Oracle files to sell $20B in new stock, bonds to finance AI buildout

Original Article ↗ Hacker News Discussion ↗

Oracle’s plan to raise $20 billion via new stock and bonds to fund AI data center expansion is prompting questions about shareholder dilution, mounting debt, and whether the company is betting its stable enterprise and ERP business on a low-margin, rapidly evolving “AI infrastructure” play. Commenters debate if AI equity is already overpriced as a hedge against job loss, whether Oracle’s heavy leverage and commoditized role in the AI stack create real default risk, and how a potential AI bubble or Oracle failure would ripple through markets and long‑entrenched customers. Others focus on governance and politics, expressing deep skepticism about bailouts, regulatory capture, and the broader sustainability of current AI spending.

Equity raise, dilution, and market reaction

  • Oracle plans to raise $20B, half in new equity (2.5% dilution at current cap), half in debt.
  • Some argue dilution is “purely ownership/voting” and economically neutral if new capital grows the “pie.”
  • Others stress dilution is real: existing shareholders own a smaller percentage, EPS falls, and if the share price drops during the raise, more shares must be issued, increasing dilution.
  • Debate over simple examples of buybacks vs. issuances and how market cap vs. equity value are treated; no consensus, but all agree Oracle stock is down sharply since 2025.

Debt load, default risk, and core business

  • Oracle already carries heavy debt (~$125B mentioned) with interest around $5.7B/year; debt is just above junk.
  • CDS pricing suggests a non-trivial default probability by 2031 (around 20–22%).
  • Some say the core database/ERP business is “healthy” so default is unlikely; others note debt service could starve R&D and still allow a restructuring where shareholders are wiped and bondholders own a “new Oracle.”

AI buildout strategy and hardware risk

  • Oracle is seen as providing largely commodity compute to AI labs (e.g., a very large multi‑hundred‑billion‑dollar “Stargate” deal with OpenAI).
  • Concern: rapid GPU advances (Blackwell → Rubin) could obsolete data centers before full monetization, especially given massive power and cooling jumps.
  • One side claims Nvidia offers modular upgrades and buybacks of older boards; others dispute any formal buyback/trade‑in program and note major architectural/power differences that limit “drop‑in” upgrades.

ERP, cloud, and commoditization

  • Some argue ERPs’ best days are over: AI agents could slash consulting headcount and customization revenue.
  • Counterpoint: ERPs still have strong compliance/audit value; they’re deeply embedded and already heavily customized.
  • Reports that Oracle is putting legacy telecom and ERP products into maintenance/sunset while pushing hard on cloud and AI infrastructure, possibly swapping high‑margin software for lower‑margin commodity data center business.

AI jobs, hedging, and bubble/bailout debates

  • Discussion on whether investing in AI firms hedges job loss: many think valuations already price in extreme success; there’s a wide region where AI harms jobs but doesn’t make investors rich.
  • Some see the AI buildout as a systemically dangerous bubble, likening it to 2007 CDS and predicting inevitable bailouts.
  • Others argue an AI crash would not be “too big to fail”: employment impact is limited, tech is geographically/politically concentrated, and the federal government may lack fiscal capacity or political will to rescue it.

Sentiment toward Oracle

  • Strong negative sentiment: many openly hope for Oracle to misstep, default, or be broken up; some specifically mention reclaiming assets or trademarks.
  • Others caution that a sudden collapse would disrupt many institutions reliant on Oracle but acknowledge that, in bankruptcy, critical software units would likely be sold and continue operating under new ownership.