Bloomberg Tech Podcast Summary - March 19, 2025

Episode Title: Nvidia’s New AI Offerings, Musk Buys More X Shares
Host(s): Caroline Hyde (A), Tim Stanback (B)
Notable Guests: Mandeep Singh (Bloomberg Intelligence), Kim Forrest (Becker Capital Partners), Matthew Prince (Cloudflare CEO), Jay Jacobs (BlackRock), Christine Esserman (Accel), Kurt Wagner (Bloomberg), David Wadhwani (Adobe), Henry Ren, Katie Roof

Overview

This episode of Bloomberg Technology dives deep into the rapidly evolving landscape of AI hardware and software, led by Nvidia’s latest chip announcements at their annual GTC conference, the ongoing “AI arms race” among hyperscalers and tech giants, surging investments in AI infrastructure, and updates from key companies including Cloudflare, Adobe, Tencent, and Musk’s X. The episode also explores venture trends and competition in AI developer tools.

Key Themes & Discussion Points

1. Nvidia Unveils New AI Chips and Technology at GTC

(00:01:33–05:30, 28:09–31:46)

• Event Recap: Nvidia CEO Jensen Huang launched new chips and an updated AI roadmap, including the Blackwell Ultra platform (with future platforms “Vera Rubin” and “Feynman” teased). Investors reacted coolly, with shares bouncing back slightly after a prior sell-off.
• Performance & Roadmap: Nvidia continues to showcase advances in AI chip performance, promising substantial gains—Vera Rubin chips expected “to 3x the improvements on Blackwell Ultra” (03:52).
• Analyst Perspective:
  • Mandeep Singh comments on the demand for “reasoning” compute (“reasoning requires 100x more compute... the demand trends seem to be reassuring,” 03:04), yet flags heightened competition as hyperscalers build their own ASIC chips.
  • “You just can't rule out the fact that ASICs are a real possibility... That competitive environment will get challenging.” (04:27)
• Competitive Landscape: Nvidia’s dominance is evident, but major customers developing custom silicon pose a long-term risk. The field is increasingly defined by “frenemies”—partners and competitors at once.
  • “Your competition is the four or five main hyperscalers that are still looking to develop that ASIC option.” (04:27–05:30)

Notable Quote

“To me it's still all about data centers and servers. And they seem to have a very clear roadmap till 2027 in terms of what kind of performance improvements they're looking [for] with the next version of their chips.”
— Mandeep Singh, Bloomberg Intelligence (03:04)

2. The Innovation Race: Hardware vs. Software in AI

(05:30–10:41)

• Nvidia’s Moat: Kim Forrest (Becker Capital Partners) observes that while Nvidia currently “rules the hardware world,” human inventiveness may soon find ways to optimize via software.
  • “Just throwing really expensive chips and tons of chips at a problem is not always the answer.” (05:56)
• DeepSeek Disruption: Conversation pivots to DeepSeek and innovations that may reduce dependence on ever-more expensive hardware by improving software efficiency.
• Economics of AI: High costs of training, maintaining, and monetizing large language models remain a challenge. Models are “85 to 95% correct and that is underwhelming” (08:34).
  • Forrest argues, “Why not use [software] on AI itself—to use these data centers we already have and rethink … training, testing and validation—why not make that better by better software? Crazy.” (07:24)
• Ongoing Challenges: Even with new open-source offerings like Nvidia’s Dynamo, the problem of incorrect AI outputs persists. “Wrong answers are still a problem” (10:00).

Notable Quote

“If you're going to build your business on something that's incorrect, 10 to 15% of the time... these are real world problems that we have to get unexcited about and fix.”
— Kim Forrest (08:34)

3. Cloudflare’s AI Security Push

(12:34–15:59)

• New Product Launch: Cloudflare for AI offers guardrails and controls to help companies avoid data leaks and misuse when deploying AI.
  • CEO Matthew Prince explains: “What we've heard from our customers is that they want guardrails… so that's what we're launching with Cloudflare for AI.” (12:59)
• Broad Context: The AI rush leads to mistakes and vulnerabilities—Cloudflare for AI aims to mitigate such risks.
• Industry Role: Cloudflare integrates deeply with AI firms: “Most of the major AI companies also use Cloudflare... we’ve been able to integrate those two sides in a way that gives you a holistic offering.” (13:56–14:45)
• AI Reality Check: Prince acknowledges rampant hype but states, “Even if only 1% of what's being done today turns into value... that 1% has the opportunity to have enormous returns.” (14:57–15:50)

Notable Quote

“Anyone who's not doing at least something experimenting in AI is probably missing the boat here.”
— Matthew Prince, Cloudflare CEO (14:57)

4. Big Tech Updates: Google Pixel, Tesla, Xiaomi, Tencent, and Wiz Acquisition

Various Segments: 18:03–20:04, 33:08–35:29

• Google: Launches new budget Pixel 9a ($499), competitive timing with iPhone 16.
• Tesla: Approved to carry passengers in California (with human drivers), paving the way for potential robocar services.
• Xiaomi: Accelerating EV production, raising 2025 delivery targets.
• Tencent: Reports fastest revenue growth since 2023; huge capital expenditure jump—quadruple YoY—on urgency to strengthen AI capabilities amid surging GPU demand (20:04).
• Google Acquires Wiz for $32B, signifying major consolidation in cloud security. Regulatory climate contributed to deal speed (33:08–34:52).

5. Elon Musk Buys More of X (Twitter)

(21:03–24:08)

• Musk’s Stake: Bought an additional $150M of shares, now owns ~75% of X.
• Signal to Market: Bloomberg’s Kurt Wagner suggests the move is “more a sign … that this company is still valued at where he, or close to where he purchased it.” (21:26)
• Business Health: Signs of improvement include raising capital at a $44B valuation and banks’ willingness to offload X’s debt at original loan prices. Nonetheless, advertiser lawsuits create challenges.
• Platform Shift: Wagner notes, “The news value of Twitter or now X, in my opinion, has gone down.” Changes to verification and ranking have reduced X’s utility as a real-time news source, especially in a second Trump presidency context (23:23).

Notable Quote

“Twitter was really like a must have for me … I just don’t feel like I’m getting news through apps in the way I used to through Twitter. And I think the verification thing in particular has a lot to do with that.”
— Kurt Wagner, Bloomberg (23:23)

6. AI Investment: BlackRock's Thematic Focus & Sector Insights

(25:52–29:42)

• Thematic Investing Dominates: Jay Jacobs (BlackRock) says 2025 is defined by thematic investing (AI, geopolitics) over strictly macro/interest rate trends. “Developments in AI… that’s what’s driving a lot of the investment this year” (25:52).
• Valuation Debate: Despite volatility and share price pullbacks, “people are buying the dip” in AI. “Incredible AI companies… trading at the same PE as 75-year-old fast food companies.” (26:49)
• Long-Term Opportunity: Focus shifting from hardware ("picks and shovels") to unique data and transformative applications. “That unique high quality data is going to be where a lot of value lies.” (28:53–29:42)
• Broader Allocations: Both end investors and advisors see this as a long-term entry opportunity.

7. AI Infrastructure Frenemies: Musk, X AI, Nvidia-Team Up

(30:16–31:46)

• Major Alliance: Musk’s X AI, Nvidia, Microsoft, BlackRock, and UAE-backed MGX partnering on $30B+ in U.S.-centric AI infrastructure, with focus on data centers and energy. Despite being rivals in other areas, collaboration is fueled by massive demand and opportunity.
  • “The whole world of the AI space Seems to be one built on frenemies.” (30:58)
• OpenAI’s Stargate Project: 400,000 Nvidia chips to be installed in a Texas data center, signifying real-world hardware demand and continued infrastructure build-out.

8. AI Developer Tools and Venture Trends

(35:29–41:11)

• Graphite (Accel-backed): New $52M Series B to scale an AI-powered code review platform, aiming to streamline and enhance human+AI collaboration for code development.
  • “Graphite is the collaboration layer where human developers and AI agents can collaborate on code changes.” (36:01)
• Industry Crowding: Many startups and major players (e.g., Copilot, Poolside) compete in this booming market; Graphite bets on both speed of iteration and in-person NYC-based team culture.
• Talent Demand, Not Replacement: Despite AI, “it's actually a war for talent… software developers can just be much more productive with tools like Graphite.”(36:51)
• Tech Excitement: Accel focuses on productivity, collaboration, security as the most promising investment areas in the current software/platform paradigm shift.

Notable Quote

“We're seeing incredible productivity gains and companies are going multiproduct faster. I actually think this is going to be an accelerant to the software development career.”
— Christine Esserman, Accel (36:51)

9. Adobe: AI-Driven Creative Suite Expansion

(42:07–45:28)

• Product Innovations: Adobe is integrating AI and “agentic” technologies into its creative suite to enable large-scale, personalized content for enterprise clients.
  • “The entire show... has been really around enabling enterprises to create more content at scale through AI and agentic experiences.” (42:07)
• Interoperability: Partnerships now include Google and Runway among others, cementing Adobe’s position as a one-stop AI creative platform.
• Monetization: AI products already at $125M run-rate, with expectations to double in nine months; AI is “influencing billions of dollars” across Adobe’s business.
  • "We expect that to double in the next nine months. ... So we're very bullish about that continuing to grow and actually accelerate in the years ahead." (45:02)

Memorable Quotes (with Timestamps)

• Nvidia's AI pivot:
  “Reasoning requires 100 times more compute ... the demand trends seem to be reassuring coming from Nvidia.” — Mandeep Singh (03:04)
• Software’s opportunity in AI:
  “More software, less hardware.” — Kim Forrest (05:56)
• Cloudflare on AI risk:
  “What we've heard from our customers is that they want guardrails, they want controls... that's what we're launching with Cloudflare for AI.” — Matthew Prince (12:59)
• X’s diminishing news value:
  “The news value of Twitter or now X, in my opinion, has gone down.” — Kurt Wagner (23:23)
• BlackRock on AI investment:
  “Incredible AI companies ... are trading at the same PE as 75-year-old fast food companies right now.” — Jay Jacobs (26:49)
• Adobe’s AI ambition:
  “The entire show... has been really around enabling enterprises to create more content at scale through AI and agentic experiences.” — David Wadhwani (42:07)

Key Timestamps for Major Segments

| Time | Segment | |-------------|---------------------------------------------------------------| | 01:33–05:30 | Nvidia GTC news, AI chip landscape | | 05:30–10:41 | Hardware vs. software debate, DeepSeek, AI economics | | 12:34–16:35 | Cloudflare releases AI security suites | | 18:03–20:04 | Google hardware, Tesla ride-hailing, Xiaomi, Tencent surge | | 21:03–24:08 | Musk’s stake increase in X, platform analysis | | 25:52–29:42 | Jay Jacobs (BlackRock): AI thematic funds and value | | 30:16–31:46 | X AI/Nvidia/Microsoft/BlackRock mega-AI-infra partnerships | | 33:08–35:29 | Google $32B acquisition of Wiz | | 35:29–41:11 | Accel’s Graphite and the race in AI code review/collaboration | | 42:07–45:28 | Adobe’s new AI-powered creative suite, monetization |

Conclusion

This episode illustrates the rapid evolution and interdependencies in the AI ecosystem. Nvidia’s hardware dominance continues amidst gathering competitive clouds. Software innovation and security rise in parallel, with Cloudflare, Tencent, and Adobe serving as examples of firms racing to keep pace. Meanwhile, investors—retail, institutional, and VC—are still betting big on AI's disruptive future, with “frenemy” alliances and enormous bets on both infrastructure and developer tooling. Even as some tech stocks stumble, conviction in AI’s long-term value seems as strong as ever.