A pillar of the UK AI Opportunities Action Plan is lowering the barriers to constructing billions of pounds in new data centers in AI growth zones. This may be a good thing for the UK more broadly, but not necessarily for some of the reasons mentioned, like jobs and tax revenues. There are some important lessons to consider from Northern Virginia, the data center capital of the world.
Sam Altman recently posted “Three Observations,” suggesting that AI will benefit all of humanity with more chips, improved efficiency, and socio-economic value increasing super exponentially. But meaningful progress also requires including the human interests and curiosity of the other 99.9%, not to mention the missing T-word.
Jeffrey Emanuel’s essay on the short case for NVIDIA is widely attributed to sparking the largest-ever single-day drop in the stock market. Although he is bullish on AI in general, he does feel that highly perishable AI infrastructure is being built faster than it can be economically consumed.
The impressive cost/performance gains demonstrated in DeepSeek have important implications for AI strategy and investment. Better AI will require thinking about how to lower the barriers to experimentation as much as increasing data center footprints and energy consumption.
The current crop of pixel-based tools for capturing and rendering 3D are considerably different from how we perceive the world as light fields. This has important implications for the future of better digital twins and smart glass user interfaces.
Vendors at the recent Consumer Electronics Show were pitching a bevy of new smart glasses that sounded like they could provide a bigger screen for desktop apps. But don’t get your hopes up around any serious productivity.
The US’s permissive personal data scraping regime has ‘unlocked’ AI innovations like raising insurance rates and allowing hackers to track your every move. These recent developments suggest valuable lessons in implementing the UK AI Opportunities Action Plan.
The outgoing Biden administration has enacted “AI diffusion” rules that impose new KYC burdens on the export of AI model weights and associated “computational operations.” At the very least, they will complicate AI business models and associated cloud infrastructure if they go into effect in three months as planned.
NVIDIA has made considerable progress on its tooling for building agentic AI, including new foundation models, new agentic AI blueprints, and partnerships with several open source vendors to develop AI blueprints for processes in software development, conversational chatbots, generating reports, revenue generation, video analytics and industrial asset troubleshooting.
NVIDIA officially launched Cosmos, a new world foundation model platform for streamlining the development of physical AI applications in robotics, autonomous vehicles, logistics, and manufacturing. This will make it easier to develop embodied AI applications that learn from interacting with the world rather than just what people have written.
New generative AI techniques are finally making it easy to activate all that unstructured data lying around in enterprise data stores. The problem is that there are many ways to do it wrong, and it's important to decipher how it might add value to the bottom line, not to mention for customers and employees.
The generative AI boom has sparked a lively discussion on the future of trust and safety and calls for new regulations in 2024. This has sparked an unflinching look at some of the practical issues leading to the growth of ethical debt required to prepare for the future.
Salesforce has introduced a new AI readiness index that suggests a framework that outlines some of the key factors required to accelerate the broader adoption of AI, particularly in B2B.
