One-Line Summary

Microsoft moves to own more of its AI stack post-OpenAI deal revision, Salesforce turns Slack into an AI coworker, and private markets pile into Anthropic while SpaceX’s IPO could drain liquidity.

Big Tech

Microsoft Is on a New AI Journey After Reworked OpenAI Deal

Microsoft, best known for Windows, Office, Azure, and the Copilot suite, is recalibrating its AI strategy to be more self-sufficient after a reworked arrangement with OpenAI. Under AI CEO Mustafa Suleyman, the company signals a push to control more of the AI stack—models, data, and distribution—reducing single-partner risk while keeping strategic ties. Think of it as moving from renting a power plant to gradually building your own, without turning off the current lights. ¹

Why shift now? Regulatory scrutiny, supply chain constraints in compute, and rising competition make dependency expensive. By investing in internal research and diversified model access, Microsoft can tune performance, cost, and safety for its enterprise base—where privacy, compliance, and uptime are non-negotiable. This also strengthens Azure’s pitch as the neutral, enterprise-grade AI cloud. ¹

For customers, this likely shows up as more tightly integrated Copilot features across Microsoft 365, Dynamics, and Azure, with clearer governance and cost controls. Expect more multi-model orchestration behind the scenes, giving you better accuracy and resilience—without you having to switch tools. ¹

The bottom line: Microsoft is de-risking its AI future by broadening capabilities beyond a single partner while doubling down on enterprise distribution. If you’re building on Azure or Copilot, expect steadier roadmaps and more knobs to manage cost, latency, and data boundaries. ¹

Salesforce Rolls Out a Major AI Upgrade For Slack

Salesforce, the CRM giant that owns Slack, rolls out 30+ new AI capabilities for Slackbot, positioning it as a practical “AI coworker.” The assistant can read messages, channels, and files, connect to Salesforce records, and trigger actions—like pulling client data when a name is mentioned or summarizing meetings—so work moves from chat to execution in one place. ²

Under the hood, Slackbot adds reusable “AI skills” and integrates via Model Context Protocol (MCP) and Salesforce’s Agentforce. For teams, this means fewer tabs and faster follow-through: think “create a Q2 renewal forecast” and get a draft with linked Salesforce objects, next steps, and a meeting invite. It reframes Slack from a conversation pit into an operational surface. ³

Why it matters: if your company already runs on Slack + Salesforce, these features collapse context switching and standardize workflows across sales, service, and ops. Early coverage frames this as a “fleet” of agents inside Slack—useful so long as admins set access boundaries and audit trails for compliance. ⁴

OpenAI Calls for Grid Upgrades and Stronger Safety Nets

OpenAI publishes policy recommendations for the AI era, urging governments to accelerate electric grid buildouts, create fast-response social safety nets, and even consider a public wealth fund to share gains from AI productivity. It also flags prospective governance for superintelligence. Translation: AI won’t just need GPUs; it needs power, people protections, and policy. ⁵

This is notable because it ties AI scaling to real-world infrastructure and labor markets. Grid delays already slow AI data center timelines; safety nets can cushion job transitions as software automates tasks faster than organizations can retrain. Think of it as prepping roads and guardrails before rolling out faster cars. ⁶

For enterprises, expect more scrutiny on where compute runs and how workforce plans adapt. Procurement teams may face new ESG-style checklists around energy sourcing and job impact plans as regulators turn guidance into rules. ⁵

Industry & Biz

Anthropic Heats Up in Private Markets; SpaceX Could Soak Up IPO Cash

Secondary-market demand for Anthropic shares turns “almost insatiable,” with one broker citing buyers lining up roughly USD 2 billion, while around USD 600 million of OpenAI shares look for takers at a discount to its latest primary valuation. The vibe: investors want exposure to multiple frontier labs, but momentum favors Anthropic right now. ⁷

OpenAI is trying to police secondary access via authorized bank channels, reportedly with fewer fees, while banks charge carry for Anthropic exposure—an indicator of demand. But a new variable arrives: SpaceX’s confidential IPO filing, potentially raising USD 50–75 billion at a USD 1.75 trillion valuation, could absorb a lot of public-market liquidity. ⁸

If SpaceX goes first, others may face tighter checkbooks and tougher questions. For founders considering 2026 listings, sequencing matters: follow a blockbuster, and you compete with excitement and investor fatigue; go first, and you price-test sentiment. Either way, AI issuers will need crisp narratives on unit economics, safety, and model moat. ⁹

Spain’s Xoople Raises $130M to Map Earth for AI

Xoople secures a USD 130 million Series B and a deal with L3Harris to build spacecraft sensors, aiming to create a high-fidelity Earth map purpose-built for AI training and inference. Better geospatial data means better downstream AI for logistics, climate, and autonomy—think “Google Maps for machine perception,” refreshed more often. ¹⁰

Space-data infrastructure is having a moment. Separately, the “orbital data centers” narrative is gaining airtime as a way to skirt Earth-side siting battles—more sizzle than steak for now, but it signals how compute scarcity and NIMBY pushback shape capex plans. Expect partnerships across launch, sensors, and edge-AI to accelerate. ¹¹

Early-stage rounds underscore the trend: Antaris closes USD 28 million to expand AI-driven satellite mission design and autonomous operations, while India’s Xovian Aerospace raises USD 2 million to build an AI-native RF intelligence constellation. The mosaic is clear—data, from RF to optical, is the new AI fuel, and satellites are the pumps. ^{12 13}

What This Means for You

• If you run on Microsoft 365 or Azure, expect steadier Copilot roadmaps and more enterprise-grade controls as Microsoft diversifies model providers and builds more in-house. That means better knobs on cost, latency, and privacy without ripping out existing workflows. ¹

• Slack-first teams can turn chatter into actions with Slack’s new AI skills—auto-summaries, CRM lookups, and meeting follow-through. The productivity unlock comes from wiring AI to your systems of record, but you’ll need admin guardrails on data access, logging, and approvals. ^{2 3}

• For finance and founders: Anthropic’s momentum shows investor appetite for model diversification; SpaceX’s IPO timing reminds us that liquidity is finite. If you’re planning a raise or exit, align your timelines with market absorption and sharpen your margins-and-moat story. ⁷

• Data is the defensible edge: Xoople, Antaris, and Xovian point to a world where proprietary geospatial and RF streams power AI advantage in logistics, defense, climate, and insurance. If you’re buying AI, ask vendors what unique data they own—not just which model they call. ^{10 12}

Action Items

1. Pilot Slack’s new AI skills with real workflows: Pick one sales or support use case (e.g., renewal prep), connect Slack to Salesforce, and measure time-to-action before/after.
2. Tune Copilot for cost and compliance: In Microsoft 365, review data access scopes and usage analytics; set prompts, retention, and DLP policies for your top three teams.
3. Audit your AI data moat: List unique data you own (logs, geospatial, RF, transactions), and run a 2-week POC to improve one KPI (forecast accuracy, SLA compliance) using retrieval or fine-tuning.
4. Scenario-plan funding around SpaceX’s IPO window: If you’re raising in 1–2 quarters, build alternate timelines and investor lists assuming tighter liquidity post-IPO.