AI Daily Signal: Nvidia Changes CEO and Fractile Raises $720M

Jensen Huang resigns as Nvidia CEO, while Fractile raises $720 million, OpenAI reports stronger revenue, and enterprise agents expand.

Follow in Google Search

Leadership, capital, and product design all shifted across the AI market in the past 24 hours. Nvidia began an immediate CEO transition after Jensen Huang resigned. Fractile raised $720 million to commercialize a new inference chip architecture, while OpenAI reportedly crossed $10 billion in quarterly revenue. Databricks and Google also introduced products that move AI from question answering toward longer-running, collaborative work.

Jensen Huang resigns as Nvidia CEO

Nvidia announced that Jensen Huang resigned as president, chief executive, and a board member, effective immediately. Chief Financial Officer Colette Kress will serve as interim CEO while continuing in her finance role. The board has started a search that will consider internal and external candidates, and it named Harvey C. Jones as Nvidia’s first board chair.

The transition is unusually significant because Huang co-founded Nvidia and led it for 33 years, including its transformation into the central supplier of computing infrastructure for the current AI boom. Nvidia said customer commitments, operations, and its product roadmap are expected to continue without interruption. The next test comes quickly: the company is scheduled to report fiscal third-quarter results on November 17.

OpenAI’s quarterly revenue reportedly reaches $10 billion

TechCrunch reported that OpenAI generated more than $10 billion in third-quarter revenue, citing a Bloomberg report based on information shared with investors. That would be an increase of almost 50 percent from the reported $6.7 billion in the second quarter. The company’s adjusted operating margin also reportedly improved to negative 109 percent from negative 150 percent.

The figures show stronger demand and better unit economics, but they also underline the cost of competing at the frontier. OpenAI is investing in models, consumer products, enterprise tools, and agents at the same time. Revenue growth gives it more operating room, yet the reported loss rate means pricing, compute efficiency, and enterprise adoption remain central to the business case.

Fractile raises $720 million for inference chips

British chip startup Fractile raised a $720 million Series C backed by a group that includes Thrive Capital, Founders Fund, Lightspeed, Redpoint, and Britain’s Sovereign AI Fund. The British government described the valuation as more than £5 billion, while other reports placed it near $7 billion. Fractile plans to ship its first chips and systems in 2027.

Fractile is betting that stacking DRAM directly above its processors can reduce the memory bottleneck that slows inference for large models and long agent workflows. It is also building a compiler to optimize models for the hardware. The funding is a large vote of confidence before production shipments, so the important evidence will be customer deployments, performance under sustained workloads, and manufacturing scale.

Databricks turns Genie into an always-on enterprise agent

Databricks introduced a second generation of Genie that can run in the background, delegate work to subagents, track long projects, and ask users for decisions through a priority inbox or email. Genie Search is also expanding to more than 100 data sources, including Google Drive, Jira, and Confluence, while actions remain subject to Unity Catalog permissions and Unity Gateway rules.

Databricks says Genie and Genie Ontology are generally available, while the new always-on mode and Genie Search are coming to preview and beta programs. That distinction matters. The product direction is clear, but enterprises should evaluate how reliably the agent preserves context, respects approval boundaries, and recovers from mistakes before assigning it long-running operational work.

Google Labs launches collaborative Tabletop spaces

Google Labs introduced Tabletop, an experiment that turns links, files, and prompts into shared interactive spaces. Groups can generate trip itineraries, dinner polls, expense splitters, or tournament brackets, then collaborate through an invite link. The product is rolling out to adults in the United States through a waitlist.

Tabletop applies generative AI to coordination rather than a private chat. Its value will depend on whether the generated tools stay understandable and editable as several people contribute. The experiment also gives Google another way to test AI interfaces that sit between documents, lightweight apps, and group decision-making.

SE3 raises $30 million for spatial intelligence

Munich-based SE3 raised a $30 million Series A led by Accel, with participation from Keen Venture Partners, Lakestar, and Seedcamp. Its software combines computer vision and sensor data to build a shared three-dimensional view for drones and other unmanned systems. SE3 says the system runs at the edge, works across hardware vendors, and is already deployed in Ukraine.

The company is targeting environments where satellite navigation and communications can be disrupted. That makes edge processing and interoperability practical advantages, but military autonomy also raises difficult accountability questions. Voice control and multi-system coordination can reduce operator workload, while making clear authorization, audit trails, and human responsibility even more important.

What matters today

Today’s news shows the AI industry becoming both larger and more operational. Nvidia must manage a historic leadership transition without slowing its roadmap. OpenAI’s reported revenue growth and Fractile’s funding indicate that investors still expect rapid expansion across software and chips. At the product layer, Databricks, Google, and SE3 are pushing AI into persistent work, shared planning, and physical systems, where governance and reliability will matter as much as model capability.

Author

Dr. Rajesh Patel

PhD in Electrical Engineering and Computer Science, MIT (2016); Postdoctoral research, UC Berkeley BAIR. Research on efficient training algorithms, multimodal architectures, and model robustness.