Singularity University

Look who dropped by! Ideas are always in motion at Singularity. Consider this your sign to think bigger this weekend.

Let's set aside 10 minutes for a Futures Wheel experiment. Here are some prompts to get you started:

1. Photosynthesis is engineered into human skin
2. Average human lifespan is 150 years
3. Human teleportation becomes possible and widely adopted

Need a refresher on why this matters? Here's a quick example.

When Steve Jobs announced the iPhone, the CEO of a chewing gum company probably wasn't thinking about their bottom line. But the widespread adoption of the smartphone tanked gum sales. Why? Because people usually buy gum while waiting in line at the checkout counter. Once everyone's eyes dropped to their phones, gum sales followed.

The futures wheel trains your brain to trace the ripple effects of new technology before they catch you off guard.

Try it and report back. What innovations are you watching that may surprise others in your industry? Did this exercise take you somewhere interesting? Somewhere crazy?

Comment which prompt you chose and where it led you.

If your robotics strategy starts with "how do we replace workers," you should fire your consultant.

Here's what that framing misses: attempting to automate workers out of a job often increases the overall demand for labor. The unpredictable, messy, human elements of real work are exactly what automation keeps failing to account for. You can't engineer out the judgment call, the improvisation, the moment where a person reads a situation and does something no workflow anticipated.

The organizations getting this right are asking a different question: what are the worst parts of this job, and how do we take those off the person doing it?

80% of the global workforce never sits at a desk. They're in warehouses, on job sites, in operating rooms, on factory floors. Technology has spent decades largely ignoring them while optimizing for the office. Robotics is one of the few tools that can actually reach them, and used well, it can take the physically dangerous, repetitive, injury-causing work off their plates while making them more capable, not less employed.

A surgeon in New York operating on a patient in a rural village in real time. Drones handling warehouse inventory so nobody spends a shift counting shelves. Robotic systems managing energy infrastructure at speeds no human operator could match.

The technology exists. The question is what problem we're pointing it at.

The 2026 FIFA World Cup kicks off tomorrow, and the SU community spans both sides of the opening match. Our partners SingularityU Mexico Summit and Singularity South Africa are repping their nations. Tag an SU alum from or in the comments and let's connect! Where are you watching from?

AI hyperscalers are willing to pay 8 to 80 times more for compute capacity than for the electricity required to generate it.

The consequence: they are effectively outbidding ordinary consumers for power. Tech companies are purchasing existing nuclear and coal plants just to keep their data centers running. Over 30 data centers currently cannot break ground because there is no available energy to grant them permits.

The energy grid is fragile and slow to expand. Building new transmission infrastructure takes up to 15 years.

This goes beyond supply chain issues for tech. It is a geopolitical and economic constraint that will shape which regions host the infrastructure behind AI, which businesses can access affordable compute, and which governments have to make hard choices about who gets power.

Energy access is becoming a competitive moat. Is your leadership team thinking about it that way?

Data centers are the hidden backbone of the digital economy, and their proliferation is accelerating in ways that have serious implications for business, geopolitics, and democratic governance.

In this conversation, Sharron will take us inside data centers: what they actually are, how they operate, and why the next generation looks very different than what exists today. From offshore platforms to orbital infrastructure, the physical home of our data is moving.

She'll then trace the thread from data center proliferation to AI advancement to Earth observation technology, and raise a question every leader should be thinking about: as the capacity to gather, store, analyze, and transmit data scales exponentially, who controls it, who benefits from it, and what does that mean for democratic institutions and global equity?

Join the Discussion Series this Thursday.
Link to Register: https://us06web.zoom.us/webinar/register/4917805964188/WN_3E9L72PRTLy9WJCc43hNZQ

Data centers now consume 6 percent of all electricity in the United States. Two years ago that figure was considerably lower. The jump is almost entirely driven by AI.

The biggest facilities now draw as much power as small cities. Global data center consumption has hit 67.7 gigawatts, a 36 percent increase in two years. The US accounts for 43 percent of that total. In Germany, data centers consume 9.5 percent of national electricity. In the UK, 5.8 percent.

A new report from the International Data Center Authority identifies a threshold: significant community and political pushback tends to begin once data centers surpass 5 percent of a country's power supply. The US crossed that line.

Hundreds of state-level bills to regulate data centers have been introduced. In Maine, the legislature passed a bill halting construction of large data centers until 2027 before the governor vetoed it. Developers in Northern Virginia's Data Center Alley, already the densest concentration of facilities in the world, cannot launch new projects until 2032 due to energy scarcity.

Water is equally contested. A single large facility can consume as much water daily as 6,500 households.

There is also waste baked into the existing system. An estimated 13 percent of US cloud consumption comes from abandoned test environments and unused applications that continue drawing power around the clock without doing anything useful.

Annual global data center spending is approaching $1 trillion, with up to $700 billion anticipated in the US alone this year. The industry shows no sign of slowing.

Whether the grid can absorb what's coming, and how hard communities push back, may determine whether the AI boom continues or runs into a wall it can't compute its way through.

Read the full story, link in comments.

The AI your organization builds is only as good as the data behind it.

Alix Rübsaam challenges leaders to scrutinize the underlying data and biases of any AI model they build, use, or sell. A decade ago, Amazon trained a hiring algorithm on their historical recruitment data. The model learned to favor men, because their previous hiring practices already did. They were flagged, course-corrected, and repeated the pattern.

This is garbage in, garbage out made visible.

When a dataset reflects historical decisions that don't actually align with the problem you're trying to solve, the model inherits those distortions and amplifies them at scale. The technology doesn't introduce the bias. It crystallizes it.

Rübsaam arms leaders with the questions that cut through this:
"How would the data we're using influence the outcome?"
"What do I know about the data we're using, and what do I not know?"
"Who or what is included in the data, and who or what isn't, and why?"
"How does my own background influence how I interpret the outcome?"

And then the harder layer of self-assessment: do you know the answer to these questions? Do you know who does? Do you know where to find out?

AI literacy starts long before the model is deployed. It starts with the data, and with the leaders willing to ask uncomfortable questions about it.

What question from this list would be hardest for your organization to answer honestly?

Large Language Models do not store facts. They predict them. These systems are probabilistic, not deterministic. They function as giant statistical maps, generating the next most likely word based on patterns across billions of data points. As Tricia Wang puts it, the intelligence in these models is extremely narrow. At their core, LLMs are highly advanced auto-suggest tools.

If an AI has been trained on sensitive data, or simply has access to it in its context window, it can unpredictably surface that information as a probabilistic output. This happens by design, a direct result of how these systems fundamentally work.

AI cannot be trusted to naturally safeguard sensitive data it has been exposed to. Those who don't understand this are making decisions with an incomplete mental model of the technology guiding their organizations.

Too many are already implementing AI without asking the foundational questions first. Before going further, Neil Desai urges executives to answer: what growth do you honestly need to achieve?

Traditional business education trains leaders to analyze their competitive landscape within a very fixed, narrow industry vertical, which leaves them highly vulnerable to disruptive changes coming from unexpected places.

Enter The Futures Wheel: a strategic foresight tool used to systematically map out the non-obvious first-, second-, and third-order consequences of a new technology or trend.

This exercise forces teams to break out of mechanistic, cause-and-effect thinking, allowing them to spot potential disruptions and identify net-new category opportunities before they become mainstream.

While most people are talking about AI hallucinations and deepfakes, not enough people are talking about these 5 risks:

1 - Agentic Accountability.
When your AI co-worker makes a decision, who signed off - and could you actually get the audit trail in twenty-four hours?

2 - Human-In-The-Loop.
Reviewers cannot evaluate at the speed AI produces.

3 - Shadow Agentic Deployment.
Employees are wiring agents into production through no-code tools, without IT or risk awareness.

4 - Embedded AI In Your Supply Chain.
Which of your vendors are now training on your customer data, and when did procurement sign for it?

5 - Director Liability.
Your D&O policy was not written with autonomous AI in mind. Coverage of AI-driven failures is contested and untested.

Agentic AI isn't waiting for your governance framework to catch up. It's already making decisions, acting autonomously, and leaving a trail your legal team may not be able to follow.

It doesn't matter how your LLM performed. If you're the one that deployed it, you'll be on hook for its actions.

Introduction to AI: Here, Real, and Yours to Lead with Kellie Nuttall

You can build the most powerful AI tool in the world. If your people are not using it, it is like a treadmill sitting in your garage.

That is the insight Dr. Michael "House" Housman brought to this month's SU Discussion Series, and it is one that every leader navigating the AI transformation needs to hear.

MIT research shows that 90 to 95% of AI implementations fail in production. Not because the technology is wrong. Because of change management, lack of buy-in, and a workforce that is afraid of what comes next.

In this session, Dr. Housman draws on 15+ years building and deploying machine learning platforms across startups and Fortune 500 companies to cover:

- Why most AI initiatives stall, and how to build momentum that lasts
- How to run an "AI census" in your organization
- How to build a true AI-first culture at every level
- What the road to AGI means for the decisions you make today

Watch the full discussion now. Link in comments.