SpaceX, OpenAI, Anthropic Are Going Public — So What?

Introduction

Summer 2026 brings three big names to the public markets in quick succession: SpaceX in June, OpenAI in September, Anthropic in October. These aren't abstract names — they're foundational providers of AI infrastructure, compute, and agentic tooling.

SpaceX **filed its S-1 with the SEC on May 20, 2026**, with trading expected to begin on Nasdaq under the ticker SPCX as early as **June 12**. OpenAI began preparing a **confidential prospectus** with Goldman Sachs and Morgan Stanley that same week, with Sam Altman targeting a **September 2026** debut. Anthropic, according to The Information, is focused on an **October 2026** listing — it raised $30 billion in its Series G in February 2026, reaching a post-money valuation of $380 billion.

As a developer, you can read this as a "finance story." But it's also news that "the owners of my tools now answer to public markets."

What "Answering to Public Markets" Actually Means

When a company goes public, investors demand growth, margin, and predictability. That pressure shows up in product decisions: features slow down, prices rise, enterprise comes first, community gets pushed to the back.

Here's what's striking: Sam Altman himself is uncomfortable with the pressure. His answer on the Big Technology Podcast in December 2025 became a headline:

And it's not just personal preference. According to Reuters' reporting on The Information, **OpenAI CFO Sarah Friar** has internally raised concerns about Altman's 2026 IPO schedule and the plan to commit up to **$600 billion in spending** over five years. Friar reportedly questioned whether OpenAI is organizationally ready for a listing and whether slowing revenue growth can support that compute spend.

In other words: the "answering to markets" pressure is already showing up internally, before the listing.

This isn't determinism — plenty of companies kept shipping good products after going public. But as a pattern, the risk is real.

Pricing Shifts: The Copilot Case

GitHub Copilot is **moving to an AI Credits system starting June 1**: flat rate is over, every token counts. According to GitHub's own announcement, **1 AI Credit = $0.01**, and token consumption covers input, output, and cached tokens. Chat, cloud agent, code review — all of it consumes credits. Code review also burns GitHub Actions minutes on top of that.

Per Visual Studio Magazine's analysis of the FAQ thread, developers share one core worry: token-heavy workflows (agentic coding, chat, code review) become more cost-sensitive. You get less for the same money. One community comment captures it: "You Will Get Less, but Pay the Same Price."

For annual plan holders, it's harsher. WellWells' breakdown based on GitHub docs shows that someone making ~300 Claude Sonnet 4.6 calls per month under the old system drops to **33-50 calls under the new model multipliers** — a 6-9x cost increase.

This pattern isn't unique to Copilot. **Simon Willison** (creator of Datasette, co-creator of Django) put it bluntly on the High Leverage podcast in April 2026:

The phrase "invisible price hike" is the key one. Sticker price stays the same, but a tokenizer change raises the effective cost — exactly the kind of maneuver you'd expect from companies under earnings-call pressure.

Is the Lock-in Risk Real?

Being dependent on Anthropic or OpenAI starts to mean something different once they're public. API pricing and rate limit decisions are now part of quarterly earnings calls.

This isn't theoretical. In January 2026, Anthropic blocked third-party harnesses — including **OpenCode** — from accessing Claude via subscription OAuth. The reaction was sharp. **David Heinemeier Hansson (DHH)**, creator of Ruby on Rails, posted what became the headline critique on X:

DHH summarized it in two words: "customer hostile." The episode wasn't just a policy fight — it was proof that even before going public, Anthropic was willing to cut off its developer ecosystem the moment commercial logic demanded it. **Peter Steinberger** (creator of OpenClaw, who later joined OpenAI) migrated his entire 12-agent development setup from Anthropic to OpenAI Codex.

This is why open source alternatives like OpenCode matter. Being able to switch platforms, staying portable, refusing to depend on a single vendor — these aren't technical preferences anymore. They're strategic ones.

How is portability achieved in practice? Because the **OpenAI-compatible API** has become a de facto standard, gateways like OpenRouter and LiteLLM reduce model switching to a config change. As open-weight models like DeepSeek R1 and Llama 4 close in on frontier capability, "being locked to a provider" is increasingly a choice — but only if you've invested in those alternatives deliberately, before you need them.

Conclusion

Where I land today: choosing tools by asking "which is best right now?" is no longer enough. "Whose decisions will steer this tool in the coming period?" has to be on the list.

Sam Altman being "0% excited" about an IPO, OpenAI's CFO pushing back internally on his timeline, Anthropic aggressively restricting developer access even while still private, Copilot's shift to usage-based billing, Simon Willison's "invisible 40% price bump" observation — all of these point in the same direction: economic pressure is shaping product decisions.

Going public doesn't change that. It just makes it visible. The quarterly earnings call serves as a reminder every three months.