It is Sunday, June 14, 2026. The AI industry does not take weekends off and neither does this roundup. SpaceX landed on Wall Street two days ago and the numbers were something. OpenAI bought a talk show. Novo Nordisk handed its drug pipeline to an AI company. And Google’s most anticipated model of the year is still sitting in the waiting room while the clock on Sundar Pichai’s promise ticks down. Here is everything that happened and why it matters.
SpaceX went public on Friday and closed up 19 percent on the largest IPO in history
It actually happened. SpaceX listed on Nasdaq under the ticker SPCX on Friday, June 12, raising $75 billion at $135 per share in what is officially the largest initial public offering in recorded history, surpassing Saudi Aramco’s $29 billion raise in 2019 by a significant margin. Shares opened at $150, peaked at $176.52 intraday, and closed at $160.95 — a 19 percent gain on the IPO price in a single session. The company ended its first day as a public company valued above $2 trillion.
A few numbers worth sitting with. Elon Musk holds approximately 42 percent of SpaceX’s equity through a dual-class structure that gives him roughly 82 percent of voting control via Class B shares. At the Friday close, his SpaceX stake alone was valued at $869 billion. He did not sell a single share. The company lost $8.7 billion between January 2025 and March 2026. The profitable segment is Starlink broadband, which generated $11.4 billion in revenue and $4.4 billion in operating income in 2025. The xAI division — which came with the SpaceX acquisition of xAI earlier this year — burned $14 billion in cash against $3.2 billion in revenue. Investors bought the future. The present is Starlink subsidizing everything else.
For the AI industry, the SpaceX debut matters as a benchmark. Anthropic and OpenAI are both in the IPO queue. If institutional appetite holds at these levels, both companies enter their roadshows with a favorable comparable. Goldman Sachs CEO David Solomon told clients this week that markets are in what he described as greed mode. That is not normally how Goldman describes a healthy market.
Read the coverage: NPR on the record debut and what it means for the market | CNBC live coverage of the first trading day | CNN on the debut and analyst reactions
OpenAI bought a Silicon Valley talk show and called it a communications strategy
In April, OpenAI acquired TBPN — the Technology Business Programming Network — a daily three-hour livestream talk show hosted by former tech founders John Coogan and Jordi Hays. The show airs weekdays from 11am to 2pm Pacific, averages 70,000 viewers per episode, and was on track to generate more than $30 million in revenue this year before the deal. OpenAI paid a reported amount in the low hundreds of millions of dollars. Sam Altman called it his favorite tech show on X. The New York Times had recently described TBPN as Silicon Valley’s newest obsession.
TBPN will continue to operate under its own brand with promised editorial independence. The hosts say they can still say whatever they want because the show is live. The show now reports to Chris Lehane, OpenAI’s chief global affairs officer — the same person who invented the phrase “vast right-wing conspiracy” as a political communications tool in the 1990s. The team sits inside OpenAI’s strategy organization.
The honest read: OpenAI is heading into an IPO roadshow and wants friendly air cover in the media environment it operates in. TBPN is where the CEOs of Meta, Microsoft, Salesforce, and OpenAI itself come to talk candidly with insider hosts. Buying that platform months before a trillion-dollar public offering is not a coincidence. Altman said he does not expect the hosts to go easier on OpenAI as a result of the acquisition. That may be true. It is also unprovable by design.
Read the coverage: TechCrunch on the deal and what it signals about OpenAI’s ambitions | OpenAI’s official announcement | Variety on the editorial independence question
Novo Nordisk handed its entire drug pipeline to OpenAI and told its workforce to learn AI or else
In April, Novo Nordisk — the Danish pharmaceutical company behind Ozempic and Wegovy — announced a sweeping strategic partnership with OpenAI to apply AI across drug discovery, manufacturing, supply chain, and commercial operations, with full integration across the company targeted by the end of 2026. This is not a pilot. It is a company-wide transformation with a hard deadline.
The specific application: OpenAI’s models will analyze complex datasets, identify drug candidates earlier in the research process, and shorten the time between a compound showing promise in a lab and a patient receiving it. The deal also includes a workforce upskilling commitment — OpenAI will train Novo’s global organization in AI tools, which is a polite way of saying the company is restructuring around AI fluency as a baseline job requirement.
The competitive context makes this easier to understand. Novo Nordisk is losing ground in the GLP-1 weight-loss drug market to Eli Lilly, which has signed 16 AI partnerships since 2025 totaling billions of dollars, including a $2.75 billion deal with AI drug discovery firm Insilico Medicine. The race to find the next Ozempic is now substantially an AI race, and Novo is betting that OpenAI’s general-purpose model capabilities can close the gap against rivals using more specialized biotech AI tools.
Read the coverage: CNBC on the partnership details and competitive stakes | Quartz on the drug discovery race against Eli Lilly | BioPharma International on the governance and data protection framework
Google’s flagship model is still not out and June is running out of days
At Google I/O on May 19, Sundar Pichai announced Gemini 3.5 Pro would ship the following month. The audience reportedly groaned. That was 26 days ago. It is now June 14. June 30 is 16 days away. Gemini 3.5 Pro has not shipped.
What is confirmed: the model targets a two-million-token context window, a Deep Think reasoning mode, and frontier multimodal capability across text, images, and other formats. It is designed to absorb the use cases Google previously routed to Gemini Ultra — the hardest reasoning tasks, complex long-context work, and deep agentic workflows. Pricing is expected in the range of $2 to $4 per million input tokens and $12 to $25 per million output tokens, putting it in direct competition with Claude Opus 4.8 and GPT-5.5.
The model is currently in internal use and limited Vertex AI enterprise preview. Prediction markets at Polymarket have traders pricing a late June release, with a small but nonzero probability on the model missing the month entirely. The irony worth noting: Apple announced last week that its new Siri runs on a $1 billion-per-year Gemini license, making Google’s model the default assistant on two billion iPhones starting this fall. The model that will power all of that has not finished shipping yet.
Read the coverage: TechTimes on the confirmed specs and what Deep Think reasoning actually means | Polymarket odds on the release window | The realistic breakdown of dates, pricing, and how it stacks up against Claude Opus 4.8 and GPT-5.5
Also worth reading this weekend
- Researchers gave top AI models a classic psychology attention test and found a significant flaw. The models performed well on short tasks but deteriorated sharply as complexity increased. The study raises questions about how AI performs on extended, real-world reasoning chains rather than benchmark-length prompts. (ScienceDaily, June 10)
- OpenAI is reportedly backing away from fully autonomous AI research by 2028, reframing its vision as a “tandem” between humans and machines rather than a handoff to AI. CEO Sam Altman and chief scientist Jakub Pachocki are also calling for an international body with authority to slow frontier AI development if necessary. (LLM Stats news feed)
- Taiwan is considering restricting AI chip sales to all Chinese customers, not just those on US blacklists, to align with American export controls. If implemented, it would represent the most significant tightening of the semiconductor supply chain to Chinese AI developers since the original NVIDIA export restrictions in 2023. (LLM Stats news feed)
That is your Sunday AI update. SpaceX is public, OpenAI owns a talk show it promises it will not interfere with, Novo Nordisk is letting an AI find the next blockbuster drug, and Google still owes the world its best model of the year. See you Monday.