🚀 The Startup Khan Newsletter

September 17, 2026 — Hang Ten Raises $53M as AI Starts Rewriting the Economics of Consulting

🧠 Top Story: Former Infosys CEO Vishal Sikka Adds $53M to Hang Ten Systems

What happens when you rebuild a consulting firm around AI from day one?

Investors are putting serious money behind that experiment.

Hang Ten Systems has raised an additional $53 million seed financing, just five weeks after closing its initial $32 million round.

That takes total funding to $85 million only four months after launch.

The new financing was led by Xora, the early-stage investment platform backed by Temasek, with participation from Mayfield and Aramco Ventures.

The company’s investor base also includes Intel CEO Lip-Bu Tan, Micron CEO Sanjay Mehrotra and Yahoo co-founder Jerry Yang, who has joined Hang Ten’s board.

And the founder brings serious enterprise pedigree.

Vishal Sikka previously served as CEO of Infosys and CTO of SAP.

What Hang Ten Is Actually Building

This isn’t another foundation-model startup.

Hang Ten is targeting massive enterprises—typically companies generating more than $10 billion in annual revenue—and helping them:

Develop AI strategy → Modernize legacy software → Deploy agentic development → Redesign enterprise workflows

Its approach combines agentic code generation, reusable AI skills and human domain expertise.

Hang Ten says it has already secured multiple seven-figure customer contracts.

Those customer economics are company-reported, but the financing itself—and the speed at which it arrived—are the bigger signal.

Why founders should care

The common AI-consulting thesis is:

Give consultants AI tools → Consultants become more productive.

But Hang Ten suggests something more ambitious:

What if you rebuilt the consulting company itself?

Traditional consulting:

Large human team → Months of analysis → Custom implementation

AI-native consulting:

Small expert team + Agents + Reusable intelligence → Faster implementation

If that model works, the economics could look radically different.

Higher revenue per employee.

Faster delivery.

Smaller teams.

More reusable IP.

Potentially better margins.

👉 Founder takeaway: Don’t only use AI to improve the existing business model.

Ask whether AI lets you design a completely different business model.

---

🔊 Funding Watch: Treble Raises $18M to Build the Sound Layer for Physical AI

AI doesn’t only need eyes.

It needs ears that work in the real world.

Icelandic startup Treble Technologies has raised $18 million in a Series A-2 led by Paladin Capital Group, with KOMPAS VC, Frumtak Ventures and the European Innovation Council Fund participating.

The company has now raised approximately €36 million in total.

Treble builds physics-based acoustic simulations and synthetic audio data for developers working on:

Voice AI → Smart glasses → Headphones → Robots → Vehicles → Drones

The company names Amazon and Logitech among its customers.

Why this matters

Yesterday we covered AI infrastructure around:

Compute. Networking. Security. Agent orchestration.

Treble exposes another layer:

Simulation.

A voice model trained in a clean laboratory still has to understand someone speaking:

Inside a moving car.

Beside industrial machinery.

In a crowded restaurant.

Outside in heavy wind.

Through a tiny wearable microphone.

You could collect real recordings for every possible environment.

Or you could simulate them.

Model environment → Generate synthetic data → Train AI → Test AI → Deploy physically

That’s potentially a massive picks-and-shovels category.

👉 Founder takeaway: Before physical AI can operate reliably in the real world, it needs increasingly realistic synthetic worlds in which to learn.

---

🧪 Trend Watch: The Simulation Stack Is Getting Interesting

Physical AI creates a fundamental problem.

Failure in software can mean an error message.

Failure in a robot, vehicle or industrial system can mean damaged equipment—or worse.

So intelligent physical systems need places where they can fail cheaply.

Millions of times.

That creates an emerging infrastructure stack:

Physics simulation

Synthetic video

Acoustic simulation

Digital twins

Robot training environments

Evaluation systems

Think of simulation as the testing environment for physical intelligence.

The better the synthetic world becomes, the more learning can happen before deployment touches reality.

👉 The physical-AI boom may create an enormous software market before the robots themselves reach mass adoption.

---

💰 Valuation Watch: One-Month-Old Emulate Reportedly Approaches $4B

Investor appetite for elite AI research talent continues to stretch traditional venture logic.

British startup Emulate, founded in August by former Google DeepMind researchers Jack Parker-Holder, Matthew McGill and Philip Ball, is reportedly discussing a financing of as much as $700 million at a valuation approaching $4 billion.

The founders previously worked on DeepMind’s Genie world-model research.

But an important qualification:

The financing has not closed.

Terms could still change.

Founder signal

At this stage, investors aren’t underwriting conventional startup metrics.

They’re underwriting:

Founder pedigree + Research capability + Scarce talent + Future model potential

That’s effectively venture capital competing for intellectual capital before the business fully exists.

But enormous valuation doesn’t eliminate the fundamental requirements.

Eventually every company still needs:

Product → Customers → Distribution → Economics

👉 Founder reminder: A valuation is financing evidence—not product-market-fit evidence.

---

🧰 Tool of the Day: Langfuse

As companies deploy more AI agents and complex workflows, understanding what those systems actually did becomes increasingly important.

Langfuse provides observability, tracing, evaluation and prompt-management infrastructure for LLM applications.

That becomes useful when one customer request triggers multiple agents, model calls, tools and retrieval steps.

Useful for:

• Agent tracing
• LLM observability
• Evaluation
• Prompt management
• Cost tracking
• Debugging complex AI workflows

👉 Best use: When “the AI gave a bad answer” isn’t enough information to figure out what actually broke.

Complex AI systems need observability just like complex software systems do.

---

📊 Trend Check: AI-Native Means Redesigning the Economics

There’s an important difference between:

AI-enabled

and

AI-native.

An AI-enabled consulting firm gives existing employees copilots.

An AI-native consulting firm asks whether the existing staffing model should exist at all.

An AI-enabled robotics company collects more training data.

An AI-native infrastructure company asks whether most of that data should be simulated instead.

This distinction matters across industries.

Don’t ask:

“Where can we add AI?”

Ask:

If the answer is no, there’s probably an opportunity.

---

🌟 Founder Spotlight: Attack the Cost Structure

One of the best places to search for AI startup ideas is inside an industry’s cost structure.

Pick a market and identify its three biggest expenses.

For consulting, that might be:

Human labor → Delivery time → Custom implementation

For physical AI:

Real-world data collection → Testing → Failed physical experiments

Then ask:

Which of these costs does AI fundamentally change?

Not by 10%.

By 10x.

👉 That’s where AI stops being a feature and starts becoming a new company architecture.

---

🔴 Founder Opportunity Radar

🍁 TODAY — Toronto Startup Pitch Night

Startup Valley Toronto Pitch Night runs tonight, September 17, bringing founders, operators and investors together for startup pitches and networking.

For Toronto-area founders, this is a same-day networking opportunity rather than an accelerator deadline.

🚀 TOMORROW — Startup World Cup Philadelphia Regional

The Philadelphia Regional Startup World Cup competition takes place September 18 at Temple University.

Ten selected technology startups will deliver three-minute pitches followed by two minutes of Q&A.

Applications are already closed, so this is now primarily an attendance and networking opportunity.

🇫🇮 September 23 — Pitch What’s NEXT with Anthropic

Maria 01 and Anthropic are hosting Pitch What’s NEXT with Anthropic on September 23 in Helsinki.

Selected startups pitch before an expert jury, receive Claude-related benefits and can earn an opportunity to appear alongside Anthropic at a Slush side event in November.

👉 Founder reminder: A startup event doesn’t need to produce a trophy to be valuable. One customer, investor or senior hire can justify the room.

---

⚡ Quick Win: Run an AI-Native Reset

Take your company’s core workflow.

Pretend your current product and organization don’t exist.

Then ask:

“If we started today, knowing AI agents were available from day one, how would we build this?”

Would you need the same headcount?

The same workflow?

The same software stack?

The same pricing model?

The same customer onboarding?

The same service delivery?

Where the answers change dramatically, you’ve found something worth investigating.

---

🧩 Final Thought

Hang Ten and Treble represent two very different AI opportunities.

Hang Ten: Rebuild a human-heavy industry’s economics around AI.

Treble: Build infrastructure required for physical AI to operate reliably.

One attacks the business model.

The other attacks the technology stack.

But the principle is the same.

The strongest AI-native companies won’t simply bolt intelligence onto yesterday’s product.

They’ll ask why yesterday’s economics, workflows and infrastructure should exist at all.

Don’t add AI to the old architecture.

Redesign the architecture around AI.

— The Startup Khan

Your 5-minute startup edge — funding news, tools, and founder playbooks.