Digital Colliers Daily Briefing — June 12, 2026
Public markets, private capital, and developer tooling each posted defining moments today. SpaceX priced the largest IPO in history at a $1.77 trillion valuation, setting a benchmark the next wave of AI listings will be measured against. Jeff Bezos's Prometheus added another $12 billion to a war chest aimed at automating physical engineering. And GitHub moved agentic automation into public preview inside Actions, threading AI agents directly into the software delivery lifecycle.
1. SpaceX prices at $135, raising $75B and concentrating control in Musk
What happened. SpaceX priced 555.6 million shares at $135 ahead of Friday's Nasdaq debut under the SPCX ticker, raising $75 billion at a $1.77 trillion valuation, according to TechCrunch. The offering eclipses Saudi Aramco's 2019 record of $24.9 billion by a factor of three. Underwriters hold a greenshoe option on an additional 83.3 million shares that could push the haul past $86 billion. The book was reportedly four times oversubscribed, with Bloomberg citing $100 billion in retail orders alone and a $5 billion order from BlackRock. Hyperliquid's synthetic SpaceX market was pricing shares at $167 on the eve of trading, implying a 24% first-day pop.
Why it matters. The structure is as consequential as the size. Musk retains 85.1% of voting power through Class B shares carrying 10 votes apiece, including a billion-share tranche contingent on a million-person Mars colony. As Wired notes, "the only way he can be removed as CEO is if he votes to fire himself" — a governance posture that public pension funds urged him to soften before the offering and that drew the "novel and extreme" label from skeptical investors. SpaceX also reserved roughly 30% of the float for retail — well above the 5–10% Fidelity cites as typical — and Nasdaq-100 rule changes will force index funds to hold the stock almost immediately, embedding it across 401(k)s whether buyers opt in or not.
Who is affected. Beyond Musk, the windfall list includes Valor Management's Antonio Gracias (503.4 million shares, ~$68 billion at IPO price), board member Luke Nosek, COO Gwynne Shotwell, and roughly 400 venture backers from two decades of private fundraising. SPV investors face a murkier picture: TechCrunch reports that backers in multi-layered vehicles — some stacked four or five deep — may wait eight or nine months for share distribution as rolling lock-ups lift, and some may discover their allocations were "eroded by fees" or, in worst cases, never existed. Anthropic and Anduril have already moved to disallow such structures. Outside the financials, Safe AI Now staged a Times Square effigy protest tying SpaceX's valuation to xAI's Grok and outstanding litigation over nonconsensual sexual imagery, arguing that "this IPO is a liability shift" to public shareholders.
What to watch next. Friday's open and the size of any pop; whether bankers exercise the greenshoe; the trajectory of the rolling lock-ups and the SPV reckoning Unicorns Exchange's Idan Miller predicts will expose "scammers or fraud"; and the read-through for OpenAI and Anthropic, both of which have filed confidential S-1s. SpaceX's absorption of xAI has already made the consolidated entity unprofitable, and its compute sales to Anthropic and Google through Colossus mean a stumble would propagate across the AI infrastructure stack.
Sources:
- SpaceX officially prices shares at $135 in the largest IPO ever — TechCrunch AI
- Why You Might Already Own SpaceX Shares, Siri's AI Makeover, and Knicks Owner's Surveillance Machine — Wired
- Massive Effigy of Elon Musk Raised Over Times Square to Protest Grok — Wired
- SpaceX SPV investors won't know their true holdings until post-IPO lock-ups lift — TechCrunch AI
- SpaceX IPO Puts Elon Musk's 'Extreme' Ownership to the Test — Wired
- You Probably Won't Get Rich Off the SpaceX IPO — Wired
2. Prometheus draws $12B at $41B valuation to automate physical engineering
What happened. Prometheus, the physical AI company co-founded by Jeff Bezos and former Verily co-founder Vik Bajaj, raised $12 billion at a $41 billion valuation, TechCrunch reports. Bezos personally participated alongside JPMorgan Chase, Goldman Sachs, and BlackRock. The round follows an initial $6.2 billion raise late last year, bringing total capital to roughly $18 billion in under a year. Prometheus is pursuing what it terms an "artificial general engineer" — software capable of designing and manufacturing complex physical systems ranging from jet engines to drug compounds — with 150 employees across San Francisco, London, and Zurich. Specific products remain undisclosed, and Bezos signaled most of the new capital will fund compute.
Why it matters. The check size places Prometheus among the largest AI funding events ever, and crucially it is targeted at physical rather than purely digital problems. Investors backing the round are betting that the moats around real-world engineering — tooling, materials science, regulatory data, manufacturing process knowledge — are deeper than those around language models. Bezos used the announcement to stake out a contrarian labor thesis, predicting "labor scarcity" rather than mass displacement, with productivity gains converting two-earner households into one-earner ones. That framing sits awkwardly against Amazon's own record under CEO Andy Jassy, where automation has accompanied tens of thousands of layoffs over the past year.
Who is affected. Aerospace primes, pharmaceutical R&D organizations, and industrial design firms are the most direct targets — sectors where engineering cycle times and tacit expertise have historically resisted software automation. The round also reshapes the competitive map for physical AI peers and contract engineering firms, and it pulls another tier-one syndicate (JPMorgan, Goldman, BlackRock) deeper into the AI capex cycle just as SpaceX prices.
What to watch next. Any product disclosure or design partner announcements; how Prometheus sources the proprietary engineering data needed to train against jet engines and drug compounds; and whether the company's hiring pulls senior engineering talent from incumbents in aerospace and pharma. The compute allocation also matters: at this scale, Prometheus becomes a meaningful customer for the same infrastructure providers serving the frontier labs.
Sources:
3. GitHub Agentic Workflows opens to public preview, with org-level billing and tightened bot controls
What happened. GitHub moved Agentic Workflows into public preview, letting developers define agent-driven automations — issue triage, CI failure analysis, documentation updates, cross-repository changes — in natural-language Markdown files that compile to standard Actions YAML. The platform ships with layered safeguards: read-only permissions by default, sandboxed execution behind an Agent Workflow Firewall, an integrity filter on accessed content, a safe-outputs validation step, and a dedicated threat detection job that scans proposed changes before they apply. In two simultaneous updates, GitHub eliminated the need for personal access tokens by enabling use of the built-in GITHUB_TOKEN — with AI credits billed directly to the organization when copilot-requests: write is set — and allowed github-actions[bot]-created pull requests to run CI/CD workflows pending user approval, closing a gap that previously let bot PRs merge without CI.
Why it matters. This is the operational plumbing the agentic-coding narrative has been missing. Reusing Actions runners, policies, and permission models means enterprises don't have to stand up a parallel agent infrastructure or accept the security debt of long-lived PATs. Organization-level billing changes procurement dynamics for Copilot CLI usage, and cost-center controls give finance teams a credible way to cap spend per workflow run. The bot-PR-CI fix addresses a concrete merge-safety hole — one Hud.io CTO May Walter summarized as "getting an agent to open a pull request was never the hard part. Trusting it enough to merge is."
Who is affected. Carvana and Marks & Spencer are the named early adopters, with M&S CTO James Hoare describing reusable workflow catalogues spanning security, quality, and delivery that compress hours of triage and remediation into minutes. More broadly, every Actions-using organization now has a sanctioned path to put agents into the SDLC; the feature is available across Copilot Free, Pro, Pro+, Business, and Enterprise tiers. Security and platform teams pick up a new threat surface to govern, while DevEx teams gain a building block they previously had to assemble from third-party tooling.
What to watch next. Adoption metrics during preview, the rate at which prebuilt workflows in GitHub Next's agentics repository get forked, and whether competitors — GitLab, Atlassian, and the standalone coding-agent vendors — respond with comparable native integrations. The economics of organization-billed agent runs at scale, and how cost centers hold up against runaway loops, will determine whether this graduates from preview into default infrastructure.
Sources:
- GitHub Agentic Workflows is now in public preview — GitHub Changelog
- Agentic workflows no longer need a personal access token — GitHub Changelog
- Bot-created pull requests can run workflows if approved — GitHub Changelog
Three different layers of the AI economy moved in concert today. Public markets absorbed a $1.77 trillion bet that one operator's judgment is worth surrendering shareholder oversight; private capital extended an $18 billion thesis that the next defensible AI moats lie in atoms rather than tokens; and the tooling layer quietly made agents a first-class citizen in how code reaches production. The common thread is concentration — of capital, of control, and of the infrastructure on which the rest of the industry will increasingly depend.