ENERGY

AI could cause a power shortfall

AI firms continue planning astronomical AI infrastructure. But can the US power supply hack it? 

Anthropic has joined the slew of AI firms investing billions in massive data centers throughout the US. On Wednesday, the company announced that it would invest $50 billion in American AI infrastructure, starting with data centers in Texas and New York, in partnership with Fluidstack. 

Anthropic joins OpenAI, Nvidia, Oracle, Softbank and more in the race to develop these sites and evolve its AI models. But the power demands of these data centers may exceed the power grid’s capacity. 

In a note published earlier this week, Morgan Stanley analysts warned that AI demand could leave the US with a “power shortfall totaling as much as 20%” for data centers through 2028, reaching a deficit of up to 13 gigawatts. 

Though tech leaders claim that the need for compute is the biggest problem facing the evolution of AI, energy supply and grid reliability present an even greater risk. The problem is that the building and deploying of these colossal server farms is far, far outpacing utility companies’ ability to upgrade the grid, Sebastian Lombardi, chair of the energy and utilities practice at law firm Day Pitney, told The Deep View.

While the problem is currently deepest felt in “pockets” of the US that have high concentrations of data centers, it’s only a matter of time before the stress on the grid and energy demand are felt all over the country, he said, possibly resulting in issues with reliability and affordability for utility payers. The rapid pace and magnitude of these buildouts are leaving utility companies and regulators scrambling to play catch-up.

“The AI data center story has complicated things. It’s created some questions about how we are going to maintain reliability,” said Lombardi. “The amount of energy that is expected to be used to power that infrastructure is quite significant.”

Despite the issue at hand, tech firms show no signs of tempering their all-out sprint. The solution might be what Lombardi calls an “all of the above” strategy for increasing the energy supply. This means a sharper focus on renewable energy, as well as more stakeholders buying into the “renaissance” that nuclear power is having. Even moonshot ideas, such as Google’s space-based Suncatcher data centers, could be worth exploring. “We all want the lights to stay on,” said Lombardi. “We may have to get away from picking winners and losers if we're going to meet the pace and the magnitude of demand.”

However, how much help these tech companies will be in solving the problem that they're effectively causing, especially as many struggle with their net-zero goals, is still unclear.

TOGETHER WITH AUTH0

Secure your AI agents with Auth0 for AI agents

Enable your AI agents to more securely access your tools, workflows, and users’ data. Outsource the various complexities (like auth) so you can focus on what really matters – building the app. That’s why we’re such fans of Auth0 for AI Agents.

Auth0 for AI Agents handles all the tricky aspects of adopting AI for your business with just a few lines of code. Handle and secure access tokens, process authorization requests, and implement AI with more confidence… without sacrificing any security. It’s exactly the kind of support builders like those that read The Deep View need to take their idea to the next level. 

Try Auth0 for AI agents in Developer Preview now, the complete auth solution for AI agents.

STARTUPS

AI works better with proprietary data

As large, foundational models get larger and larger, they start to act the same. The differentiator is in the data. 

Alembic Technologies, a San Francisco-based AI lab, is dedicated to solving the problem of AI homogeneity, making models that are actually distinct from one another, founder and CEO Tomas Puig told The Deep View. 

“As we see the capabilities of these models converge … this creates a very large problem for corporations,” said Puig. “While I think generalized intelligence is really good, where we've really focused on is building the best intelligence in the world from private data sets.” 

The startup, which develops custom AI models for enterprises using their proprietary data, announced a $145 million Series B funding round. The round multiplies the company’s valuation more than 15-fold, bringing it to $645 million, Puig said. 

Alembic’s focus lies specifically in causal AI models, or those that think using cause and effect. 

• For example, using a company’s own data, a causal model may analyze which kinds of marketing perform best for a company and why those tactics do well. 

• “The benefits of the cause and effect side of the house is you actually know what you can affect and what you cannot, what is worth pursuing and what's not worth pursuing,” Puig said. 

Additionally, Alembic announced that it is deploying a DGX AI Supercomputing cluster running the NVIDIA AI Enterprise software suite. The architecture, to be constructed in California, marks the company’s second such cluster, the first being in Virginia. Given that it’s building models with private and sensitive data, owning its own hardware is key in ensuring privacy, Puig said. 

“For our clients at their security level, they want to know that literally anything we compute never leaves our own private house,” said Puig. “We work with the type of data that nobody in the world wants to give somebody else access to.”

TOGETHER WITH WIZ

How 100+ Security Leaders Are Tackling AI Risk

AI adoption is accelerating – and new research shows most security programs are still working to catch up. Wiz’s latest guide, AI Security Readiness: Insights from 100 Cloud Architects, Engineers, and Security Leaders, explores how organizations are approaching AI security today, where the biggest gaps exist, and what actions leading teams are taking to reduce risk.

In this free report, you’ll learn: 

• How organizations are adopting and securing AI 

• The strategies teams are using to manage shadow AI and visibility gaps

• A maturity model to benchmark your organization’s progress

If you want to make smarter AI security decisions that are grounded in what real teams are doing right now, download your free copy right here.

PRODUCTS

Cursor hits $30 billion valuation

AI-powered coding is proving to be one of the most lucrative use cases for the tech. And Cursor is raking it in. 

On Thursday, the company announced a $2.3 billion funding round, boosting its valuation to $29.3 billion, up from $2.5 billion in January. The round was led by Accel, with new investors including Google and Nvidia joining in, signalling that tech giants see no signs that AI coding tools are going anywhere. 

Cursor has bloomed over the past year as companies began automating the work of young coders. The startup’s platform has drawn more than one million users, and counts tech heavyweights like OpenAI, Stripe and Adobe as customers. According to the Wall Street Journal, it’s turned down multiple acquisition offers from major companies. The company also said it has surpassed $1 billion in revenue. 

The platform allows users to toggle in between different AI models as they code, including those by Anthropic and OpenAI, and includes an autoselect feature that picks the right model for the job.

• However, it may be seeking to bring AI in-house: In late October, the company launched Composer, its own in-house coding model, built for low-latency agentic coding and capable of completing most tasks in 30 seconds or less.  

• In a post on X earlier this week, the company published its popular and fastest-growing models on its platform as of November. Claude Sonnet 4.5 topped the list for popularity, with Composer coming in second and GPT-5 in third. 

The company’s investors told the Wall Street Journal that the company’s main objective is to add more users, and CEO Michael Truell said that AI model companies remain “important partners to us.” 

But in boosting usage of its proprietary AI model, the company may be able to actually keep some of its revenue, rather than forking it over to major model providers like Anthropic, OpenAI and Google. 

LINKS

• XAI reportedly in talks to raise $15 billion, though Elon Musk denied it

• Salesforce acquires agentic startup Doti AI for $100 million

• Satya Nadella talks about preparing for AGI on the Dwarkesh Patel podcast

• Baidu reveals two new AI chips for "powerful, low-cost, and controllable” AI compute

• Facebook Marketplace gets an AI upgrade

• Mira Murati’s Thinking Machines is in talks to raise funds at $50 billion valuation

• Anthropic disrupted a sophisticated AI-powered espionage campaign

• Baidu Ernie 5: Baidu has unveiled the next generation of its flagship Ernie chatbot, with reduced token options and a nicer personality. 

• World Labs Marble: A foundational world model for a “spatially intelligent future.”  

• LinkedIn AI-powered people search: find the person you are looking for on LinkedIn easier, using nothing but natural language. 

• Google SIMA 2: A new video game agent from Google DeepMind, capable of navigating and solving problems in virtual worlds. 

OpenBuilder: An open source alternative to vibe coding platforms like Lovable.

$100K Fees Just Redefined Hiring

Beginning Sept 21, 2025, each new H-1B visa will cost $100,000.

That’s stalling growth across U.S. startups and tech giants alike.

The alternative? Remote hiring done right.

Build across Latin America: high-caliber engineers, AI experts, 60% cost savings — and zero visa limits.

The future is borderless.

No bureaucracy. No $100K barrier. Just brilliant teams, built fast.

👉 See “The H-1B Talent Crunch 2025”

GAMES

POLL RESULTS

Can AI truly understand the world without a physical body?

• Yes - simulation is enough (19%)

• No - embodiment is essential (30%)

• Depends on the task (robots yes, chatbots no) (26%)

• Understanding is overrated - prediction is enough (15%)

• Something else (write in!) (10%)