Bloomberg Tech Event Special | Bloomberg Tech 6/04/2026

Key Concepts

• AI Infrastructure Playbook: The focus on "picks and shovels" (compute, memory, optics, networking) as the primary beneficiaries of current capital expenditure.
• Agentic AI: The shift from simple chatbots to autonomous agents capable of executing tasks, connecting to databases, and collaborating with humans.
• Scaling Laws: The empirical observation that increasing compute, data, and algorithmic resources leads to higher intelligence.
• Data Context: The critical missing link for AI; the ability to feed enterprise-specific data into models to make them actionable.
• Capital Formation Cycle: The transition of AI from a mere technology product cycle to a massive financial cycle involving record-breaking equity raises and IPOs.
• SaaS Apocalypse: The debate over whether AI will render traditional software companies obsolete or force them to evolve into "agentic" platforms.

---

1. Market Dynamics and Broadcom Earnings

The market experienced significant pressure, with the NASDAQ 100 seeing its worst performance since mid-May. A primary driver was Broadcom, which reported $16 billion in custom silicon (TPU/ASIC) sales, missing the street’s expectation of $17.2 billion.

• Key Insight: Despite the miss, analysts note that Broadcom has visibility into 2028, suggesting long-term demand remains robust.
• Market Sentiment: The stock had added $270 billion in market cap leading up to the earnings, leading to concerns that the stock was "priced for perfection."

2. The AI "Bubble" Debate

Industry leaders and investors are divided on whether the current AI rally is a bubble.

• The Bullish Perspective: Andrew Feldman (CEO of Cerebras) argues it is the "opposite of a bubble" because builders cannot keep up with the $25 billion backlog of demand.
• The Investor Perspective: Altimeter Capital’s Apoorv Agrawal notes that AI has become a massive capital formation cycle. He distinguishes between those receiving capex (compute, energy, memory, networking) and those spending it (labs like OpenAI, Anthropic, and Google).

3. DataBricks and the "Context" Problem

Ali Ghodsi, CEO of DataBricks, argues that we have already achieved Artificial General Intelligence (AGI) in terms of raw capability, but it lacks the necessary context to be fully productive.

• Methodology: DataBricks is focusing on "Genie," a tool designed to infuse enterprise data context into AI agents.
• Case Study: Novo Nordisk (maker of Ozempic) uses Genie to ingest clinical trial data, reducing the time to analyze study results from weeks to minutes.
• Software Thesis: Ghodsi supports Jensen Huang’s view that software is not in trouble; rather, agents will become the primary users of software, leading to more software being written in the next two years than in all of human history.

4. Identity and Security in the Agentic Era

Todd McKinnon, CEO of Okta, discussed the "agentic buildout."

• Framework: Okta is positioning itself as the "identity layer" for AI agents. As agents gain the ability to access credit cards and sensitive data, the risk of "rogue agents" increases.
• Argument: The barrier to widespread agent adoption is not model capability, but the ability to securely connect these models to disparate enterprise databases and systems.

5. Economic Impact and Federal Reserve Perspective

Mary Daly, President of the Federal Reserve Bank of San Francisco, provided a macroeconomic view on AI.

• Productivity: Daly noted that while there is tremendous enthusiasm and investment, the economy has not yet seen widespread, transformative productivity gains. She views the next year as the "litmus test" for these gains.
• Inflationary Impact: She addressed the debate on whether data center buildouts are inflationary. While they compete for limited resources (energy, infrastructure equipment) in the short term, they represent long-term infrastructure investment.
• Labor Market: Businesses are currently cautious about hiring, preferring to "interrogate" how AI can handle tasks before expanding their headcount.

---

Synthesis and Conclusion

The overarching theme of the event is the transition from the "hype" phase of generative AI to the "infrastructure and integration" phase. While market volatility persists due to high expectations for chipmakers like Broadcom, the consensus among leaders is that the AI super-cycle is in its early stages. The primary challenges identified are not model intelligence, but rather secure connectivity (Okta), data context (DataBricks), and energy/infrastructure capacity. The "SaaS Apocalypse" is viewed as a transition period where companies that fail to integrate agentic capabilities will become legacy vendors, while those that adapt will thrive in a new era of automated digital work.