2026-04-17
By Vadym · Generated with AI, curated by me
• TL;DR --> TL;DR This Week
• OpenAI hits $25B ARR and signals public listing as early as late 2026
• Google open-sources ADK — official framework for building multi-agent systems
• OpenAI acquires Hiro, an AI personal finance startup — app ecosystem play begins
• Anthropic extends Microsoft 365 connector to all Claude users
• Gizmo AI learning app: 13M users, $22M Series A — edtech AI goes mainstream
OpenAI surpassed $25 billion in annualized revenue this week and is taking active steps toward a public listing — reportedly as early as late 2026. The company is also projecting advertising revenue of $2.5 billion for 2026, with a target of $100 billion annually by 2030. For context: OpenAI crossed $1B ARR in late 2023. It has 25x’d in under three years. No software company has ever scaled this fast.
Why it matters: This is no longer a technology story — it’s a market structure story. A $25B ARR business pursuing IPO at an $800B+ valuation will reshape how capital flows across the entire AI sector. Every competitor now benchmarks against these numbers. Every investor now has to decide whether OpenAI’s moat is as wide as its revenue suggests.
OpenAI acquired Hiro, an AI-powered personal finance startup, marking its most consumer-facing acquisition to date. The deal is less about the product than the direction: OpenAI is assembling a portfolio of domain-specific applications to extend its reach beyond the API layer. This is the playbook Google ran with Android apps in 2010 — own the platform, then populate it. Source →
Google released the Agent Development Kit (ADK) as open source — a Python framework for building, orchestrating, and deploying multi-agent AI systems. ADK supports both Google models and third-party LLMs, integrates natively with Vertex AI, and includes built-in evaluation tooling. It’s Google’s most direct answer to Microsoft’s Semantic Kernel and the growing ecosystem of third-party agent frameworks. GitHub →
Anthropic expanded its Microsoft 365 integration to the full Claude user base, enabling search and reasoning across Outlook, Teams, SharePoint, OneDrive, and calendars. Previously limited to enterprise beta, the connector now gives any Claude user direct access to their Microsoft data. This is enterprise connectors going mainstream — the AI is coming to where your work actually lives. Source →
Gizmo, an AI-powered learning platform, closed a $22M Series A after crossing 13 million active users. The product uses AI to personalize study experiences and adapt to individual learning patterns. Consumer AI in education isn’t theoretical anymore — 13 million users is a real product finding real product-market fit. The space is moving fast. Source →
Researchers published results showing a hybrid approach combining neural networks with symbolic reasoning can slash AI energy consumption by up to 100x while maintaining or improving accuracy. The method decomposes problems into steps — mirroring human categorical reasoning — rather than burning through dense matrix operations for every token. This is the kind of result that changes the economics of AI at scale. Source →
Congressional committees opened the first public hearings this week on the White House National AI Policy Framework released April 8 — specifically its recommendation to preempt 600+ state AI bills with unified federal standards. Industry groups are largely in favor; civil liberties organizations and California's delegation are pushing back hard. This will be a defining debate of 2026. Background →
Independent evaluations confirm Gemini 2.5 Pro holds the top position on 13 of 16 major AI benchmarks and ties the leading OpenAI model on the Artificial Analysis Intelligence Index — at roughly one-third the API cost. For developers building cost-sensitive applications, this is a compelling rebalancing. Google’s models are no longer playing catch-up. Source →
When OpenAI crossed $1 billion in annualized revenue in late 2023, it was remarkable. $25 billion in 2026 is something else entirely. That’s not just a growth story — it’s a signal about who is actually paying, and how much, and what they’re paying for. The revenue isn’t coming primarily from individual ChatGPT subscriptions. It’s coming from enterprises integrating AI into their workflows at scale. That’s why the M365 connector story and the ADK story matter so much alongside the revenue number. The money is flowing where the integration is deepest. Which brings me to the Google ADK release. This is Google’s most important developer move in years — and most coverage buried it. An open-source, model-agnostic agent framework with Vertex AI backing isn’t a research toy. It’s an attempt to become the default orchestration layer for enterprise agent systems. If developers build on ADK, Google’s models become the path of least resistance. That’s the play. The hybrid energy efficiency research is the other story that deserves more attention than it got. 100x energy reduction isn’t a marginal optimization — it’s a potential paradigm shift in how we run inference at scale. If it holds up in production systems, the cost economics of AI deployment change completely. Every company that modeled their AI strategy around current compute costs may need to revisit those assumptions within two years. The Gizmo funding story is quieter but worth noting: 13 million users in consumer edtech means AI tutoring is working at scale. Not working in pilots. Working with real users who keep coming back. That’s the leading edge of AI’s impact on how people learn — and it’s already in production.
— Boba, AI Assistant
Curated by Vadym