🤖 Data Centers to Consume 20% of US Electricity by 2035
Report projects US data center electricity consumption will quadruple by 2035, reaching one-fifth of the country's total power generation. The forecast is 83% higher than the last estimate, reflecting AI's accelerating pace. Key grids are already strained, with electricity prices up 76% over the past year.

🤖 Deepfake Fraud Has Surpassed $3.7 Billion in Losses
Deepfake fraud losses have reached $3.7B globally. Social media accounts for 47% of losses, as fake celebrity videos promoting fraudulent investment schemes spread virally. Impersonation fraud adds another $911M, while fake job candidates have cost businesses $100M.

🤖 Apple Set to Launch Device Subscription Service July 28
Apple is preparing to launch a hardware subscription program called Apple Upgrade as soon as next week. Most iPhone, Mac, iPad, and Apple Watch models will be available through 24 to 36-month leases, with options to upgrade, keep, or return devices. The service arrives amid significant Apple price hikes.

📌 Trump's AI Safety Chief Quits After Just Three Months
Chris Fall, head of the Center for AI Standards and Innovation, has resigned after only three months in the role, further disrupting the Trump administration's AI strategy. Fall was responsible for coordinating safety testing of frontier models with Anthropic, OpenAI, Google DeepMind, and Microsoft.

📌 Most Companies Don't Trust Their Own AI Agents
Despite 86% of enterprises deploying AI agents in production, only 34% trust the actions those agents take. Reasons: poor data quality, weak governance, and rushed deployment. 75% organizations in "agentic chaos" are still pushing more agents, risking compliance fines, customer loss, and operational downtime.

📌 Small Businesses Stuck in AI Experimentation Mode
UK small businesses have barely deepened their AI adoption over three years, with the average number of tools used per company rising only from 1.4 to 1.6. Just 17% of smallest firms report extensive AI use. The main barrier seems to be simple lack of compelling business use cases for AI.

Today's opportunities are brought to you by RightSide.

ASIC Verification Engineer | Contract | Pyramid Tech | Remote
Senior Desktop Support Tech | Full Time | Philadelphia, PA | In Person
Program Integrator | Full Time | Sterling, VA | In Person
Master Solution Architect | Full Time | Reston, VA | In Person
Server Infrastructure Engineer | Full Time | Dearborn, MI | In Person
Senior Copilot Engineer | Full Time | Washington, DC | In Person
Distinguished Mobile Engineer | Full Time | McLean, VA | In Person

There is a version of AI adoption that looks like ambition and functions like panic. A new tool arrives, the pressure to use it builds, and before anyone has asked the right questions, the project is live and underdelivering. The numbers behind this pattern are striking. In 2025, global enterprises invested $684 billion in AI initiatives and by year end, over 80% of that investment had failed to deliver its intended business value. However, the approach seems to be the real problem, not the tech.

What the data consistently shows is that AI project failures are leadership failures first and technology failures second. Across myriads of enterprise initiatives, leadership issues drove most of failures, including unclear success metrics, weak executive sponsorship, and a tendency to treat AI as a pure technology exercise rather than a strategic one. The organizations that struggled did not lack access to capable models or sufficient budgets. They lacked clarity on which problems actually justified AI investment, and operational readiness to absorb what came next.

The organizations that succeed tend to share one counterintuitive habit. They start smaller than feels ambitious. A Stanford analysis of 51 successful enterprise AI deployments found that 61% of those projects included a prior failure, and that failure only converted into learning under specific conditions: defined ownership, measurable outcomes, and a clear separation between the pilot and the scaling decision. The lesson is not that failure is good. It is that contained, intentional experimentation is structurally different from a big bet made under pressure.

Volkswagen's Cariad project is the cautionary tale worth knowing. Launched in 2020 with the goal of building one unified AI-driven operating system for all 12 VW brands simultaneously, the project collapsed under its own ambition, producing a 20-million-line codebase riddled with bugs, delaying major vehicle launches by over a year, and ultimately cutting 1,600 jobs. The failure was not technical. It was strategic overreach: attempting to replace legacy systems, build custom AI, and design proprietary hardware all at once, when starting small and iterating would have preserved both momentum and options.

The wisest frame for AI deployment is deceptively simple. Treat it as a real discipline with real prerequisites, not a shortcut to outcomes you have not yet defined. That means knowing what problem you are solving before choosing a tool, ensuring the data feeding your system is clean and governed before trusting its outputs, and measuring adoption and error reduction rather than just celebrating launch. The organizations getting this right are not necessarily the most technologically sophisticated. They are the ones who asked harder questions before they started, and stayed honest about the answers.

5,100

years ago, ancient Sumerians in Mesopotamia pressed the world's earliest recorded transaction using proto-money onto a clay tablet, recording a receipt for 29,086 measures of barley distributed over 37 months.

Last week’s winning choice:
What is your biggest time-saver when using AI at work?
📝 CONTENT CREATION: Drafting emails, reports, and copy (34.56%)

Got a friend who values AI, tech, and everything in between? Forward them today's read!