In 2014, Infosys hired Vishal Sikka to save the company from technological disruption. He was the first non-founder CEO in the company's history — a product visionary recruited from SAP, where he had spent 12 years building enterprise software, to drag India's second-largest IT services firm into the future.
It didn't work. By 2017, Sikka had resigned amid a bitter governance dispute with the founders. The vision of an AI-first Infosys died with his departure.
On June 24, 2026, Sikka launched Hang Ten Systems — an AI-native enterprise services startup that aims to do exactly what Infosys once paid him to prevent: replace the headcount-driven IT services model with one built on agentic code generation, reusable AI skills, and compound leverage that grows with every project instead of every hire.
The startup has raised $32 million in seed funding led by Mayfield, with a strategic investment from Aramco Ventures and participation from angel investors including Yahoo co-founder Jerry Yang, who sits on the board. It already has enterprise customers, including Siemens Gamesa Renewable Energy and Fresenius.
The timing is not coincidental. One week before Hang Ten's launch, Accenture suffered its worst single-day stock decline on record — a 20% crash that erased billions in market capitalization after the company cut revenue guidance and reported its lowest quarterly new order bookings in nearly two years. The Nifty IT index is down 29% for calendar year 2026. Infosys shares have fallen over 35% this year.
The $1.8 trillion IT services industry is watching the market price in a structural question it has been avoiding for two years: does the headcount-driven services model survive AI?
Hang Ten is betting the answer is no. Here is why that matters — and the two frameworks enterprise leaders need to evaluate what comes next.
The Structural Problem With IT Services in 2026
The traditional IT services business model is elegant and simple: enterprises outsource technology work — customizing software, integrating systems, maintaining infrastructure — and services firms deliver it by deploying armies of engineers. Revenue scales linearly with headcount. Margins come from labor arbitrage. Growth comes from more projects requiring more people.
This model built a $1.8 trillion industry in 2026 and made Infosys, TCS, Wipro, Accenture, and the Big Four consulting firms among the most valuable professional services organizations on earth.
AI breaks this model at the foundation.
When 80-90% of a software implementation can be agent-driven — and industry analysts are already documenting this in production environments — the linear relationship between headcount and revenue output is structurally impaired. The "entire GSI industry is based on" customizations that agents now increasingly obviate, as one analyst put it in the Matterfact newsletter's coverage of the IT services disruption debate.
Mayfield managing partner Navin Chaddha, who led Hang Ten's funding round, captured the thesis in one sentence: "Traditional services scale linearly with headcount. Hang Ten is built so its leverage grows with every project."
This is the compound-leverage model that AI-native startups are betting on: every project generates reusable AI skills, every automation compounds, and the ratio of output to human headcount widens with every engagement. A SaaS company that required 50 people in 2022 can now be built by a team of 8 to 12 with AI augmentation. Apply that same ratio to IT services delivery, and the economic structure of a $1.8 trillion industry starts to look very different.
The Accenture Warning Shot
The theoretical disruption argument became a market event on June 18, 2026.
Accenture reported Q3 FY2026 results that beat EPS expectations but terrified investors. Revenue came in at $18.7 billion — a 6% increase year-over-year — but new bookings dropped 2%, and the company guided Q4 revenue of $17.75 billion to $18.40 billion, below the $18.47 billion Wall Street expected. The stock crashed 20% in a single day — the worst decline in Accenture's history as a public company.
The ripple effect was immediate. Indian IT stocks suffered a ₹1.35 lakh crore ($16.2 billion) single-day meltdown. TCS fell 2.9%, Wipro 2.8%, Infosys 2.7%. Jefferies maintained its underweight stance on the entire IT services sector, warning that Accenture's outlook increases the risk of further earnings estimate cuts.
The debate that followed was revealing. CLSA argued that the guidance cut reflected macro weakness — Middle East tensions, subdued discretionary spending — more than AI disruption. But the market was pricing in something deeper: the structural fear that generative AI is replacing the fundamental need for traditional IT services at the margin.
S&P Global titled its sector analysis "Disruption Without Dislocation... Yet." The operative word is yet.
The Two Camps: Nilekani vs. Sikka
The most fascinating dimension of the Hang Ten launch is the intellectual debate it creates between two people who know Infosys better than almost anyone alive.
Nandan Nilekani, Infosys co-founder and chairman, told shareholders at the company's AGM this week that AI-first services could represent a $300 billion to $400 billion market by 2030. He revealed that Infosys has crossed $1 billion in annualized AI services revenue. His thesis: AI will expand the addressable market. Enterprises need more help, not less, to adopt AI — and IT services firms are best positioned to deliver that help.
Vishal Sikka, former Infosys CEO and now founder of Hang Ten, is betting on the opposite outcome. His thesis: AI doesn't expand the services TAM — it restructures it. The value shifts from labor-intensive delivery to AI-augmented leverage, and the winners will be AI-native firms that build compound advantages, not traditional firms that bolt AI onto a headcount model.
Both positions are internally consistent. Both are backed by evidence. And the answer matters enormously for enterprise buyers who are deciding where to place their next $10 million, $50 million, or $100 million services contract.
Jefferies' analysis earlier this year argued that IT services may be among the first sectors to face meaningful AI disruption. The investment bank maintained its underweight on the sector. But Nilekani's $300-400 billion figure is not fantasy — enterprise AI adoption requires massive amounts of integration, training, change management, and domain expertise that pure-play AI tools cannot deliver alone.
The resolution likely lies somewhere in the middle: the total market may grow, but the margin structure and delivery model within it will be unrecognizable within five years. That is the bet Hang Ten is making.
Who Is Hang Ten, Actually?
Hang Ten is not Sikka's first post-Infosys venture. After leaving in 2017, he founded VianAI, which emerged from stealth in 2019 with $50 million in seed funding and later raised $140 million in a 2021 round led by SoftBank Vision Fund 2. VianAI focused on enterprise AI applications and analytics tools.
Hang Ten is different. Where VianAI built tools, Hang Ten delivers services — but with an AI-native delivery model. The company describes itself as an enterprise AI services company built around three pillars:
- Agentic code generation — AI agents that write, test, and deploy enterprise software
- Reusable AI skills — every engagement produces reusable capabilities that compound across future projects
- Domain expertise — deep enterprise knowledge from a team that has collectively spent decades inside SAP, Infosys, and Fortune 500 environments
The founding team is a reunion of Sikka's inner circle:
- Navin Budhiraja, CTO — worked with Sikka across SAP, Infosys, and VianAI
- Sanjay Rajagopalan, Chief Design Officer — same career trajectory
- Tao Liu, SVP Forward Deployed Engineering — responsible for on-site enterprise delivery
The "forward deployed engineering" title is telling. It mirrors the model pioneered by Palantir — engineers embedded inside enterprise customers who combine domain context with technical capability. But where Palantir embeds humans who write software, Hang Ten embeds humans who orchestrate AI agents that write software.
Mayfield told TechCrunch that the company "just got started a month back" and already has customers — a velocity that would be impossible in traditional IT services, where sales cycles for enterprise contracts typically run 6-18 months.
Framework #1: Traditional vs. AI-Native IT Services — Decision Matrix
Enterprise leaders evaluating their next services engagement need a framework for comparing traditional and AI-native delivery models. Here is how they stack up across eight dimensions:
| Dimension | Traditional IT Services | AI-Native Services (Hang Ten Model) | Winner |
|---|---|---|---|
| Delivery cost structure | Linear: cost scales with headcount | Compound: cost per project decreases as AI skills accumulate | AI-Native |
| Time to first delivery | 3-6 months (scoping, staffing, ramp) | Weeks (AI agents + embedded engineers) | AI-Native |
| Domain expertise depth | Deep (decades of enterprise knowledge) | Growing (founder pedigree, but limited history) | Traditional |
| Scale and global coverage | Proven at 500K+ employee scale | Unproven beyond early customers | Traditional |
| Regulatory and compliance support | Established governance, audit trails | Nascent, untested in regulated environments | Traditional |
| Innovation velocity | Slow (consensus-driven, risk-averse) | Fast (small team, AI-augmented decision-making) | AI-Native |
| Vendor lock-in risk | Moderate (switching costs are high but manageable) | Unknown (new model, new dependencies) | Depends |
| Long-term cost trajectory | Flat to rising (labor costs increase) | Declining (AI compound effects reduce marginal cost) | AI-Native |
When to choose traditional: Regulated industries (banking, healthcare, defense) where compliance history matters more than speed. Multi-year transformations requiring 500+ person teams with proven global delivery. Engagements where the CIO's primary concern is risk mitigation.
When to evaluate AI-native: Greenfield AI implementations where speed matters. Projects where the 80-90% agent-driven delivery model can apply. Engagements where the CFO's primary concern is cost per outcome, not cost per engineer.
When to hedge: Run a parallel pilot. Give the AI-native vendor a well-scoped $500K-$2M project alongside your traditional services partner's larger engagement. Measure cost per feature, time to delivery, and defect rate. Let the data decide.
Framework #2: Enterprise AI Services Vendor Evaluation Checklist
Before engaging any AI-native services provider — Hang Ten or its emerging competitors — enterprise procurement teams should evaluate along these 12 criteria:
Capability Assessment (Score 1-5)
1. AI Delivery Model Maturity
- Does the vendor use AI agents in production delivery, or just in demos?
- Can they show evidence of compound leverage (reusable skills from prior engagements)?
- What percentage of delivery work is agent-driven vs. human-driven?
2. Domain Expertise Depth
- Does the team have verifiable enterprise software experience (SAP, Oracle, Salesforce, ServiceNow)?
- How many enterprise customers have they delivered to in production?
- Can they provide customer references with measurable outcomes?
3. Security and Governance
- How do they govern AI agents during delivery? (See our coverage of the OWASP Agentic Top 10)
- Do AI-generated code artifacts undergo human review before production deployment?
- How is customer IP and data protected during AI-augmented delivery?
4. Contractual Structure
- Is pricing based on outcomes (features delivered, systems integrated) or inputs (hours, headcount)?
- What happens when AI agents make errors in production? Who bears liability?
- Is there a transition plan if the engagement needs to move to a traditional services model?
Risk Assessment (Score 1-5)
5. Vendor Viability
- How much runway does the vendor have? (Seed-stage = high risk)
- Who are the investors, and do they have enterprise services track records?
- Is there a credible path to profitability, or is this a growth-at-all-costs play?
6. Concentration Risk
- Is the vendor's expertise concentrated in a few key individuals?
- What is the bus factor? Can delivery continue if the CTO or CEO is unavailable?
- How much institutional knowledge exists vs. individual heroics?
7. Regulatory Readiness
- Has the vendor delivered in regulated environments (SOX, HIPAA, PCI-DSS)?
- Can they produce audit trails for AI-generated deliverables?
- How do they handle data residency requirements?
8. Integration Complexity
- Can their delivery model integrate with existing enterprise toolchains (Jira, ServiceNow, Azure DevOps)?
- How do their AI agents interact with legacy systems that lack APIs?
- Is there a clear handoff process for ongoing maintenance?
Strategic Assessment (Score 1-5)
9. Intellectual Property
- Who owns AI-generated code and configurations?
- Are the "reusable AI skills" derived from your project shared with competitors?
- How is IP assignment documented in the master services agreement?
10. Talent and Knowledge Transfer
- Does the engagement include training your internal teams?
- Can your engineers learn to operate the AI-augmented delivery tools independently?
- Is there a path to eventual in-sourcing?
11. Scalability
- Can the vendor scale from a pilot ($500K) to a transformation ($50M)?
- What is the ceiling on their current delivery capacity?
- How quickly can they onboard new domain expertise?
12. Exit Strategy
- How difficult is it to transition away from the vendor?
- Are deliverables in standard formats that other providers can maintain?
- What is the switching cost if the engagement fails?
Scoring: 48-60 = Strong candidate for pilot engagement. 36-47 = Proceed with caution and strong contractual protections. Below 36 = High risk, consider traditional alternative.
The Bigger Picture: What Happens to 1.5 Million IT Services Jobs?
The question nobody at the Infosys AGM or the Hang Ten launch event wants to answer directly is what AI-native services means for employment.
The Indian IT services industry alone employs approximately 5.4 million people. The top five firms — TCS, Infosys, HCLTech, Wipro, and Tech Mahindra — collectively employ over 1.5 million. If the agent-driven delivery model achieves even half of its projected efficiency gains, the math on future headcount is uncomfortable.
Nilekani's counter-argument — that AI expands the addressable market so total employment grows — has historical precedent. ATMs were supposed to eliminate bank teller jobs, but the number of bank tellers actually increased because ATMs made it cheaper to open new branches. Cloud computing was supposed to eliminate infrastructure jobs, but it created an entirely new category of cloud engineering roles.
But the ATM analogy has a limit. ATMs automated one task (cash withdrawal) in an otherwise human-intensive job. AI agents automate the core task — writing software, integrating systems, maintaining codebases — that defines the entire value proposition of IT services.
The honest answer is that both outcomes will happen simultaneously. The total market may grow to Nilekani's $300-400 billion, but the composition of that market will be radically different. Fewer engineers doing more customization work. More AI orchestrators managing more complex systems. And a new class of AI-native services firms — Hang Ten being the first prominent example — that compete on leverage rather than headcount.
What Enterprise Leaders Should Do Now
-
Pressure-test your current services contracts. Ask your Accenture, Infosys, or Wipro account team exactly how AI is changing the delivery model on your current engagements. If the answer is "we're using Copilot" — that is not a strategy. Ask what percentage of delivery is agent-driven, what reusable skills have been generated, and how the cost-per-outcome compares to 24 months ago.
-
Run a controlled pilot with an AI-native vendor. Pick a well-scoped project — a system integration, a data migration, a custom application build — and run it with an AI-native vendor alongside your traditional services partner. Compare time, cost, quality, and defect rate. Let the data break the tie.
-
Rethink your procurement framework. Traditional IT services procurement evaluates on hourly rates, headcount commitments, and geographic coverage. AI-native services require evaluation on outcome pricing, delivery velocity, IP ownership, and compound leverage. Your procurement framework was built for a world that is disappearing.
-
Watch the Accenture earnings trajectory. The June 2026 crash was a single data point. If Q4 confirms the trend — declining bookings, compressed guidance, accelerating AI-driven efficiency gains cannibalizing traditional delivery — the market will reprice the entire sector. That repricing creates both risk and opportunity for enterprise buyers with long-term services commitments.
-
Follow the founder talent. Sikka is not the only senior IT services executive launching an AI-native alternative. Watch where the next wave of founder talent comes from — former partners at Accenture, Deloitte, and McKinsey who see the same structural shift. The pattern of experienced operators building AI-native challengers will accelerate.
The Bottom Line
Vishal Sikka's career is a narrative arc for the entire IT services industry. He was hired to drag a traditional services firm into the AI era. He couldn't do it from inside. Now he is building the alternative from outside.
Hang Ten Systems may succeed or fail — $32 million in seed funding and two early customers do not make a market winner. But the thesis it represents is not going away. AI-native services delivery is not a product demo or a research paper. It is a business model that is now funded, staffed, and selling to Fortune 500 companies.
The $1.8 trillion IT services industry built its empire on a simple equation: more work requires more people. For the first time, that equation has a credible challenger.
The man who was hired to prevent this moment just made it real.
Continue Reading
- OpenAI Just Put 150 Engineers Inside Your CFO Office
- 5 Acquisitions in 14 Days. The AI Agent Stack Is Being Bought.
- PwC Trained 30,000 on Claude. Big 4 AI Stack Just Reset.
- EY Just Put Agentic AI in Front of 130,000 Auditors
- The Agent Sprawl Crisis: 96% Deployed AI Agents. Only 12% Can Govern Them.
Rajesh Beri is Head of AI Engineering at Zscaler, focused on enterprise AI security, governance, and deployment at scale.