🚀 AgentKit Shrinks Months of Agent Building into Hours
The toolkit combines four building blocks: Agent Builder lets you design workflows on a visual canvas; ChatKit embeds chat‑based experiences into your app; a Connector Registry securely links to data sources; and Evals for Agents helps you test and optimize prompts and models. Early adopters like Ramp report cutting iteration cycles by 70% and achieving higher accuracy.

🛠️ Apps Turn ChatGPT into an App Store
OpenAI’s Apps platform allows ChatGPT users to call services from partners like Booking.com, Expedia, Spotify, Figma, Coursera, Zillow and Canva by name or let ChatGPT suggest them. The preview Apps SDK lets developers build and monetize their own conversational tools. The system works seamlessly within ChatGPT and surfaces apps contextually, making the assistant feel more like an operating system than a chat bot.

🤝 AMD and OpenAI Forge a 6 GW Compute Alliance
Over several years, OpenAI will deploy six gigawatts of AMD Instinct GPUs, beginning with one gigawatt of MI450 units in late 2026. AMD granted OpenAI warrants to purchase up to 160 million shares as capacity milestones are met. AMD’s CFO expects the partnership to generate tens of billions of dollars in revenue.

📈 ChatGPT Soars to 800 Million Weekly Users
At OpenAI’s DevDay, Sam Altman revealed that ChatGPT has 800 million weekly active users, processes over 6 billion tokens per minute, and has attracted more than 4 million developers. The surge sets the stage for the new Apps ecosystem and highlights ChatGPT’s entrenched position in productivity workflows.

Windsurf is where developers today are doing their best work. Powered by the team behind Devin, the best software engineering agent, Windsurf is once again revolutionizing the AI IDE space by delivering the smartest and most reliable agentic workflows.

Try Windsurf for free today!

After nearly a decade of stalled revenue and strategic pivots, Replit has pulled off a dramatic turnaround—growing from $2.8 million to $150 million in annualized revenue in under a year, thanks to its AI coding assistant and a bold shift away from pro developers. The company’s new focus? Empowering everyday knowledge workers to build software, not just seasoned engineers.

That move, combined with a lean 110-person team and gross margins as high as 90% on enterprise deals, just earned it a $250 million raise and a $3 billion valuation. But with foundation model giants like OpenAI and Anthropic eyeing the same space, Replit’s edge will be tested. → Read the full article here.

Join the waitlist to get early access to curated jobs, and find your next big thing. 

The Recap: Surging investment in AI data centers, projected to hit $5.2 trillion globally by 2030, is raising fears of a bubble reminiscent of the 1990s telecoms crash. While demand for generative AI remains uncertain, firms like OpenAI and NVIDIA are driving massive, debt-fueled infrastructure projects in remote regions. The Economist explores how novel factors, such as financing from private lenders, risky tenants, and isolated locations, could spell trouble if expectations don’t match reality.

Highlights:

Forward Future Takeaways:
This story underscores how AI's future isn't just about algorithms, it's increasingly about physical infrastructure, finance, and risk tolerance. The scramble to build generative-AI capacity may outpace real-world demand, exposing tech and financial systems to a familiar boom-bust cycle. As investment surges ahead of returns, the key question is whether AI’s economic promise will arrive in time to justify the colossal bet. → Read the full article here. (Paywall)

Security researchers have uncovered three serious vulnerabilities—dubbed the “Gemini Trifecta”—in Google’s Gemini AI that exposed users to prompt injection, cloud exploitation, and data exfiltration. The flaws affected key components like Gemini Cloud Assist, the Search Personalization model, and the Browsing Tool, allowing attackers to manipulate AI behavior and siphon sensitive user data via poisoned logs, browser histories, and embedded prompts.

Google has patched the issues, but the findings underscore a growing concern: AI tools aren’t just targets—they’re becoming attack vectors. As adoption accelerates, securing these systems is now a frontline priority. → Read the full article here.

🧠 Teen Builds AI Memory API, Snags Top Backers: 19-year-old Dhravya Shah raised $2.6M for Supermemory, an AI tool that extracts insights from messy data, Google and Cloudflare are in.

👨‍🔧 CodeMender Tackles Code Bugs Autonomously: Google researchers unveiled CodeMender, an AI agent that spots and patches security flaws, 72 fixes shipped already.

✨ Codex Now Generally Available: OpenAI’s AI coding agent rolls out with Slack integration, SDK access, and admin tools, daily use is up 10x, with GPT-5-Codex serving 40T tokens in 3 weeks.

🤔 Eufy Paid Users for Staged Thefts: Anker offered $2 per video to Eufy camera owners—even for faked crimes—to train AI, sparking new concerns over privacy, data use, and transparency.

💸 Figma Stock Jumps on ChatGPT Integration: Shares rose 7% after Sam Altman demoed how users can run Figma tasks directly in ChatGPT—monetization tools for third-party apps en route.

Tired of predictable answers? Talk to yourself, from a parallel universe.

📢 Want to advertise with Forward Future? Reach 550K+ AI enthusiasts, tech leaders, and decision-makers. Let’s talk—just reply to this email.

🛰️ Want more Forward Future? Follow us on X for quick updates, subscribe to our YouTube for deep dives, or add us to your RSS Feed for seamless reading.

Thanks for reading today’s newsletter—see you next time!

Matthew Berman & The Forward Future Team

🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀