Vishal Sikka Raised $32 Million to Build an IT Services Firm That Runs on AI. The Cost Claim Has No Number Yet.

Vishal Sikka ran Infosys until 2017. His new company wants to take business from firms like Infosys.

On Wednesday, that company, Hang Ten Systems, said it raised $32 million in seed funding led by Mayfield, with a strategic investment from Aramco Ventures and a group of angel investors. The board includes Yahoo co-founder Jerry Yang. Hang Ten is based in Palo Alto, and by Mayfield's own account it "just got started a month back."

Start with what it does, because the category is the whole story. Hang Ten calls itself an "enterprise AI services company." Strip the label and it is an IT services and consulting shop: the kind of outfit paid to build, customize, integrate, and maintain the software a large company runs on. That is the work Infosys, TCS, Accenture, and the rest have sold for decades, mostly by assigning a lot of people to it. Sikka is betting AI can do much of that work with far fewer of them.

What the product actually is

Hang Ten's own description, translated:

• Agentic code generation. AI models that write and change code with limited human steering.
• Reusable skills library. Prebuilt routines the agents call instead of rebuilding the same thing on every engagement.
• An "FDE Bench." A pool of forward deployed engineers, meaning humans who sit with the customer. Worth noting, because it means this is fewer people, not no people. The pitch is leverage, not magic.
• Domain focus on enterprise transformations, finance, HR, and new product development.

The company says it can build, change, and run enterprise software "at a fraction of the cost and time." That is the sentence to watch. A fraction of what, measured how, against which baseline? The press release does not say. Sikka does not say. "Faster and cheaper" with no denominator is a marketing claim, not a benchmark, and it is the exact claim every services firm now makes about its own AI tooling, including the ones Hang Ten is trying to unseat.

The customer list is real: Siemens Gamesa Renewable Energy and Fresenius, both named for "AI-native project delivery," with a supportive quote from Siemens Gamesa CEO Vinod Philip. Named logos are a useful signal that someone signed a contract. They are not a result. Nobody has put a number on how much time or money the AI-native model saved on those projects, which is the only number that would tell you whether the model works.

The bet against his old industry

Hang Ten launches into an open argument about whether AI helps or guts the IT services business. Analysts at Jefferies argued earlier this year that IT services could be among the first sectors to feel real AI disruption. Infosys chairman Nandan Nilekani has taken the other side this week, framing "AI-first services" as a $300 billion to $400 billion opportunity by 2030 rather than a threat. The market has been siding with the pessimists for now: Infosys shares are down more than 35% this year.

Sikka has effectively picked the disruption side, and he is selling against the model he used to run. That is a coherent position. It is also a convenient one for a founder raising money, since "the incumbents are doomed and I know their playbook" is a clean pitch to a venture firm.

The VianAI question

This is not Sikka's first enterprise-AI company. After leaving Infosys in 2017, he founded VianAI, which came out of stealth in 2019 with a $50 million seed and raised another $140 million in 2021 in a round led by SoftBank Vision Fund 2. That is roughly $190 million into a startup most people would struggle to describe today.

Mayfield's Navin Chaddha says Hang Ten is a different company aimed at a different market, and on paper that holds: VianAI built analytics and decision tools, while Hang Ten sells services built around code generation. Still, this is the second enterprise-AI swing from largely the same group. The early team includes longtime Sikka collaborators from SAP, Infosys, and VianAI, among them CTO Navin Budhiraja, chief design officer Sanjay Rajagopalan, and forward-deployed-engineering lead Tao Liu.

Why $32 million for a one-month-old company

Thirty-two million dollars is a large seed, and it is not buying a proven model. It is buying a team and a head start. Mayfield, which says it invests "$3 billion in AI companies, with 70% of investments at inception," was explicit about the logic: "At Mayfield, we invest in people first." Chaddha's pitch for the economics is that "traditional services scale linearly with headcount" while Hang Ten's "leverage grows with every project."

Maybe. That is the thesis of nearly every AI-services and AI-coding company funded in the last two years, and the part nobody has shown at scale is the second half: that the leverage holds once real enterprise software, with its compliance rules and twenty years of accumulated exceptions, has to actually run in production without a large bench of engineers babysitting it.

The thing to watch is not the next funding round or the next marquee logo. It is the first time Hang Ten, or one of its customers, publishes a real before-and-after: this project, this many people, this much time, versus what the old way cost. Until then, it is a strong resume, a believable team, and a claim with the number left blank.