Hang Ten’s $85M Seed Bet Says 30-Person Software Teams Are Finished
Thirty-person software teams are not a delivery model. They are an invoice. Hang Ten just raised $85 million to prove two to four people can do the work.
A four-month-old company has raised $85 million to attack one of corporate Australia’s favourite rackets: putting 30 people on a software project that should not need 30 people.
Vishal Sikka’s Hang Ten Systems says AI can let teams of two to four deliver work that once took roughly 30. If that holds up in production, plenty of IT services firms are about to discover that “headcount” was never a business model. It was just expensive camouflage. ([techcrunch.com](https://techcrunch.com/2026/09/16/former-infosys-chiefs-ai-startup-adds-50m-to-seed-weeks-after-initial-raise/?utm_source=openai))
Hang Ten just raised $53 million five weeks after its first seed round
On September 16, Hang Ten Systems announced an additional $53 million in seed funding led by Xora, Temasek’s early-stage investment platform. Mayfield and Aramco Ventures also participated. That came only five weeks after a first $32 million seed close, taking total funding to $85 million. ([techcrunch.com](https://techcrunch.com/2026/09/16/former-infosys-chiefs-ai-startup-adds-50m-to-seed-weeks-after-initial-raise/?utm_source=openai))
That is not normal seed-stage behaviour. It is a very loud market signal.
The company was founded in May by Sikka, the former CEO of Infosys and former SAP executive. It is not trying to be another frontier-model lab. Hang Ten sells advisory, transformation and applied-AI delivery to large enterprises, particularly businesses with more than $10 billion in annual revenue. In plain English: it wants to help giant companies decide what to build, modernise their clunky systems and deliver the software using AI-heavy workflows. ([techcrunch.com](https://techcrunch.com/2026/09/16/former-infosys-chiefs-ai-startup-adds-50m-to-seed-weeks-after-initial-raise/?utm_source=openai))
The investor list matters because it tells you what sort of company this intends to be. Xora brings the Temasek connection. Mayfield has backed it from the start. Aramco Ventures is there. Intel CEO Lip-Bu Tan and Micron CEO Sanjay Mehrotra are investors. Yahoo co-founder Jerry Yang has joined the board. That is a cap table built for enterprise introductions, technical credibility and global sales—not for posting clever demos on social media. ([techcrunch.com](https://techcrunch.com/2026/09/16/former-infosys-chiefs-ai-startup-adds-50m-to-seed-weeks-after-initial-raise/?utm_source=openai))
Hang Ten says it already works with Fresenius Kabi, Saudi Aramco and Siemens Energy, has secured multiple seven-figure contracts, and is chasing eight-figure deals. Sikka told TechCrunch one customer signed a multimillion-dollar contract within 25 days of their first meeting. Anyone who has sold software to a global enterprise knows why that last number raises an eyebrow. Enterprise procurement can take longer than some marriages. ([techcrunch.com](https://techcrunch.com/2026/09/16/former-infosys-chiefs-ai-startup-adds-50m-to-seed-weeks-after-initial-raise/?utm_source=openai))
The real product is not AI. It is accountability.
Most companies do not have an AI problem. They have an implementation problem.
They have executives who have watched a chatbot summarise a document and suddenly think the whole company should be “AI-first.” Then they hand the work to a consulting firm, which runs workshops, writes slides, builds a proof of concept that nobody uses, and invoices everybody involved until Christmas.
Hang Ten’s pitch is more useful than that. Its team says it combines agentic code generation, a reusable skills library called Hobie, and deep domain knowledge to deliver actual production software. The promise is not simply faster code. It is moving the bottleneck away from typing code and toward defining the right requirement, validating the output and safely deploying it inside a real business. ([techcrunch.com](https://techcrunch.com/2026/09/16/former-infosys-chiefs-ai-startup-adds-50m-to-seed-weeks-after-initial-raise/?utm_source=openai))
That distinction is enormous.
Writing code has always been only part of a software project. The painful parts are usually deciding what the business actually needs, untangling legacy systems, dealing with data quality, navigating compliance, winning internal approval and owning the result when it breaks at 2am.
AI can reduce the cost of producing code. It cannot magically turn a confused executive team into a decisive one. It cannot make rotten internal data clean. It cannot make a bank, hospital or energy company stop caring about security and certification.
So the winning firms in this next phase will not be the ones saying, “Look, our agent can write Python.” Every kid with a laptop can say that now. The winners will take responsibility for a commercial result: a working system, deployed safely, with a clear owner and a number attached to the outcome.
Why Infosys, Accenture and the services crowd should pay attention
The traditional IT-services model has a structural weakness: it often earns more when more humans spend more time on the job.
That does not mean every consultant is useless. Far from it. Big companies need people who understand their ugly old infrastructure and know how to ship safely. But the old arithmetic is under pressure when a small, capable AI-native team can do a meaningful share of the delivery work.
Hang Ten says some engagements can be handled by two to four people where a 30-person team might previously have been required, while customers or third parties still conduct final quality checks and certification. It is targeting a tenfold improvement in cost, speed, or a combination of both. Those are company claims, not proven industry law—but they explain why investors moved quickly. ([techcrunch.com](https://techcrunch.com/2026/09/16/former-infosys-chiefs-ai-startup-adds-50m-to-seed-weeks-after-initial-raise/?utm_source=openai))
Here is the brutal bit: even if Hang Ten achieves only half that improvement, the old model has a problem.
A client that can get a decent outcome in half the time, at materially lower cost, will not be sentimental about preserving an oversized vendor team. Nor should it be. Businesses exist to create value, not to fund elaborate calendars full of status meetings.
For large incumbents, this is not necessarily a death sentence. They have distribution, trusted relationships, armies of domain experts and huge delivery capacity. But they now face a choice. They can use AI to genuinely shrink delivery teams and pass some savings to customers, or they can pretend nothing has changed and wait for nimbler competitors to price them into irrelevance.
The overlooked angle: Hang Ten may be building a better consultancy, not a software company
This is where founders and investors need to keep their heads screwed on.
An $85 million seed round is impressive. A famous founder is impressive. A few fast enterprise wins are impressive. None of that proves a durable software business.
Hang Ten’s model still contains a large services component: advisory, transformation and custom software delivery. Services can be a fantastic business. They can throw off serious cash, build trusted relationships and create an excellent wedge into giant customers. But services businesses scale differently from pure software businesses. Revenue often rises with experienced people, and experienced people are not infinitely available.
The company’s challenge is to turn what it learns from each engagement into repeatable intellectual property: reusable workflows, tested AI skills, implementation playbooks, industry-specific guardrails and deployment machinery that gets better every time. Hobie is important precisely because it could be part of that productisation layer. ([techcrunch.com](https://techcrunch.com/2026/09/16/former-infosys-chiefs-ai-startup-adds-50m-to-seed-weeks-after-initial-raise/?utm_source=openai))
If Hang Ten merely hires clever people to build bespoke systems faster, it can still become a very valuable services firm. Nothing wrong with that. But it will not earn software-like economics simply because AI is in the slide deck.
If it converts delivery experience into a system that lets a small team repeatedly produce reliable outcomes across regulated industries, then it becomes far more interesting. That is the prize investors are buying into.
The other uncomfortable truth: cheap code makes judgement more valuable
There is a silly narrative floating around that AI makes technical talent less important.
No. AI makes average output cheaper. It makes good judgement more valuable.
When generating a first version of software becomes fast and cheap, the scarce resource becomes knowing what should be built, what cannot fail, where the risks live, which shortcuts are safe and which ones will cost you a fortune six months later.
That is good news for capable operators. It is bad news for passengers.
The people who win will be able to frame a problem clearly, make decisions with incomplete information, understand their customers, and hold a high bar for execution. They will use AI to remove drudge work, not to outsource thinking.
What this means for you
If you are a founder, stop bragging about how much AI your product contains. Customers do not buy AI. They buy faster revenue, lower costs, fewer mistakes or less risk. Pick one. Put a number on it. Then own the result.
If you run an operating business, find one software process where you are paying for a bloated external team or months of internal delay. Do not launch a grand “AI transformation.” Pick a painful workflow, define the commercial result, appoint one accountable executive and run a 90-day test.
If you sell services, audit your delivery model before your clients do it for you. Ask a nasty question: if we started this company today with AI available, would we still staff projects this way? If the answer is no, fix it while you still have the relationship.
And if you invest, do not confuse a large round with a moat. The question is whether a company has built repeatable distribution, trusted customer access and a delivery system that improves with every job. Money gets attention. Repeatability gets rich.
Hang Ten has bought itself a serious chance to prove that enterprise AI can do more than make demos look slick. Now comes the hard part: delivering software that works, at a cost incumbents cannot match, without becoming just another consulting firm wearing a black t-shirt.