The Device Layer

Apple opened WWDC today by announcing that Siri has been rebuilt on Google Gemini — a custom 1.2 trillion-parameter model licensed at approximately $1 billion per year. iOS 27 ships to 1.4 billion active iPhones. Users can choose between Gemini, ChatGPT, and Claude as their AI engine. AI is no longer a tool employees choose to adopt — it is now a layer in the operating system they already carry. The enterprise AI adoption question just changed character entirely.

In this edition: Apple WWDC — Siri on Gemini, 1.4 billion devices, model choice  ·  Colorado AI Act enforcement killed: first U.S. AI deadline delayed and scaled back  ·  SoftBank commits €75 billion to French AI data centers  ·  OpenAI workspace agents reach enterprise globally

When AI Becomes the OS

Apple announced at WWDC 2026 today that Siri has been rebuilt on a custom version of Google Gemini — a 1.2 trillion-parameter model licensed at roughly $1 billion per year. iOS 27 will ship this fall to 1.4 billion active iPhones. Users will be able to choose between Gemini, ChatGPT, and Claude as their AI engine. When iOS 27 ships, enterprise AI adoption will not be a question of whether employees will use AI. It will be a question of whether the AI they are already using is connected to anything that makes it useful for the business.

The rebuilt Siri functions as a genuine AI assistant: persistent conversation history, contextual memory across apps, multi-step reasoning, web search, content summarization, image generation, and file analysis. Apple is processing all Gemini queries through its Private Cloud Compute infrastructure — model weights inside Apple's hardware-isolated enclaves, data not retained after processing, nothing shared with Google. For enterprise security teams in regulated industries, this architecture matters as much as the capability.

The model choice feature is new for Apple and significant for enterprise strategy. Users can select Gemini as the default, or switch to ChatGPT or Claude depending on the task. Apple is treating AI models as interchangeable infrastructure, not as differentiated products. If 1.4 billion iPhone users come to expect model portability by default, enterprise employees will expect the same from internal AI tools. The lock-in strategies that AI vendors have been building since 2023 are being quietly eroded by the world's most valuable consumer technology company.

Apple's choice of Google Gemini over OpenAI or its own models carries its own signal. Google won the default position in the world's most premium consumer device not because of model benchmarks but because of the commercial, infrastructure, and partnership terms Apple required. The enterprise implication that matters most is not the AI features Apple shipped today — it is the question those features make urgent: when every employee carries an AI assistant by default, what does your enterprise need to have built so that AI is connected to your systems and data, rather than running alongside them as a general-purpose chatbot?

Moving Pieces

Three developments that matter to enterprise leaders this week

Colorado AI Act: The First U.S. Enforcement Deadline Was Quietly Killed

The June 30, 2026 enforcement date for the Colorado Consumer Protection for Artificial Intelligence Act will not happen. A federal magistrate judge stayed the original law on April 27, and Governor Polis signed a replacement bill — SB 26-189 — on May 14, delaying enforcement to January 1, 2027 and drastically scaling back requirements. The replacement drops risk management programs, annual algorithmic impact assessments, and algorithmic discrimination duties in favor of a narrower transparency and notice framework. An ongoing federal injunction — triggered by xAI's legal challenge — also extends to the successor legislation, meaning even the scaled-back version cannot be enforced until rulemaking is complete. This is a material counter-signal to last Friday's account of 2,081 state AI bills: the first enforcement deadline in U.S. history just became a disclosure framework scheduled for 2027. State AI legislation is moving, but enforcement is considerably softer than the bill counts suggest.

SoftBank Commits €75 Billion to Build 5 GW of AI Data Centers in France

SoftBank announced at the Choose France investment summit that it will develop and operate 5 gigawatts of AI data center capacity in France, representing up to €75 billion in investment. The first phase commits €45 billion to deliver 3.1 GW in the Hauts-de-France region by 2031, with sites in Dunkirk, Bosquel, and Bouchain. SoftBank is partnering with EDF on energy and Schneider Electric on manufacturing infrastructure. This is the largest single AI infrastructure commitment in European history. The strategic context: European AI infrastructure has significantly lagged U.S. and Asian capacity, and EU regulatory requirements increasingly favor data residency within European jurisdiction. SoftBank's bet is that European AI demand — for cloud inference and sovereign compliance alike — will justify infrastructure at U.S.-comparable scale. For enterprise AI teams with European data obligations, the capacity expansion changes the economics of compliant AI inference.

OpenAI Brings Workspace Agents and ChatGPT for Spreadsheets to Enterprise Globally

OpenAI expanded two enterprise capabilities globally this week. Workspace agents — shared AI agents that teams configure to work across tools — are now available across Business, Enterprise, and Edu plans, with admin controls for monitoring activity and usage visibility in the admin console. The free period for workspace agents runs through July 6, with credit-based pricing beginning after that date. Separately, ChatGPT for Excel and Google Sheets is now globally available for Enterprise, Edu, and K-12 workspaces, bringing a native ChatGPT sidebar for building, updating, and reviewing spreadsheets. The spreadsheet integration is operationally significant: Excel and Google Sheets are the most widely deployed data tools in enterprise environments, and a native sidebar eliminates the copy-paste workflow that has characterized most enterprise AI use to date. Enterprise revenue now accounts for more than 40% of OpenAI's total revenue, on track for parity with consumer by year-end.

iPhones that will run Google Gemini by default when iOS 27 ships — the largest single AI distribution event in history.

No AI product has ever reached 1.4 billion users simultaneously. ChatGPT reached 500 million weekly users across two and a half years of remarkable growth. When iOS 27 ships, Google Gemini arrives on every modern iPhone at once — not as an app users download, but as the default AI assistant in the operating system. For enterprise leaders, the number matters not because of what Apple shipped but because of what it makes unavoidable: every employee with a company iPhone will have AI embedded in their operating system by the end of 2026. The question is not whether they will use it. It is what they will use it on.

Sources: MacRumors / WWDC 2026  ·  Enterprise DNA  ·  June 8, 2026

The Adoption Question Is Over

Enterprise AI teams have spent two years working on adoption. Awareness campaigns, lunch-and-learns, AI champions, usage metrics. iOS 27 retires the question.

When Gemini-powered Siri ships on 1.4 billion iPhones, adoption is no longer a change management problem. It is a given. The employees who were not using AI will be using AI before the year is out — not because your organization convinced them, but because their phone came with it. The question that replaces it is harder and more valuable: what does your organization need to have built so that the AI your employees are already using is actually useful for the business?

That is an integration question. It requires knowing which systems hold the data that matters — the CRM, the ERP, the document repositories, the proprietary knowledge bases — and building the connectors, permissions, and governance frameworks that allow AI to reach that data without creating security and compliance exposure. Most enterprise AI programs have not built this. They have built procurement processes, usage policies, and awareness campaigns. The infrastructure layer is still pending for the majority of organizations.

iOS 27 is not a problem. It is a forcing function. The organizations that build the integration layer before the update ships will be the ones who turn ambient AI into operational advantage. The ones that wait will have a workforce running a very capable general-purpose AI against very general-purpose knowledge while their competitors' AI is running against the actual business.

Build the integration layer. That is the work.

AK  /  Spearhead  /  Building AI systems, not tools