What the AI Boom Looks Like From Inside Cisco with Sam Badri | The Real Eisman Playbook Ep 77

SUMMARY

Steve Eisman interviews Sam Badri, Head of Investor Relations at Cisco, to explore how the AI boom is reshaping Cisco’s business and the broader technology sector. Badri explains the drivers behind Cisco’s recent revenue acceleration, the evolving role of AI infrastructure, and why he believes the industry is still in the early stages of AI adoption.

MAIN POINTS

• Sam Badri outlines his background in consulting and equity research before joining Cisco as Head of Investor Relations. [2:00]
• Cisco’s core business is described as providing networking equipment, including switches, routers, and Wi-Fi access points, primarily to enterprises. [6:00]
• Cisco’s recent revenue growth is attributed to a surge in hyperscaler business and a campus refresh cycle, with hyperscaler sales nearly doubling but still under 20% of total revenue. [10:00]
• Cisco differentiates itself from Arista by designing its own silicon for switches, while Arista relies on third-party chips and advanced software. [15:00]
• Discussion shifts to the concentration of AI infrastructure demand among a few companies like OpenAI and Anthropic, and the implications for the tech supply chain. [21:00]
• Badri explains the concept of 'skills files' and AI agents, noting that most users remain at the basic prompt stage and only a minority have adopted advanced workflows. [24:00]
• Major tech firms like Google and Amazon are transitioning from capital-light to capital-intensive models due to massive investments in AI and data center infrastructure. [29:00]
• Community and regulatory pushback is increasing against data center expansion, particularly in states like Texas, due to concerns over power and resource allocation. [32:00]
• Badri emphasizes that Cisco has fundamentally changed in recent years, achieving record financial results and adapting its communication with shareholders and analysts. [39:00]

DETAILED ANALYSIS

Sam Badri’s career trajectory, from consulting at PwC to equity research at Credit Suisse and ultimately to his current role at Cisco, provides him with a unique vantage point on both Wall Street expectations and the realities inside a major technology company. His experience covering large-cap technology and data center infrastructure companies, including the likes of Equinix and Digital Realty, has informed his understanding of the sector’s evolution, particularly as AI adoption accelerates.

Cisco’s business model centers on providing the essential networking hardware that underpins global connectivity, such as switches, routers, and Wi-Fi access points. While these products are largely invisible to end-users, they form the backbone of enterprise and data center operations. In recent years, Cisco has expanded into silicon development, with its Silicon One chip portfolio representing a significant technological leap.

These network processing units (NPUs) are designed to efficiently transmit data across networks and have become increasingly relevant as hyperscalers—large cloud providers like Amazon, Google, and Microsoft—scale up their infrastructure to meet AI-driven demand.

The company’s financial performance has seen a marked shift. When Badri joined Cisco around 2023, revenue growth was in the mid-single digits, a rate that analysts viewed skeptically. However, recent quarters have seen revenue growth accelerate to 18%, a level not consistently seen since the dot-com era.

This growth is driven by two primary factors: a near doubling of sales to hyperscalers, which, while still less than 20% of Cisco’s total revenue, represents a rapidly expanding segment; and a 'campus refresh cycle,' where enterprises are replacing aging networking equipment, some of which has been in service for over 15 years. The refresh is partly motivated by Cisco’s decision to end support for legacy devices, prompting customers to upgrade to newer, more capable hardware.

The AI boom has been a catalyst for this transformation. Badri identifies a clear inflection point for Cisco’s AI relevance about six to nine months prior to the interview, when demand for Silicon One-based systems surged. These systems are now integral to hyperscaler data centers, which require high-performance, low-latency networking to support AI workloads.

Cisco’s main competitor in this space, Arista Networks, differentiates itself by focusing more heavily on hyperscaler clients and leveraging third-party silicon, whereas Cisco’s vertical integration allows for greater control over hardware design.

A significant portion of the discussion addresses skepticism around the sustainability of the AI boom. Eisman raises concerns about the concentration of demand among a handful of AI companies, particularly OpenAI and Anthropic, and the potential risks if these firms falter. Badri argues that the commanding position of these companies is driven by explosive consumer demand for AI-generated tokens, which fuels the entire supply chain.

He likens their dominance to a hypothetical scenario where Google Search was a private company, emphasizing the unprecedented scale and influence of these AI providers.

Despite the rapid growth, Badri notes that the majority of AI tool users remain at a basic level, primarily submitting simple prompts. Only a small fraction have progressed to advanced workflows involving 'skills files'—customized instruction sets that guide AI models to perform complex, domain-specific tasks. Badri describes how he and other power users leverage these files to dramatically increase productivity, particularly in research-intensive fields like finance.

However, he estimates that about 90% of users have yet to move beyond the beginner stage, suggesting that the market for advanced AI applications remains largely untapped.

The conversation also touches on the broader industry shift among tech giants from capital-light to capital-intensive operations. Companies like Google, Amazon, and Meta, once known for generating massive free cash flow, are now investing heavily in AI infrastructure and data centers, sometimes even raising equity capital to fund these expansions. Badri points out that while this behavior is not entirely new—hyperscalers have always been aggressive spenders on IT infrastructure—the scale and urgency have increased dramatically as companies race to secure strategic positions in the AI landscape.

Community and regulatory resistance to data center expansion is becoming more pronounced, even in traditionally business-friendly states like Texas. Power grid allocation audits and concerns over resource usage, such as water consumption, are introducing new complexities to the build-out of large-scale facilities. The physical and electrical density of modern data centers has increased severalfold compared to just a few years ago, reflecting the growing computational demands of AI workloads.

Internally, Badri has leveraged his Wall Street background to recalibrate Cisco’s investor relations strategy, fostering more direct communication with analysts and shareholders. He positions himself as both an external representative of Cisco and an internal advocate for shareholder interests, ensuring that market feedback is integrated into corporate messaging and disclosure practices.

Looking ahead, Badri believes that both Cisco and the broader sector are still in the early stages of AI adoption. He draws a parallel to the early days of Microsoft Office, suggesting that as more users move beyond basic prompts to advanced AI workflows, demand for AI infrastructure—and thus token consumption—will continue to grow. While software developers currently lead in AI tool adoption, Badri anticipates that financial services and other sectors will soon follow, further accelerating the trend.

LINKS

• The Real Eisman Playbook Premium subscription page.
• YouTube channel for The Real Eisman Playbook.
• Linktree hub for Steve Eisman’s content and social links.
• Granola AI-powered notepad for meeting notes, as mentioned in the episode.
• Granola AI trial link for podcast listeners.

KEYWORDS