top of page

LATEST INSIGHTS

Expert insights across your SaaS environment

Blogs and papers from the SAASiQ team on enterprise SaaS, cloud and AI.

Anthropic and OpenAI File for IPOs, SpaceX Lists, and Microsoft Launches Its Own AI Models

Writer: SAASiQ.ai
SAASiQ.ai
Jun 15
6 min read

Updated: 6 days ago

Title: Anthropic and OpenAI File for IPOs, SpaceX Lists, and Microsoft Launches Its Own AI Models

Date: 15 June 2026

Type: Blog

Author: SAASiQ (contact@saasiq.ai)

Word count: 1414 words

Reading time: 6 min

Published: 15-06-2026


Anthropic confidentially submitted a draft IPO registration to the US Securities and Exchange Commission on 1 June, and OpenAI said on 8 June that it had done the same. On 12 June SpaceX, which has owned xAI since February, began trading on Nasdaq after the largest IPO on record. Microsoft used its Build conference on 2 June to release its first family of models trained without OpenAI, and Oracle reported a year-end order book of $638 billion, most of its recent growth coming from AI contracts.


What Anthropic filed

Anthropic said on 1 June that it had submitted a draft registration statement on Form S-1, which gives it the option to go public once the SEC has finished its review, and neither the number of shares nor the price had been set. A confidential submission does not commit a company to a date, but US rules require the registration statement to be filed publicly at least 15 days before the roadshow, so the audited figures will be published before any shares are sold.


The filing came four days after Anthropic closed a $65 billion Series H on 28 May at a post-money valuation of $965 billion. Altimeter Capital, Dragoneer, Greenoaks and Sequoia Capital led the round, and it included $15 billion of previously committed investment from cloud providers, $5 billion of it from Amazon. Anthropic said its run-rate revenue had crossed $47 billion earlier in May. TechCrunch reported that the figure was $9 billion at the end of 2025. In February, Anthropic's Series G had raised $30 billion at a valuation of $380 billion.


Bloomberg reported on 3 June that Anthropic had chosen Morgan Stanley and Goldman Sachs to lead the offering, with JPMorgan also involved, and that it was considering going public as soon as October. That is the reported earliest date, and Anthropic itself has not given one.


OpenAI follows a week later

OpenAI announced its own confidential draft S-1 on 8 June. It said it had not decided on timing and that listing might be a while off, because some of what it wanted to do was easier as a private company.


OpenAI was last valued at $852 billion post-money, when it closed a $122 billion round on 31 March. Bloomberg reported that Amazon agreed to put in $50 billion of that, and Nvidia and SoftBank $30 billion each, with $35 billion of Amazon's share contingent on OpenAI going public or reaching artificial general intelligence. So Anthropic now carries the higher private valuation of the two, and it filed first.


SpaceX lists, with xAI inside it

SpaceX priced its offering on 11 June at $135 a share, selling 555.6 million shares to raise $75 billion and valuing the company at about $1.77 trillion. The previous record was Saudi Aramco's $24.9 billion in 2019. The shares opened at $150 on 12 June and closed the day at $160.95, 19 per cent above the offer price.


SpaceX merged with xAI, the maker of Grok, in February, so its prospectus carries the first audited numbers for a frontier AI developer. TechCrunch's reading of the S-1 on 20 May showed xAI with an operating loss of $6.4 billion on revenue of $3.2 billion in 2025, and capital spending of $12.7 billion in 2025 followed by $7.7 billion in the first quarter of 2026 alone.


The filings also show who is renting that capacity. Anthropic is paying SpaceX $1.25 billion a month through May 2029 for the compute at Colossus 1, xAI's data centre near Memphis, and a filing on 5 June showed Google agreeing to pay SpaceX $920 million a month from October 2026 for about 110,000 Nvidia GPUs.


Microsoft's own model family

At Build on 2 June, Microsoft announced seven models under its MAI name, trained from scratch without distillation from OpenAI or any other outside model. The largest, MAI-Thinking-1, is a mixture-of-experts reasoning model with 35 billion active parameters and a 256,000-token context window, in private preview in Microsoft Foundry. Microsoft says it scores 53 per cent on SWE-Bench Pro, alongside Anthropic's Opus 4.6, and that independent raters preferred it to Sonnet 4.6. Those are Microsoft's own figures.


The coding model, MAI-Code-1-Flash, has 5 billion parameters and scores 51 per cent on SWE-Bench Pro, according to Microsoft, and is available in GitHub Copilot and rolling out as one of the default models in VS Code. The rest of the family covers images (MAI-Image-2.5 and a Flash version, already live in PowerPoint), speech recognition in 43 languages (MAI-Transcribe-1.5, built into Copilot, Teams, GitHub and Dynamics 365 Contact Center) and voice (MAI-Voice-2 and a Flash version). The models are also sold through Fireworks AI, Baseten and OpenRouter.


The contract behind this changed on 27 April. Microsoft's licence to OpenAI's models and products now runs to 2032 but is no longer exclusive, OpenAI can sell through any cloud provider, and Microsoft has stopped paying OpenAI a revenue share, while OpenAI's payments to Microsoft continue to 2030 under a cap. Microsoft remains a major shareholder, with about 27 per cent after OpenAI's restructuring in October 2025.


For Microsoft 365 customers the practical point is that the model behind a given Copilot feature is Microsoft's choice and can change in a product update. MAI-Transcribe-1.5 is already built into Teams, so any AI policy that names an approved model provider needs a way of finding out which models sit behind the features in use.


Google's agent tools

Google released Gemini 3.5 Flash at I/O on 19 May and made it generally available the same day, in the Gemini app, the Gemini API, Gemini Enterprise and the Gemini Enterprise Agent Platform. Google says it outperforms Gemini 3.1 Pro on coding and agentic benchmarks, with 76.2 per cent on Terminal-Bench 2.1 and 83.6 per cent on MCP Atlas, a test of tool use. In Gemini Enterprise the model was switched on by default in all regions on 9 June, with the option to turn it off removed. Google said at I/O that it expected to roll out Gemini 3.5 Pro in June.


Managed Agents, launched in preview in the Gemini API on 19 May, provisions a remote Linux environment in which an agent can plan, call tools, run code and browse the web, with the agent's instructions written in AGENTS.md and SKILL.md files. It runs on Gemini 3.5 Flash and is also offered through the Agent Platform. Since 28 April, Gemini Enterprise has accepted custom servers built on the Model Context Protocol, the standard for connecting models to tools and data that Anthropic published in November 2024.


Google also announced Gemini Omni, a model that generates from any input, starting with video, and began rolling out Gemini Omni Flash to AI Plus, Pro and Ultra subscribers through the Gemini app and Google Flow.


Oracle's year-end numbers

Oracle reported its fiscal 2026 fourth quarter on 10 June. Revenue was $19.2 billion, up 21 per cent, with cloud infrastructure up 93 per cent to $5.8 billion and cloud applications, which include Fusion and NetSuite, up 10 per cent to $4.1 billion. Remaining performance obligations, the value of contracted revenue not yet recognised, reached $638 billion, $85 billion more than at the end of the third quarter.


Oracle said most of the increase in the third and fourth quarters came from large AI contracts where the customer either prepaid for the GPUs or supplied them, and that those prepaid and customer-supplied portions now total $75 billion. Capital spending for the year was $55.7 billion and free cash flow was negative $23.7 billion. Oracle expects total revenue to grow 27 to 29 per cent in fiscal 2027.


The suspension and the filings

Anthropic released Claude Fable 5 on 9 June. At 5:21pm Eastern on 12 June it received a directive from the US Commerce Department suspending all access to Fable 5 and Mythos 5 by foreign nationals, inside or outside the United States, and because it could not check nationality in real time it disabled both models for every customer. Its other models, including Opus 4.8, were not affected, and Fable 5 was still unavailable on 15 June.


SAASiQ's view is that the S-1s, once public, will be useful reading for anyone buying these services, because a prospectus has to disclose material risks and commitments. The SpaceX filing set out xAI's losses and compute contracts in exactly that detail.


Bloomberg reports that Anthropic could list as soon as October, OpenAI has set no timetable, and Google has said Gemini 3.5 Pro is due this month.

SAASiQ - Intelligent Solutions for SaaS ©

Optimise your SaaS licences and software subscriptions with SAASiQ

bottom of page