Good morning. Well, Sam Altman got his wish. Project Stargate is a go.

It’s not $7 trillion, but it’s a start. And it is most obviously a massive win for Altman and OpenAI (and the chipmakers and cloud companies, of course).

We break it all down below.

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

• 🏗️ AI for Good: Concrete Copilot 

• 🚨 Google, AI and the IDF

• 💰 Trump unveils $500 billion Big Tech coalition for massive AI build-out

AI for Good: Concrete Copilot

We talk a lot about the carbon footprint of AI models, but data centers are far from the only sector responsible for outsize carbon emissions. The cement industry in 2022 produced 1.6 billion metric tons of carbon emissions, amounting to about 8% of total global emissions for that year. 

And according to the World Economic Forum, concrete is “the second most widely used material after water,” meaning that production — and, therefore, emissions — is on pace to continue ramping up. 

What happened: Factoring into the solutions on offer is Concrete.ai, a firm that last year launched the commercial version of its Concrete Copilot, a generative AI-based platform designed to reduce the carbon footprint associated with mixing concrete. 

• The system works by integrating with a specific producer’s current and historical data, at which point it can — based on specifications set by the producer — spin out millions of optimal concrete mix designs in accordance with those specifications. 

• During field tests, the platform was able to achieve an average carbon reduction of 30%, and an average cost savings of $5.04 per cubic yard. 

“We built Concrete Copilot, so producers don’t have to choose between cutting cost and carbon,” Alex Hall, CEO of Concrete.ai, said. “Our ultimate goal is to reduce the annual global carbon footprint of concrete by 500 million tons just by optimizing concrete mixes with materials already in supply chains.”

Catch Hallucinations in any LLM application

Even today’s top LLMs still hallucinate too much for business-critical applications.

That’s why Cleanlab developed the Trustworthy Language Model (TLM) — a state-of-the-art BS detector that flags incorrect LLM answers in real-time.

Read comprehensive benchmarks which show TLM can slash the error rates of popular LLMs (including o1, GPT-4o, Sonnet 3.5) by over 20%.

Whether it’s RAG, chatbots, or data extraction/annotation, TLM adds trust to any LLM application.  This reliability layer is your key to scaling AI without blowing your credibility. 

Stop hoping your AI “just works” and start knowing when it does.

Experience LLM trust scoring with your own prompts — try the Interactive TLM Playground now.

Google, AI and the IDF

From the very earliest days of the Israel-Gaza war, Google has worked to provide Israel’s military forces with access to its most advanced AI technology, the Washington Post, based on exclusive internal documents, reported. 

The details: According to those documents, a Google Cloud employee escalated requests to grant Israel’s Defense Ministry access to its latest AI tech, notably including Google’s Vertex, a tool enabling customers to custom-build algorithms based on specific data. 

• These requests continued through much of 2024, with the Israeli military requesting access in November to Google’s Gemini tech in order to build its own AI assistant to process documents and audio. It is otherwise unclear how the IDF has leveraged Google’s AI tech. 

• As we have previously reported, the IDF has additionally developed and deployed AI tools designed to provide military commanders with bombing targets; they do so at a rate that is roughly 50 times faster than what teams of human analysts were previously able to achieve. It’s not clear if Google is supporting these efforts, or how. 

Google last year fired more than 50 employees who were protesting Project Nimbus, the $1.2 billion contract between Google and the Israeli government signed in 2021. The above applications are all results of Project Nimbus. 

Get A Prioritized Target Account List Instantly with AI

Stop wasting time searching for the right customers to target. Let AI do the work for you.

Leadtalk uses AI to analyze your best customers and instantly gives you a list of lookalike target accounts.

✔️ Enter your three best customers

✔️ Get a prioritized target account list 

It’s that easy! No complicated setup, no credit card required, just smarter, faster, account targeting.

Precision targeting starts here. Get your first 25 accounts free. Try for free today.

• Supercharge your organization's AI adoption journey from complex challenges to strategic success with Tines' must-read guide: "Securing AI in the Enterprise." *

• Google invests an additional $1 billion in Anthropic (FT).

• Startups’ AI revenue is booming. Some investors doubt it will last (The Information).

• Twisted crystals show hints of a new kind of superconductivity (New Scientist).

• Meta backs Databricks as the data analytics startup inches toward IPO (CNBC).

• France’s electricity generation is nearly fossil fuel-free (Semafor).

If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

• Though the TikTok ban has been delayed, the app’s future is far from decided. Trump said Tuesday that he would support ventures by Elon Musk — who already owns Twitter/X — or Larry Ellison to buy the app.

Trump unveils $500 billion Big Tech coalition for AI build-out

President Donald Trump, alongside OpenAI CEO Sam Altman and Oracle Chairman Larry Ellison, announced Project Stargate Tuesday evening, a new company made up of several layers of prominent Big Tech corporations that will invest a minimum of $500 billion into building out AI infrastructure (namely, data centers) over the next four years in the U.S. 

Here’s what we know: OpenAI wrote in a post that the startup and Softbank will function as the project’s two leaders. Softbank — with a market cap of around $88 billion — will have financial responsibility, while OpenAI will have operational responsibility. 

• The initial funders include OpenAI, Softbank, Oracle and MGX, an investment fund launched by the United Arab Emirates last year with the sole goal of investing in several layers of the AI tech stack as part of Abu Dhabi’s effort to reshape itself into the next Silicon Valley. 

• Microsoft, Nvidia, Oracle, OpenAI and Arm will serve as the project’s key technological partners, a team that will work together to build and deploy massive new data centers. 

OpenAI said it would begin “deploying $100 billion immediately,” adding that the buildout has already begun in Texas. Ellison said at the press conference that 10 data centers — each 500,000 square feet in size — were already in construction in Abilene, Texas, and 10 more are on the way.

“All of us look forward to continuing to build and develop AI — and in particular AGI — for the benefit of all of humanity,” OpenAI said. “We believe that this new step is critical on the path, and will enable creative people to figure out how to use AI to elevate humanity.”

By AGI, OpenAI is referring to something it defines as “highly autonomous systems that outperform humans at most economically valuable work,” though the term, steeped in marketing hype, lacks a universally accepted definition and remains scientifically dubious and entirely hypothetical. 

As part of all of this, Microsoft said it is no longer OpenAI’s exclusive cloud provider; the two are switching instead to a model where Microsoft has the right of first refusal whenever OpenAI seeks more compute. Microsoft has independently said that it plans to spend a further $80 billion on AI infrastructure in 2025. Microsoft CEO Satya Nadella did not clarify to CNBC whether Stargate will require an additional investment from Microsoft, and what that might amount to. 

Trump said at a press conference that this project would create 100,000 jobs “almost immediately.” 

The fiscal responsibilities of each of these companies relevant to the $500 billion total are unclear. OpenAI lost $5 billion in 2024 on roughly $4 billion in revenue, and expects those losses to continue spiking until the end of the decade. 

Elon Musk, who, despite his relationship with Trump, isn’t a part of the Stargate project, said that OpenAI doesn’t “actually have the money.” 

The context: This all comes shortly after Trump revoked former President Biden’s executive order on AI, a thin regulatory stanchion that has yet to be replaced. It also highlights issues of surging AI-related energy demand at a time when we do not have the capacity to meet it. But more on that in a minute. 

If the name ‘Stargate’ sounds familiar to you, that’s because it’s not exactly new. The Information reported in March of 2024 that Microsoft and OpenAI were drawing up plans for a supercomputer nicknamed “Stargate” that would cost around $100 billion and contain millions of GPU chips. A month before that, Altman made headlines for embarking on a world tour during which he reportedly sought around $7 trillion worth of semiconductor investments. 

That brings us to the data center problem: There are a few factors at play here. One is the energy cost associated with data centers. In 2023, data centers consumed around 4% of total U.S. energy; the Department of Energy expects that number to surpass 12% by 2028. 

• This surge in data center activity has contributed to a burgeoning public health crisis, in which, according to recent research, “the AI hardware manufacturing process, electricity generation from fossil fuels to power AI data centers and the maintenance and usage of diesel backup generators to ensure continuous AI data center operation all produce significant amounts of criteria air pollutants.” 

• According to statistical modeling, these pollutants could contribute to hundreds of early deaths, thousands of asthma cases and around $20 billion in public health costs in 2030 alone. 

At the same time, the North American Electricity Reliability Corporation (NERC) said recently that “demand growth is now higher than at any point in the past two decades.” And the U.S. isn’t generating enough capacity to meet future demand, indicating “critical reliability challenges” for the U.S. electric grid. 

A recent report from Bloom Energy on data centers in 2025 found that most data center operators expect power availability to worsen, since the U.S. grid “has not been able to keep pace with this demand.” Because of this, data center executives interviewed by Bloom expect 30% of data centers to be powered, at least in part, by on-site energy by 2030, a setup that could include gas turbines (such as those in operation at xAI’s Memphis data center), in addition to alternative energy sources. 

During his first day in office, Trump expanded efforts to increase oil and gas production, saying: “we have something that no other manufacturing nation will ever have: the largest amount of oil and gas of any country on Earth. And we are going to use it.”

I contacted OpenAI for comment regarding the expected power requirements of Stargate, and the ways in which the team plans on meeting those requirements (on-site generation, gas turbines, green energy and battery storage, partnerships with utilities, etc.). The startup did not respond. 

Dr. Srinivas Mukkamala, an AI expert and a member of El Paso Electric Company’s board, told me that Stargate represents a “good move to get what’s needed for AI to be mainstream.”

• “We see what the demand is. Everybody is waiting in line for power and no utility can give them power for the next few years,” he said. “This $100 billion, most of it is going to go into the ground to get the basic infrastructure, power, water and land. They won’t say that — they’ve been very careful how they’re wording it … Without power, there is no AI.”

• He added that, though the project represents an initial investment of $500 billion, “it’s going to get into trillions,” something that, in an environment where AI revenue today represents a drop in the bucket of the cost being laid out for it, re-ensures that the cloud providers will remain the dominant winners of this first round of the AI revolution. 

“Things like this, they’re just gonna get worse. The barrier of entry is going to be impossible for others,” Mukkamala said. “Yes, it creates jobs, (but) for who? You’re reskilling your workers, but they’ll never be wealthy. The wealth will be concentrated in 30 or 40 people. This is an insane amount of money.” 

“This is just helping OpenAI,” he added. “This is an OpenAI thing.” 

He expects the sector to eventually be able to make trillions of dollars, but said that, for these companies to recoup their investments in AI would require two things: time and deregulation. 

“They will go all in to make this a reality,” he said. “That’s the only thing I’m worried about. At what expense? Let us drive AI and don’t let AI drive us. We are humans and this should always be human first.” 

At what cost? 

The entire field of AI is centered around cost-benefit analyses being conducted by a handful of companies who stand to gain a lot of money by uncautiously pushing ahead.

For this to happen without regulation in place to ensure mitigations regarding pollution, energy stability, public health, etc. is a concerning development that functions as only the latest evidence of the sheer might of tech companies today. 

🤔 Your thought process:

Selected Image 2 (Left):

• “This image looks very familiar and often seen.”

Selected Image 1 (Right):

• “I can't believe I got this wrong. I will def not be able to tell well the matrix takes over.”

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

Here’s your view on the barriers to adoption:

Roughly half of you said hallucination and real-world mistakes represent the biggest barrier to adoption. 13% said it’s regulatory compliance; 13% said data and 16% said cost.

Hallucination:

• “Hallucination and mistakes are closely aligned - it boils down to trust. I use ChatGPT a lot but still find myself validating against other sources - which, of course, could be inaccurate themselves.”