AI Furnace Newsroom: Meta's $2B Manus AI Acquisition, NVIDIA's $20B Groq Deal, OpenAI's Ad Monetization Move and JP Morgan's AI Adoption Strategy

Welcome to this week’s AI Furnace Newsroom

In the last week of 2025, we saw a culmination of the major trends of the year coming to the fore and acquisitions/acqui-hires taking centre stage as companies position themselves for 2026. Manus AI was acquired by Meta, Groq agreed to an acqui-licence deal with NVIDIA, Cursor acquired Graphite.

At the infrastructure level, hardware giants continue the chip wars and electricity, power, and data centres continue getting investment as compute capacity stays the bottleneck.

At the foundation model level, competition between foundation models continues to increase in velocity and pressure with 2025 seeing the launch of Google’s Gemini 3 Pro, OpenAI’s GPT-5.2 and Anthropic’s Claude Opus 4.5. Meta’s Llama saw a slowdown as Meta figures out its AI strategy, with Yann Le Cun (Chief AI Scientist at Meta) leaving and Alexandr Wang joining to lead out the AI strategy following Meta’s Scale AI acquisition. It looks like Meta may be moving to the application layer, with its acquisition of Manus AI, and this may be a trend for other AI labs to follow in 2026.

At the application layer, AI agents for software development, customer service and for niche verticals (think Harvey for legal or OpenEvidence for healthcare) cemented their traction with strong industry adoption; job losses continue; startups move from AI pilots to internal scaling; enterprise CIOs focus on scale and figure out if they should build internally, partner with incumbents or bring in more agile startups to help them get ahead of competitors and achieve ROI.

In today’s insights we cover:

1. Meta Acquires Manus AI for $2B

2. NVIDIA Doesn’t Acquire Groq - it Makes its Biggest $20B Financial Bet Yet

3. OpenAI Explores Ads in ChatGPT for Additional Monetization

4. Cursor Scoops Up Code Review Startup Graphite

5. Salesforce Rolls Back on Generative AI

6. JP Morgan’s AI adoption hit 50% of employees

Read time: 5 mins

💡 Furnace Insights

Meta

Meta Acquires Manus AI for $2B

Meta has acquired Manus AI, a Singapore-based general purpose AI agents company, for a reported $2B+.

This sees Meta buying not just any AI startup, but specifically an AI agent distribution play. Manus AI recorded $100M+ annual revenue run-rate just 8 months after launch. It claimed to have processed more than 147 trillion “tokens” of text and data, and supported over 80 million virtual computers.

It offers both free and paid subscription tiers. The company already had large customers, such as Microsoft which began testing Manus in Windows 11 PCs, allowing users to create websites from local files.

As part of the Meta deal, Manus employees will help to integrate AI agent capabilities across Meta’s consumer and business products. Manus will keep running as an independent subscription service, and it will close its Chinese operations and ties with Chinese investors.

What this says about Meta’s mindset: the new moat is about tech and distribution and trust. Meta did not have any AI-native products and this move allows Meta to integrate an AI agent product that already has users and revenue. 2025 was the year of AI agents, and we saw no significant moves from Meta previously. The Manus deal now puts them squarely on the AI Agent map for 2026.

NVIDIA

Nvidia Doesn’t Acquire Groq - it Makes its Biggest $20B Financial Bet Yet

Nvidia struck a $20B cash-based asset deal with Groq to deepen its AI inference stack, but this is no usual acquisition. The asset deal involves Nvidia licensing Groq’s low-latency inference IP and the transition of Groq’s founder and senior leaders to Nvidia.

GroqCloud is outside the deal, and Groq’s other operations will continue as usual under new leadership. Nvidia will be integrating Groq’s low-latency processors into its AI factory platform.

This move follows a pattern of talent-and-IP deals that have come to define the AI M&A landscape. It also signifies the shift of AI spend from training to inference.

Nvidia pulling Groq’s low-latency designs into its stack shows that demand is currently coming from chat systems, agents, search, and real-time apps that need instant responses at scale. For developers and enterprises, this points to faster inference options baked directly into Nvidia’s platform and fewer reasons to experiment outside its ecosystem.

OpenAI

OpenAI Explores Ads in ChatGPT for Additional Monetization

OpenAI is reportedly exploring new ways to monetize ChatGPT by embedding conversational ads and sponsored content directly into chatbot responses. This follows mounting pressure for OpenAI to turn its massive free user base into sustainable revenue.

According to a recent report, the company is evaluating ad formats that would surface personalized recommendations within conversations or alongside responses, rather than relying solely on traditional subscription fees. This is an effort aimed at offsetting the high costs of infrastructure and model training.

However this poses a threat to the experience, if recommendations feel biased or sales driven. While still in early planning and subject to user trust considerations, this marks a potential shift toward intent based monetization that could redefine how AI platforms generate income and interact with mainstream users.

Cursor

Cursor Scoops Up Code Review Startup Graphite

AI coding platform Cursor has acquired Graphite, an AI-powered code review startup, in a move to build an end to end software development workflow. By integrating Graphite’s review capabilities directly into its editor, Cursor is targeting one of the biggest bottlenecks in AI-assisted programming: code review and approval.

The deal, reportedly priced above Graphite’s last $290 million valuation, comes as competition in AI code generation intensifies among players like OpenAI, GitHub Copilot, and a growing set of specialist developer-tool startups racing to own the full development lifecycle. Cursor has been on an acquisition spree following it’s valuation of $29B in November.

JP Morgan

JP Morgan’s AI Adoption Hits 50% of Employees

JP Morgan’s internal LLM assistant platform went from 0 to 250,000 users in months. Today, more than 60% of employees across the bank actively use it. This wasn’t driven by top-down mandates. Adoption spread virally as employees built, customized, and shared AI assistants tailored to their own roles and workflows.

Behind that success was a deliberate connectivity strategy: rather than betting on proprietary models as a moat, it invested early in building a sophisticated, multi-modal RAG (retrieval-augmented generation) platform now in its fourth generation. This plugs directly into the bank’s most critical systems, data, documents, and tools. This “ubiquitous connectivity” lets employees apply AI to real work across finance, risk, HR, CRM, trading, and operations.

The result is a “one platform, many jobs” approach, where reusable AI building blocks can be assembled into role-specific tools at scale. JPMorgan’s view is clear: models will commoditize, but the real value and competitive advantage comes from how deeply AI is wired into enterprise knowledge, processes, and systems. Without that, even the most advanced AI remains little more than a shiny demo.

Salesforce

Salesforce Rolls Back on Generative AI

Salesforce is dialing back its use of open ended generative AI after trust and performance issues surfaced in real world deployments.

Executives point to missed instructions, task drift, and failures once prompts grew more and more complex. After rolling out AI agents and cutting around 4,000 support roles, the company is now prioritizing deterministic triggers and data foundations.

CEO Marc Benioff, who has been a vocal advocate of AI led change, is now drafting Salesforce’s annual strategy document with data foundations, not AI models, as the top priority. The company is now prioritizing rule based automation and deterministic workflows to ensure consistency, compliance, and reliable outcomes. The shift signals a broader trend in enterprise AI: moving away from “AI everywhere” hype toward controlled, auditable automation that can consistently deliver results at scale.

📈 AI Venture Deals of the Week

• Newera.ai, an enterprise generative/agentic AI builder, raised $2.1M in its Pre-Seed round.

• Dazzle AI, a consumer AI tools startup, raised $8M in its Seed round.

• Jasper, an AI content platform for marketers, raised $125M in its Series A round.

• Sharge, a consumer tech company building AI glasses, raised ~$14.29M in its Series A+ round.

• Kargo, an industrial AI platform for warehouses/logistics, raised $42M in its Series B round.

⚒️ New AI Product Launches You Don’t Want to Miss 

• GLM-4.6V, by Z.ai, an open-source multimodal LLM series (including a 106B model and a 9B “Flash” model) designed for strong visual understanding with native multimodal tool/function calling for agent workflows.

• Gambo, a “game vibe coding” agent that helps you generate and remix playable games (including assets) and includes built-in monetization options like ad integrations.

• Keak, an AI agent for website optimization that generates variants, runs A/B tests, and updates your site to improve conversions (with a lightweight pixel/script approach).

• LFM2-2.6B-Exp, by Liquid AI (LiquidAI), an experimental 2.6B-parameter text generation model checkpoint trained with pure reinforcement learning, aimed at strong instruction following/knowledge/math performance for agentic and multi-turn use cases.

Upcoming Events 📅

• AIF Founder Supper Club @ Whiskey Club - Wednesday, January 21

• AIF Founder Supper Club @ Wine Bar - Wednesday, February 18

• AIF Founder Supper Club @ Private Room - Tuesday, March 24

Interested in meeting the Who’s Who of AI?

The AI Hot 100 Summit is back for its third edition on May 7th, 2026 in New York City. The only AI Summit built for real connections where startup innovation meets enterprise implementation. Join over 500 AI Executive Leaders, Founders and Investors for a jam packed day of enterprise AI lightning talks, innovation showcases, and curated networking.

Get ready to meet and greet speakers and attendees from OpenAI, NVIDIA, Haleon plc, Legora, Anthropic, Cursor, Eleven Labs, Adobe, Walmart, Insight Partners & more.

🗓️ May 7, 2026

📍New York City