Digital Colliers Daily Briefing — May 15, 2026
Capital is flowing into AI infrastructure at scales that would have seemed implausible eighteen months ago. Today's briefing covers three data points that together describe the current shape of the market: a successful public debut for an AI chip company that spent two years stuck in regulatory limbo, a private fundraise that values one model lab at nearly a trillion dollars, and a product update that hardens the front line in the agentic coding war. Each event, on its own, would lead a normal news cycle. They landed on the same Thursday.
1. Cerebras opens at $385, closes the year's biggest tech IPO at a $66B valuation
What happened. Cerebras Systems priced its IPO Wednesday evening at $185 per share — well above the revised $150–$160 range, which itself had been raised from an initial $115–$125 — and sold 30 million shares to raise $5.5 billion. According to TechCrunch, the stock opened Thursday at $385, up 108%, before settling at $311.07 by the close, a 68% first-day gain that Bloomberg pegged at a $67 billion market capitalization. Yahoo Finance valued the company at $66 billion at the bell, with shares ticking up again in after-hours trading. CEO Andrew Feldman's stake closed worth roughly $3.2 billion and CTO Sean Lie's at $1.7 billion, per CNBC reporting cited by Techmeme — stakes of about 5.5% and 3%, respectively.
Why it matters. Cerebras had filed to go public in 2024 and was effectively frozen by a CFIUS review of its concentrated revenue relationship with Abu Dhabi's G42. The successful relisting hinges on a transformed financial profile: $510 million in 2025 revenue (up 76% year-over-year) and a swing to $237.8 million in net income from a near-half-billion loss the year prior, according to TechCrunch. That makes Cerebras the first credible public-market reference point for an Nvidia alternative at scale, and the first major tech IPO of 2026.
Who is affected. Public investors now have a liquid AI-silicon name beyond Nvidia and AMD, which is likely to pull capital toward other accelerator startups still private. Cerebras customers — OpenAI (via what TechCrunch describes as a "complicated circular-deal relationship"), G42, MBZUAI, and AWS — gain a vendor with a strengthened balance sheet for capacity expansion. Bankers eyeing the rest of the 2026 IPO pipeline get the demand signal they have been waiting for.
What to watch next. Customer concentration remains the structural risk; G42 still accounts for a disproportionate share of revenue. Watch for lockup expirations, the cadence of new inference customer announcements, and whether Cerebras' wafer-scale architecture wins design slots against Nvidia's Blackwell-successor roadmap as hyperscalers finalize 2027 capex.
Sources:
- Cerebras raises $5.5B, then stock pops $108%, in the first huge tech IPO of 2026 — TechCrunch AI
- Cerebras co-founders CEO Andrew Feldman and CTO Sean Lie own stakes worth $3.2B and $1.7B, respectively, representing about 5.5% and 3% of the company (CJ Haddad/CNBC) — Techmeme
- Cerebras shares closed up 68% at $311.07 in the company's Nasdaq debut after it raised $5.5B in the year's largest IPO, giving it a market value of $67B (Bloomberg) — Techmeme
2. Anthropic locks in $30B at a $900B valuation, narrowing the AI race to two
What happened. Anthropic has agreed to terms on a $30 billion fundraise at a $900 billion valuation, with Sequoia Capital, Dragoneer, Greenoaks, and Altimeter Capital co-leading the round, according to the Financial Times' George Hammond as reported by Techmeme. The deal is among the largest private financings on record by either headline figure.
Why it matters. A $900 billion mark sets Anthropic alongside OpenAI as the only two AI labs operating at a valuation tier formerly reserved for trillion-dollar public incumbents. The $30 billion check is, in effect, a compute commitment: at current GPU pricing, the round funds multiple gigawatts of training and inference capacity, and it locks Anthropic into the same hyperscale spending cadence as OpenAI. It also crystallizes a duopoly framing for frontier model development, with Google DeepMind, Meta, and xAI now competing against private valuations that approach the market caps of S&P 100 companies.
Who is affected. Enterprise buyers face a market in which the two leading frontier vendors each carry pricing power commensurate with near-monopoly capital backing — relevant for any organization negotiating multi-year API or seat commitments. Talent markets tighten further; compensation benchmarks at this valuation reset the floor for senior research hires across the industry. Sequoia, Dragoneer, Greenoaks, and Altimeter are now exposed to a single position whose mark-to-market swings will move their entire fund performance. Cloud partners — chiefly AWS and Google Cloud — gain renewed multi-year commitments that further entrench their AI infrastructure positions.
What to watch next. Confirmation of final round terms, secondary components, and any governance or board changes; the cadence at which Anthropic translates the capital into Claude model releases and Claude Code feature velocity; and how the round affects OpenAI's own next financing, which will inevitably be benchmarked against it.
Sources:
3. OpenAI puts Codex in the ChatGPT mobile app, escalating the agentic coding fight
What happened. OpenAI rolled Codex into the ChatGPT mobile app on iOS and Android, in preview across all plans including Free and Go. According to OpenAI's announcement, more than four million people now use Codex weekly. The mobile experience connects to live Codex sessions running on laptops, Mac minis, devboxes, or managed remote environments via a secure relay layer, streaming back screenshots, terminal output, diffs, test results, and approval prompts. Alongside the mobile launch, OpenAI made Remote SSH and Hooks generally available, introduced programmatic access tokens for Enterprise and Business plans, and added HIPAA-compliant use of Codex in local environments for eligible Enterprise workspaces.
Why it matters. The release reframes Codex from a desktop tool into a persistent, multi-device agent — a direct answer to Anthropic's Remote Control feature for Claude Code, released in February. As The Verge notes, OpenAI has been visibly cutting "side quests" to concentrate engineering and enterprise muscle on this fight; Codex's macOS app gained background execution last month and a Chrome extension earlier this month. The competitive context matters: Claude Code has been gaining share among professional developers throughout the year, and Microsoft's recent move to wind down Claude Code licensing for some internal cohorts signals how strategic the category has become.
Who is affected. Working developers are the immediate audience; the mobile interface targets the dead time — commutes, meetings, coffee runs — during which long-running agent tasks previously sat idle awaiting approval. Enterprise buyers gain HIPAA-eligible local deployments and programmatic tokens for CI/CD integration, removing two common procurement blockers. The customer testimony OpenAI published from Sea Limited frames the broader trajectory: David Chen, co-founder of Sea and CPO of Shopee, reports 87% weekly active usage among Codex users at the company, with developers shifting from "passive autocomplete" to agents operating inside CI/CD pipelines. Chen's framing — the developer evolving into a "system orchestrator" — captures the bet both OpenAI and Anthropic are making about engineering org design.
What to watch next. Windows support for the mobile-to-desktop relay (promised but not dated), enterprise pricing changes once the preview converts to general availability, and benchmark data on whether mobile-initiated sessions actually move task throughput. The competitive tell will be Anthropic's response — likely a Claude Code feature push timed to the closing of its $30 billion round.
Sources:
- Work with Codex from anywhere — OpenAI Blog
- OpenAI says Codex is coming to your phone — TechCrunch AI
- Now in preview: Codex in the ChatGPT mobile app. — YouTube · OpenAI
- [HN · 334↑] Codex is now in the ChatGPT mobile app — Hacker News
- OpenAI's Codex is now in the ChatGPT mobile app — The Verge AI
- Sea's View on the Future of Agentic Software Development with Codex — OpenAI Blog
The three stories trace a single capital flow. Public investors bid a hardware supplier to $66 billion on the prospect of inference demand; private investors price the most prominent consumer of that inference capacity at $900 billion; and the products that ultimately generate the revenue — agentic coding tools used by millions of developers weekly — get the feature work that justifies both valuations. The bottleneck across all three is the same: compute, and the talent and silicon required to expand it. Whether the math in tiers two and three eventually supports the equity raised in tier one is the question the rest of 2026 will answer.