Goldman Sachs forecasts global economy acceleration in 2027 due to falling oil prices and AI development

A recent conversation with the head of a small creative agency in a second-tier Indian city whose clients are mostly outside India showed that the influence of artificial intelligence (AI) goes beyond simply speeding up content production. Instead, AI has automated processes such as rotoscoping, routine editing, and parts of project management. As a result, the small team is now taking on projects that previously required hiring external freelancers.

The author believes something more fundamental is happening: AI is transforming the requirements for the scale of a company needed to provide complex services to the global market. For India, this presents both a significant opportunity and several difficult questions.

India's service exports reached $421.3 billion in fiscal year 26, up from $387.5 billion the previous year, with IT and business services being the largest contributors. For three decades, the IT services industry was based on a simple formula: attract more clients and hire more employees. AI is challenging this model.

Coding, testing, research, documentation, and customer support form the backbone of Indian outsourcing, and AI is showing the fastest progress in these areas. According to NASSCOM, in fiscal year 26, India's technology industry generated $10–12 billion from AI services, while total revenue grew by 6.1% to $315 billion. Although hiring continues, the total number of employees has only increased by 2.3%, weakening the link between revenue and headcount.

This shift threatens the model built on selling human potential. However, when viewed from the perspective of a five-person company, the picture looks completely different.

Small service firms have always faced a capacity problem: to win a large project, they had to build an organization capable of executing it. Now, AI is beginning to decouple capability from headcount. The author personally observed this in the work of the founder of a marketing firm with 15 people and 22 AI agents that serves clients in over seven countries. The founder categorizes every task into three types: manual tasks, such as initial introductory calls, are intentionally kept outside the scope of AI; in AI-assisted tasks, AI performs 85–95% of the work, and a human reviews the rest; fully automated tasks, such as market research, are executed end-to-end, with the only human step being sending the report to the client. Furthermore, about 40% of the firm's budget is directed towards AI agents and software, not personnel. The founder notes that the margin is 60–70%, significantly higher than the 20–25% for a typical agency.

Even the initial outreach process has changed. One client from the UK told the marketing firm's founder that they were able to find, evaluate, and close a deal with the firm within a single chat in Perplexity.

Using cross-border payments shows that thousands of Indian businesses receive payments from abroad, and it is noteworthy how small they are. In the last 12 months, 347 individual specialists earned over $100,000 from foreign clients, and 117 teams of two to ten people earned over $250,000. The exporter base is also expanding: the number of exporters actively invoicing foreign customers quarterly has grown approximately fivefold in seven quarters—from about 2,800 to 13,900. Part of this growth is due to more companies joining the network, not just the emergence of new exporters. Nevertheless, the median invoice remained around $1,500, indicating that many more people are selling rather than a few securing large deals. Eighty-two percent of these exporters have ten or fewer employees.

Exporters in AI, machine learning, data, or automation have grown by about nine times, and their median invoice is almost double. They account for about 2% of the volume, so this is currently more of a trend than a sustainable trend. One of them, a solo AI and machine learning consultant, earned about $266,000 from US clients over 12 months from just four buyers. This is atypical, but ten years ago, such work would have required a full-fledged firm.

If India's answer to AI is merely performing the same outsourcing work with fewer people and lower prices for clients, it may increase productivity but will not fundamentally improve the country's position in the global value chain. A more significant opportunity lies in shifting from selling production capacity to selling expertise and results.

AI can analyze 500 customer conversations, but it cannot independently decide what problem a client's future business should be built around. It can generate software, but someone still needs to understand what exactly needs to be created, how it fits into the client's organization, and take responsibility in case of failure. Value is increasingly found in the ability to ask the right questions, apply domain knowledge, exercise judgment, and take responsibility for the final outcome.

This can make deep expertise even more valuable, not less. Consider an Indian specialist who spent 15 years studying insurance claims, semiconductor design, pharmaceutical regulation, supply chains, or financial risks. Such a person can now combine their industry knowledge with tools that significantly expand an individual's capabilities. India does not lack such specialists. According to the Economic Survey for 2024–25, the country provides 28% of the world's STEM workforce and 23% of global talent in software development. New types of activities are already emerging. The author gives a vivid example: a healthcare worker in Kochi spends a week labeling organs and marking anomalies on fetal ultrasound scans. She is not treating patients directly; she is helping to train AI systems developed for hospitals in the US.

There is one problem. India's export structure was designed for firms with finance departments. Export documentation, GST tax filing, foreign currency conversion, and bank reconciliation assume the presence of a back-office employee handling these matters. A five-person firm invoicing clients in California and London in multiple currencies does not have such a specialist. Yet, it can move hundreds of thousands of dollars across borders annually. If AI reduces the team needed to do the work, but the founder still spends hours every month on paperwork and regulatory compliance, the bottleneck simply moves. Regulators, banks, and the broader fintech ecosystem need to make cross-border compliance as easy as the businesses they serve.

Future Indian service giants may not look like giants at all. India's first era of service exports was built by consolidating millions of skilled people into large organizations. This model will not disappear. But the next phase may be much less concentrated: thousands, if not many more, specialized Indian enterprises selling their expertise to the world. The agency discussed is interesting not because AI helps it edit videos faster. It is interesting because a handful of people in an Indian city can now compete for work anywhere in the world. Twenty-five years ago, India's breakthrough in service exports was partly about proving that global work could be done from India. The opportunity opened by AI is about proving that you no longer need to create a large company to do a large amount of work.

Despite tensions in the Middle East or the oil and gas crisis, the Indian economy is demonstrating resilience. Agencies such as the IMF and the World Bank have highly rated the Indian economy. Recently, good news came from abroad for the Indian economy: the rating agency Moody's increased India's growth forecast to 7%.

India proved to be better prepared for shocks in the global energy supply chain caused by the war in Iran. Thanks to this, Moody's adjusted the country's growth forecast for the fiscal year ending in March 2027, raising it from 6.5% to 7%.

Moody's noted that significant economic growth, strong domestic demand, and improved investment activity are helping India cope with external pressure and global turmoil. The report indicates that these strong macroeconomic foundations will allow the economy to maintain growth of around 7% in the medium term. This revised forecast came after the Indian economy demonstrated a higher growth rate in the quarter from April to June 2026, reaching 7.8%.

In its latest report, Moody's explained the forecast increase by stating that India handled the energy price shock better than previously expected. The country successfully diversified its sources of crude oil and gas supplies, which reduced the risk of disruptions in the global energy market. Furthermore, support from fuel sales companies and tax reductions helped consumers avoid the immediate impact of rising energy prices.

Despite the favorable prospects for the Indian economy, Moody's highlighted several risks that could affect the growth trajectory. The report states that rising price pressures could lead to a decline in domestic consumption. There is also a possibility of the weak southwest monsoon affecting agricultural production and weakening rural demand. Moreover, tight monetary policy could slow down the recovery of private investment, and inflation dynamics will be closely monitored in the near future.

The rating agency also pointed to signs of potential improvement in private sector investment. High-frequency indicators, such as good output and capital goods imports, rapid growth in industrial lending, and increased capacity utilization, suggest that India is moving towards a new cycle of private capital expenditure. Sustained growth in private investment, combined with consumption and government infrastructure spending, could provide an additional source of support for economic growth.

Anthropic disclosed in an update to its S-1/A prospectus, submitted to the U.S. Securities and Exchange Commission (SEC), that it may be close to creating an artificial intelligence capable of accelerating its own enhancement.

Although the company states it has not yet reached this phase, its projections suggest this could happen in the near future. This technology, known as Recursive Self-Improvement (RSI), describes a point where AI systems can improve their own development processes, boosting the evolution of their capabilities.

This perspective generates divergence: critics warn that such advances could result in extremely powerful, human-out-of-control systems capable of causing harm while pursuing their objectives. On the other hand, enthusiasts see RSI as a chance to increase productivity and speed up scientific discoveries.

Anthropic acknowledges both views. In the document sent to the SEC, the company mentions that it already uses AI in its R&D activities but warns that this approach carries a significant safety risk, especially if the pace of progress is drastically accelerated.

The prospectus update introduces new warnings in the risk factors section. Anthropic revealed that it has already used AI systems to create new models without realizing at the time the magnitude of the capacity gains achieved. The increased use of AI in research may also cause models to evolve faster than the company's ability to implement safety, alignment, and protection measures.

According to the document, internal estimates and forecasts for future models indicate that the company may soon reach what its Responsible Scaling Policy calls 'dramatically accelerated AI-attributable progress.' However, the company emphasizes that none of its current models have been classified as having exceeded this limit, and data evaluation is ongoing.

Despite this possibility, Anthropic declares that it does not foresee using Claude to autonomously design and build successive generations of advanced models without relevant human supervision or understanding. The company also plans to revise its Responsible Scaling Policy to ensure humans maintain control over the development of its most sophisticated models.

It is important to note that the filings with the SEC fulfill disclosure obligations but also serve as a means for the company to present its expectations to the market. Anthropic stresses that Claude's capabilities have improved this year, notably in programming, mathematics, and cybersecurity, although no current model has been considered by the company to have reached the stage of recursive self-improvement.