NVIDIA's AI Chips Promise 50x Revenue Boost; Cloudflare, Adobe, and AWS Unveil New Tools

[00:00:02] It's Thursday, March 20th, 2025, and I'm Dave Solt. Four things to know today. NVIDIA says the more you buy, the more revenue you get. But is it really true? More tech, more problems. Employees aren't happy with their digital tools. AI that's safer, smarter, and faster. Cloudflare, HPE, Adobe, and OpenAI announcements. And AI's next battle, champagne pricing or budget bites. OpenAI, Badoo, and AWS take sides.

[00:00:30] This is the Business of Tech. So, the big tech news is the announcements from NVIDIA's GTC 2025. NVIDIA has announced the launch of its new Blackwell Ultra and Vero Rubin artificial intelligence chips. The Blackwell Ultra chips are expected to ship in the second half of 2025, promising significantly higher performance, allowing cloud providers to potentially earn up to 50 times more revenue compared to the previous generation.

[00:00:58] The company also introduced Vera, its first custom central processing unit, alongside the next-generation graphics processing unit, Rubin, slated for release in 2026. NVIDIA aims to maintain an annual release schedule for new chip families, a shift from its previous practice of releasing new architectures every two years. NVIDIA's CEO, Jensen Wang, highlighted the rapid advancements in AI and the company's ambitious plans for the future.

[00:01:27] He introduced the Blackwell architecture, which aims to revolutionize AI computing, promising one exaflop of computing power in a single rack. He revealed that the new AI operating system, Dynamo, can deliver 40 times better performance, underscoring the shift from traditional data centers to AI factories designed for large-scale intelligence generation.

[00:01:50] He projected that by the end of the year, all of NVIDIA's operations will be AI-assisted, with 10 billion digital AI agents set to enter the workforce. He emphasized that the transition to AI-driven infrastructure is not just a trend, but an economic necessity, stating, quote, For those that want to go very deep on the technical analysis here, I included a link to Semi-Analysis, which has those details. Why do we care?

[00:02:21] Well, the announcements underscore the company's relentless push to solidify its dominance in AI infrastructure, but the implications go far beyond chip improvements. Wong's statement that data centers are evolving into AI factories is a direct appeal to enterprise and cloud customers. AI isn't just an enhancement to existing IT infrastructure. It's being positioned as a core economic driver.

[00:02:43] The claim, the more you buy, the more revenue you get, is a bold reframing of AI investments from a cost center to a revenue generator. That said, I'm skeptical. Annual investments? 50 times revenue claims? That's speculative. And open source AI models are reducing reliance on proprietary NVIDIA software stacks with the cost of intelligence going down.

[00:03:06] It's a huge bet that usage will go up to this degree. The idea that AI investment is an economic necessity conveniently aligns with NVIDIA's interests. For IT services firms, the real play is helping customers navigate the landscape without becoming captive to NVIDIA's aggressive upgrade cycle. AI infrastructure isn't a one-size-fits-all game. The winners will be those who provide flexibility, efficiency, and strategic AI adoption pathways.

[00:03:37] Today's episode is supported by Huntress. Most cybersecurity solutions are built for massive enterprises with big budgets. Not Huntress. They're the fully managed cybersecurity platform built for all businesses, not just the 1%. Huntress purposely builds security solutions like EDR, ITDR, SIM, and security awareness training to equip their team of elite threat hunters to handle the heavy lifting of security for you.

[00:04:02] When threat actors strike, Huntress's 24x7 Global Sock shuts them down before they're even on anyone else's radar. But they do more than just chase alerts. They lead the charge in industry research and knowledge, bringing expert protection and peace of mind. That's why users on G2 rate their EDR number one for growing businesses. To see how their expert threat hunting team gets the job done, visit Huntress.com slash MSB Radio.

[00:04:33] More broadly, according to a recent report by Gartner, over 20% of digital workplace applications are expected to utilize artificial intelligence-driven personalization algorithms by the year 2028. This shift aims to create adaptive experiences for employees.

[00:04:49] Despite the trend, a Gartner survey conducted with over 5,000 employees revealed that only 23% of digital workers are completely satisfied with their work applications in 2024, a decline from 30% in 2022. Tori Pullman, Vice President Analyst at Gartner, emphasized the need for workplace applications to be as intuitive and empowering as popular consumer apps.

[00:05:14] To achieve this, technology leaders are urged to implement best practices in AI personalization, ensuring transparency in how algorithms work, prioritize key business outcomes, define clear requirements during vendor selection, and continually monitor and adapt these systems to meet employees' evolving needs. And while concerns about the Chinese AI model DeepSeek are prevalent, research indicates that U.S.-based AI chatbots may be collecting even more data.

[00:05:43] According to a study by Surfshark, Google Gemini tops the list as the most data-intensive AI chatbot, gathering 22 out of 35 user data types, including sensitive information such as location, contacts, and browsing history. In contrast, DeepSeek ranks fifth, collecting an average of 11 unique data types. Almost a third of chatbots share sensitive user data with third parties.

[00:06:09] A recent survey by Writer reveals that the adoption of generative artificial intelligence is causing significant internal conflict within enterprises. Two-thirds of executives report increased tension and division, with 42% stating that the technology is tearing their companies apart.

[00:06:26] Nearly 70% of leaders note that applications are being developed in isolation, and over 30% of employees, including 41% of Generation Z, admit to undermining their companies' AI strategies by refusing to use AI tools. Additionally, while three-quarters of companies are investing at least $1 million annually in generative AI, only one-third are seeing substantial returns on investment.

[00:06:52] Only 37% of executives without a formal plan feel successful at AI adoption, compared to 80% who have a formal plan. And we like use cases. The information covers how Moderna has merged its information technology and human resources departments, driven by the rising use of artificial intelligence, particularly ChatGPT, in writing performance reviews.

[00:07:17] Bryce Chalamel, the company's vice president of AI products and platforms, noted a significant increase in ChatGPT usage among employees during performance review periods, prompting the creation of a custom ChatGPT version to aid in drafting these evaluations. Moderna, which invests over $1 million annually for ChatGPT access for 5,000 employees, reports that staff members send approximately 1.2 million messages each month to the AI platform.

[00:07:47] Why do we care? The transition from hype to practical implementation is proving messy, expensive, and divisive. AI integration isn't just about adding features, it's about solving problems. If businesses deploy AI without understanding employee workflows, productivity won't improve. Providers who help bridge this gap by tailoring AI tools to actual work habits will have an edge.

[00:08:13] AI adoption isn't just a technical challenge, it's a cultural and operational one. The Surfshark study highlights a massive blind spot in AI privacy discussions. While fears about foreign AI models like DeepSeq persist, US-based ones, especially Google Gemini, are collecting far more user data.

[00:08:33] Businesses blindly adopting AI chatbots without understanding data privacy implications expose themselves to compliance risks, regulatory fines, and potential reputational damage. AI vendors are not transparent enough, leaving a gap for services that audit and mitigate AI-related data risks. Moderna's decision to merge IT and HR due to AI adoption is a real-world use case of AI's operational impact.

[00:09:00] It isn't just changing how work is done, it's restructuring corporate functions. AI-driven automation is reducing repetitive administrative tasks. They expect other companies to consolidate or rethink traditional departments. Providers who offer AI-powered workflow automation and cross-functional AI strategy will have that competitive advantage. And AI governance and bias detection services will be critical differentiators.

[00:09:27] Cloudflare has launched Cloudflare for AI, a suite of security tools designed to help businesses manage the risks associated with artificial intelligence. This initiative addresses pressing security concerns, allowing companies to monitor employee use of AI tools, detect unauthorized services, and prevent data leaks. Cloudflare's CEO, Matthew Prince, emphasized the importance of balancing innovation with safety, stating that organizations that embrace AI strategically will thrive.

[00:09:54] The suite includes a firewall for AI, which reports on the AI applications being used within a company and can block unauthorized tools. Additionally, the Cloudflare AI Gateway provides insights into model usage and prompts submitted, helping to safeguard sensitive information. Hewlett Packard Enterprise unveiled a new unified data layer for artificial intelligence, enhancing its collaboration with NVIDIA and improving its enterprise data storage portfolio.

[00:10:22] This is designed to accelerate the data lifecycle by integrating both structured and unstructured data, enabling organizations to speed up their AI applications. HPE's executive vice president stated that this unified data approach allows enterprises to run AI at scale, regardless of data location. Adobe launched a new tool called the Adobe Experience Agent Orchestrator, designed to deploy artificial intelligence agents aimed at enhancing customer experiences.

[00:10:49] The initiative targets Adobe's competition with Salesforce and the customer relationship management sector. The company introduced 10 ready-to-use experience agents, which can be deployed in minutes, to perform tasks such as evaluating sales leads and optimizing audience segments. The company is also forming an AI agent partner ecosystem, collaborating with major firms like Amazon Web Services and Microsoft to enhance its offerings.

[00:11:16] OpenAI is set to begin beta testing its new feature, ChatGPT Connectors, which will enable business customers to link applications like Google Drive and Slack to ChatGPT. This integration aims to enhance the chatbot's ability to provide information based on internal files and conversations, making it a valuable tool for companies. The feature will initially be available to ChatGPT team subscribers and plans for expansion to other platforms like Microsoft SharePoint are in the works.

[00:11:43] OpenAI assures users that data privacy will be prioritized as Slack and Google Drive permissions will be fully respected. The company emphasizes that no synced data will be used for training purposes, aiming to alleviate concerns about sensitive business information. Why do we care? Well, that's a big pile of tactical updates for you. And here's some model updates.

[00:12:07] OpenAI has launched its most expensive artificial intelligence model to date, named O1 Pro, which is now available in the developer API. This new model is designed to provide consistently better responses by utilizing more computing power than its predecessor, O1. Currently, it's accessible only to select developers who have spent a minimum of $5 on OpenAI API services.

[00:12:30] The pricing for O1 Pro is significantly higher, costing $150 per million tokens for input and $600 for generated tokens, making it twice the price of previous models, GPT 4.5 and 10 times the cost of regular O1. Despite the high cost, OpenAI believes that the improved performance will attract developers.

[00:12:52] However, earlier reviews have been lukewarm, with users noting difficulties in solving Sudoku puzzles and understanding simple optical illusion jokes. Internal benchmarks from late last year indicated that O1 Pro only slightly outperformed the standard O1 in coding and math problems, although it did provide more reliable answers. Baidu has launched two new artificial intelligence models in a bid to compete with the rapidly rising DeepSeq.

[00:13:19] The company introduced Ernie 4.5 and Ernie X1, with Ernie X1 claiming to deliver performance comparable to DeepSeq's R1 model at half the cost. Baidu plans to offer Ernie X1 to app developers for as low as $2 won, approximately $0.28 for every $1 million tokens input and $8 won for output. The company's chatbot app, Ernie Bot, will allow users to access both Ernie 4.5 and Ernie X1 free of charge.

[00:13:47] Despite being the first major Chinese firm to launch its own AI model in early 2023, Doof faces stiff competition from DeepSeq, ByteDance, and Alibaba. Amazon Web Services has launched its Amazon Q business platform in Europe, specifically in Dublin, responding to significant demand from European customers who value data sovereignty.

[00:14:09] The platform, which allows staff to find internal data sources more effectively, will now operate entirely within the region, enhancing regulatory compliance and performance. David Pessis, the director of worldwide go-to-market for Amazon Q, emphasized that many customers prefer regional support and the security of keeping their data within Europe.

[00:14:28] In a related survey, 51% of UK organizations consider data sovereignty a crucial part of their data management strategy, highlighting the growing focus on sovereign services across the continent. AWS plans to aggressively invest in expanding its European data centers, with a notable €7.8 billion investment announced in May 2024 for establishing a sovereign cloud infrastructure in Germany. Why do we care?

[00:14:55] Well, the key theme across the updates is AI specialization and regionalization. Models are becoming more targeted in performance, pricing, and geography rather than one-size-fits-all solutions. AI pricing is diverging from OpenAI's premium pricing to Baidu's ultra-cheap models. Businesses must choose wisely, balancing capability with cost efficiency. AI optimization will be critical. The real winners won't just be those who deploy AI, but those who do it efficiently.

[00:15:27] Are you ready to get your brand in front of the tech leaders shaping the future of managed services? Here at the Business of Tech, we offer flexible sponsorship opportunities to meet your needs, whether it's live show sponsorship, podcast advertising, event promotion, or custom webinars. From affordable exposure options to exclusive sponsorships, our offerings are designed to fit businesses and vendors of all sizes looking to make an impact.

[00:15:54] Prices start at just $500 per month, making our packages a fraction of typical event sponsorship costs. Be a part of the conversation that matters to IT service providers worldwide. Join us at MSP Radio and amplify your message where it counts. Visit MSP Radio dot com slash engage today to explore all the ways we can help you grow. Thanks for listening.

[00:16:24] Today is National Proposal Day, International Day of Happiness, and apparently also Bach Beer Day and Alien Abduction Day. Oh, and National Raviol. Nerdy Ocon will be held in Palm Springs, California from April 7th through 9th. Visit NerdyOcon.com to learn all about it. The Business of Tech is written and produced by me, Dave Sobel, under ethics guidelines posted at businessof.tech.

[00:16:50] If you've enjoyed the show, make sure you've subscribed or followed on your favorite platform. It's free and helps directly. Give us a review, too. If you want to support the show, visit patreon.com slash MSP Radio and you'll get access to content early. Or buy our Why Do We Care merch at businessof.tech. Have a question you want answered?

[00:17:14] We take listener questions, send them in, ideally as a voice memo or video to question at MSP Radio dot com. I answer listener questions live on our Wednesday live show on YouTube and LinkedIn. If you've got a comment or a thought on a story, put it in the comments if you're on YouTube or reach out on LinkedIn if you're listening to the podcast. And if you want to advertise on the show, visit MSP Radio dot com slash engage.

[00:17:39] Once again, thanks for listening and I will talk to you again on our next episode. Part of the MSP Radio Network.