be proactive, more outcome based, and risk based," said Mehta. "I'm passionate about helping CXOs find their way and make their organizations more secure."

AI and cybersecurity. Perimeter and network-based approaches to cybersecurity are often flawed because "your employees are everywhere, and your data is everywhere." As a result, "AI has a role in creating a dynamic perimeter and what's going on in my environment," said Mehta. "The dynamic AI perimeter will happen, and the reason is the rise of AI means all of these problems are not solvable by human beings."

"Sophisticated, AI-driven attacks need a sophisticated response, which is AI-driven," said Mehta.

"You're going to have access to a vast amount of data telemetry, all the signals that you can analyze, and you can actually defend, including the behavior of your end users. You can defend against these attacks," he said.

What is emerging is cybersecurity platforms that use AI to become a cyber operating system.

• Zscaler acquires Avalor for security data fabric
• CrowdStrike, Palo Alto Networks duel over platforms vs. bundles
• Cybersecurity platforms spar over data, generative AI, wallet share

• CrowdStrike launches Charlotte AI, generative AI to uplevel, democratize cybersecurity analysis

ROI and cybersecurity. CXOs have said that cybersecurity budgets have been poached in the last year for AI projects. Mehta said that focus on returns is misplaced. He said:

"Security is cost of doing business. If you don't have security, you won't have business, and then you won't have ROI on anything else. It's not very clear to most leaders that they need to invest into foundational technology so that they can actually have a business, invest in generative AI and everything else in the digital transformation journey. Cybersecurity is not an optional thing."

Cybersecurity is a risk-based investment since one mistake can hit your stock price, result in SEC disclosure and harm the business, said Mehta. Enterprises will use AI to model various threat vectors and have a defense for any given situation.

By nature, defenses will have to become more autonomous. "If the idea is that a human has to get involved when there's an attack you're not going to scale," said Mehta.

Adobe: Strong Q1, outlook a bit light, GenAI on deck for Experience Cloud

Larry Dignan

Editor in Chief of Constellation Insights
Constellation Research

Larry Dignan is Editor in Chief of Constellation Insights at Constellation Research, where he leads editorial coverage focused on enterprise technology, digital transformation, and emerging trends shaping the future of business. He oversees research-driven news, analysis, interviews, and event coverage designed to help technology buyers and vendors navigate complex markets with clarity and context. ...

Read more

Adobe delivered a strong first quarter, but its outlook missed expectations.

The company, which recently dropped its plans to acquire Figma, reported first quarter earnings of $1.36 a share on revenue of $5.18 billion, up 11% from a year ago. Non-GAAP earnings in the first quarter were $4.48 a share.

Wall Street was looking for first quarter non-GAAP earnings of $4.38 a share on revenue of $5.14 billion.

As for the outlook, Adobe projected second quarter revenue of $5.25 billion to $5.3 billion with non-GAAP earnings of $4.35 a share to $4.40 a share. Analysts were looking for $4.38 a share and $5.31 billion in revenue.

For each of its product areas, Adobe posted double-digit percentage growth rates. Digital Media revenue was up 12% from a year ago, Creative revenue was up 11% and Document Cloud posted growth of 18%.

In prepared remarks, Adobe CEO Shantanu Narayen said the company is seeing strong traction with its generative AI rollout across the product portfolio. He said:

"We have innovated by delivering generative AI directly in products—with releases in Adobe Photoshop, Illustrator and Express across both desktop and mobile. AI Assistant in Acrobat and Reader unlocks the tremendous value of the trillions of PDFs around the world. We’re bringing generative AI to Adobe Experience Cloud and will demonstrate our AI Assistant for customer experience management at Adobe Summit."