News related to
Deployment

Deliveries of Model 3 and Model Y vehicles built at Tesla’s Shanghai factory reached 79,478 units in April, including exports to Europe and other overseas markets. The figure was 36% higher than a year earlier, marking a sixth straight month of year-on-year growth, though it was 7.2% lower than March as the market moved through a seasonally softer month after a stronger end to the first quarter.

The April performance signals that Tesla is regaining some stability after a difficult stretch in 2025, when competition, pricing pressure and consumer fatigue weighed on its standing in major electric vehicle markets. Shanghai remains central to that recovery because it supplies the domestic market while also serving as a major export base for Europe, Asia-Pacific and other regions. The plant produces Tesla’s two highest-volume models, leaving its monthly output closely watched as a gauge of both local demand and overseas momentum.

Tesla sold 85,670 China-made vehicles in March, giving it a stronger base heading into the second quarter. April’s decline from that level suggests the rebound remains uneven, but the year-on-year rise points to healthier order flow than at the same point last year. The recovery has been helped by adjustments to production, export allocation and the refreshed Model Y cycle, while Tesla has continued to use financing incentives and targeted promotions to defend its share.

China’s electric vehicle market remains the world’s most competitive, with new energy vehicles accounting for a growing majority of passenger-car retail sales. April industry estimates pointed to new energy vehicle retail sales of about 860,000 units and a penetration rate above 60%, underscoring how quickly battery-electric and plug-in hybrid models are displacing petrol vehicles. That shift benefits Tesla’s broader addressable market, but it also exposes the company to fierce price competition from domestic manufacturers with faster model cycles and wider product ranges.

BYD remains Tesla’s most formidable rival, with a broad line-up spanning mass-market cars, plug-in hybrids, premium models and exports. Geely, Chery, Leapmotor, Xpeng, Li Auto, Nio and Xiaomi are also pressing deeper into the market, competing not only on price but on software, cabin technology, driving-assistance features and charging convenience. Several of these manufacturers are expanding overseas, putting pressure on Tesla in Europe and other export markets that were once more heavily dominated by the US company.

Tesla’s challenge is sharpened by the age of its core passenger line-up. The Model Y and Model 3 remain strong global sellers, but rivals are launching fresh sport-utility vehicles, sedans and compact models at a faster pace. Tesla has been working on lower-cost products and software-led upgrades to protect volumes, but investors and customers are watching closely for evidence that the company can broaden its appeal beyond its two main models.

Europe has become a key test of whether Shanghai’s output can translate into renewed international growth. Tesla sales have improved in several European markets after a weak 2025, helped by stronger demand for electric vehicles and a recovery from earlier supply disruptions. Yet the company faces a more crowded field, with BYD, Xpeng and other manufacturers gaining visibility through competitive pricing, longer equipment lists and expanded dealer networks.

Regulatory approval for advanced driver-assistance technology remains another variable. Tesla’s Full Self-Driving package is central to its long-term positioning, but approvals in China and Europe have moved cautiously. Wider deployment could support margins and brand differentiation, though regulators continue to scrutinise safety, data handling and driver-monitoring requirements. Without a faster software rollout, Tesla’s sales recovery will depend more heavily on vehicle pricing, product updates and production efficiency.

Tesla’s first-quarter global deliveries stood at 358,023 vehicles, with Model 3 and Model Y accounting for 341,893 units. That concentration highlights the importance of Shanghai’s April result: the plant is not merely a regional asset but a major pillar of the company’s global volume strategy. Stronger China-made deliveries can ease pressure on other factories, support export flexibility and help Tesla manage inventory across markets with uneven demand.

Economists led by Chetan Ahya have drawn a parallel with electric vehicles, where early policy support, component localisation and aggressive production expansion turned China from a fast follower into the world’s dominant exporter. Their research projects China’s share of global manufacturing could rise to 16.5 per cent by 2030, from about 15 per cent now, as advanced automation and embodied artificial intelligence move from demonstration projects into factory and logistics use.

The bullish case rests on China’s widening lead in humanoid robot production. Industry estimates show the country produced about 12,800 humanoid robots in 2025, close to 90 per cent of global output, though the market remains tiny compared with conventional industrial robotics. China also remains the world’s largest industrial robot market, accounting for more than half of global factory robot installations in 2024, with an operational stock above 2 million units.

That foundation matters because humanoid robots require the same manufacturing strengths that supported China’s rise in solar panels, batteries and electric vehicles: precision motors, sensors, actuators, batteries, control systems, machine vision and low-cost assembly. Local companies are building ecosystems around these components, while provincial governments are offering subsidies, industrial parks and procurement support to accelerate commercial deployment.

Beijing has identified humanoid robots as a strategic industry, setting targets for mass production and global leadership while integrating robotics into wider plans for artificial intelligence and advanced manufacturing. The 15th Five-Year Plan framework is expected to place AI-powered robots near the centre of industrial upgrading, although large-scale commercialisation is more likely toward the later part of the plan period than immediately.

The sector’s momentum has drawn major private capital. Start-ups such as Unitree, UBTech, AgiBot and Fourier Intelligence have moved quickly from stage demonstrations to pilot factory and warehouse deployments. Linkerbot, a fast-growing maker of dexterous robotic hands, is seeking a valuation of about $6 billion after building a strong position in high-degree-of-freedom robotic hands used in humanoids and industrial systems. Its modular products point to an important feature of the market: some of the earliest profits may come from components rather than complete humanoid machines.

Automakers and electronics groups are also moving into the field. Xpeng has said humanoid robotics could become a long-term investment area alongside electric vehicles, while other manufacturers are testing robots for inspection, sorting, materials handling and repetitive shop-floor tasks. The logic is clear. China faces an ageing workforce, rising labour costs in coastal manufacturing hubs and growing pressure from tariff barriers. A successful humanoid robotics industry could help preserve competitiveness even as low-cost labour advantages fade.

The export implications are broader than robot sales. If humanoids improve productivity in Chinese factories, they could lower production costs across electronics, machinery, appliances and consumer goods. That would reinforce the country’s position in global supply chains at a time when the United States, Europe and several Asian economies are trying to reduce dependence on Chinese manufacturing. It could also create a new export category in robots, components, software and maintenance services.

Yet the constraints remain substantial. Humanoid robots are still expensive, often have limited battery life and struggle with unstructured environments. Many can perform scripted tasks in controlled settings but fall short when asked to handle the speed, judgement and dexterity of human workers. Even executives in China’s robotics industry acknowledge that factory humanoids remain less efficient than people in many jobs. Safety certification, liability rules, data protection and workplace integration will also affect adoption.

Dependence on high-end chips and advanced software ecosystems is another weakness. US restrictions on AI semiconductors have pushed Chinese firms to localise more hardware, but training and deploying capable embodied AI systems remains compute-intensive. Without steady access to cutting-edge chips, progress may be uneven, particularly in models that require real-time perception, movement planning and continuous learning.

Bengaluru-based Pixxel and Sarvam have set out plans to build Pathfinder, a 200 kg-class orbital data centre satellite that would test whether artificial intelligence workloads can be processed directly in space rather than routed first through terrestrial cloud systems.

The mission, announced on 4 May 2026, is scheduled to reach orbit as early as the fourth quarter of 2026. Pixxel will design, build, launch and operate the satellite, while Sarvam will provide the artificial intelligence stack needed for training and inference on board. The partners describe the project as India’s first orbital data centre satellite, placing it at the junction of three strategic races: sovereign AI, commercial space infrastructure and the search for lower-latency computing systems.

Pathfinder is intended to carry data centre-grade graphics processing units alongside Pixxel’s hyperspectral imaging camera. That combination would allow the satellite to capture high-fidelity Earth observation data and analyse it in orbit using foundation models, instead of sending large volumes of raw imagery to ground stations for processing. The aim is to transmit insights rather than just data, cutting the time between observation and decision-making.

The approach could be valuable for environmental monitoring, agriculture, mining, resource management, disaster response and critical infrastructure tracking. Hyperspectral imaging can detect patterns invisible to conventional cameras by reading hundreds of spectral bands, making it useful for identifying crop stress, pollution, mineral signatures, water conditions and changes in land use. Adding on-board AI would make such systems faster and potentially more autonomous.

Pathfinder also reflects the pressure building around AI infrastructure. Data centres have become increasingly constrained by energy demand, land availability, cooling requirements and regulation. AI-focused facilities are especially power-intensive because training and running large models depend on specialised chips. Industry forecasts point to data centre electricity use crossing 1,000 terawatt hours by 2030, with AI workloads accounting for a growing share of consumption.

Orbital computing is being explored as one answer to that problem, though it remains technically difficult and commercially unproven at scale. Satellites can draw continuous solar power in certain orbits and process space-generated data close to its source. Yet the model must overcome radiation exposure, heat dissipation, launch costs, repair limitations, chip reliability, bandwidth management and the challenge of operating high-performance hardware in an environment far harsher than a ground-based server hall.

For Pixxel, the project marks a move beyond Earth observation into space infrastructure. The company has built its position around hyperspectral satellites and already operates Firefly satellites capable of capturing detailed environmental data. Pathfinder would test whether those sensing systems can be paired with in-orbit computing to create a more responsive intelligence layer.

For Sarvam, the mission extends its sovereign AI pitch beyond ground infrastructure. The company has focused on building language models and platforms designed for deployment in local contexts, including multilingual use cases. Running its models aboard an India-built satellite would give the company a high-profile demonstration of how AI sovereignty can include hardware location, data control and infrastructure independence, not just model development.

The announcement comes as global technology groups examine space-based compute. Google’s Project Suncatcher has explored satellite constellations fitted with AI accelerators and optical links, with prototype missions planned with Planet. Elon Musk has also spoken publicly about the possibility of data centres in orbit within the wider SpaceX and Starlink ecosystem. These ideas remain at an early stage, but they show that AI infrastructure is beginning to stretch beyond conventional cloud regions.

Domestic activity is also widening. Agnikul Cosmos and NeevCloud have explored related orbital infrastructure concepts, indicating that private space and compute companies see a possible market in distributed cloud systems that span Earth and orbit. Defence, climate intelligence, telecommunications and sovereign data processing are likely to be among the first areas of interest if the technology matures.

Pathfinder’s immediate importance lies less in replacing terrestrial data centres and more in validating whether orbit can host meaningful AI workloads. The satellite is expected to test power management, thermal control, real-time inference, data processing and model performance under operational space conditions. Success would give Pixxel and Sarvam a foundation for more advanced orbital compute systems, while failure would expose the engineering and economic limits of the concept.

The project also fits a broader shift in satellite design. Spacecraft have traditionally acted mainly as sensors and transmitters, collecting data for later analysis on Earth. Pathfinder points towards satellites that can interpret what they observe, prioritise what should be sent down and support faster decisions for governments and commercial users.

Commercial adoption will depend on cost, reliability and demand. Ground-based cloud systems remain cheaper, easier to maintain and vastly more scalable. Orbital compute will need to prove that its advantages — lower latency for space data, resilience, solar access and sovereignty — justify the complexity of placing advanced AI hardware in space.

For now, Pathfinder gives the space and AI sectors a test case with wider implications. It places Pixxel and Sarvam among early movers trying to turn satellites from data collectors into active computing nodes, while testing whether the limits facing AI infrastructure on Earth can be partly addressed above it.

The Google Cloud India AI Hub, being developed with AdaniConneX and Nxtra by Airtel, is designed as a gigawatt-scale private cloud and data centre network across three campuses near the port city. The investment is planned over 2026-2030 and marks Google’s biggest infrastructure commitment in the country, combining hyperscale computing, subsea cable connectivity and clean-energy-linked operations.

Andhra Pradesh Chief Minister N. Chandrababu Naidu laid the foundation stone for the project on April 28, 2026, alongside Union electronics and IT minister Ashwini Vaishnaw, state IT and HRD minister Nara Lokesh, Google global infrastructure vice-president Bikash Koley and senior representatives of Adani and Bharti Airtel. The project follows Google’s October 2025 announcement that Visakhapatnam would host its first gigawatt-scale AI hub in the country.

The development is expected to occupy about 600 acres across locations including Tarluvada, Rambilli and Adavivaram. State officials have projected the facility’s operational launch around September 2028, though large data centre campuses of this scale typically move through phased commissioning as power, cooling, network and security systems are completed.

Google’s plan goes beyond a conventional cloud facility. The hub is intended to support AI workloads for services such as enterprise cloud computing, consumer applications, advanced model deployment and data-intensive digital platforms. Its location on the east coast gives Visakhapatnam a strategic role in subsea connectivity, with plans for international cable landings that can improve bandwidth, reduce latency and strengthen links between India, Southeast Asia and global cloud regions.

AdaniConneX, the joint venture between Adani Enterprises and EdgeConneX, is expected to lead major elements of data centre construction and power-linked infrastructure. Nxtra by Airtel will support data centre development, fibre connectivity and cable landing infrastructure. Airtel’s existing telecom backbone and Adani’s energy and infrastructure capabilities give the project domestic execution partners at a time when AI infrastructure is becoming capital-intensive and geopolitically sensitive.

The project also fits into a wider policy push to reduce dependence on overseas compute capacity. Demand for AI infrastructure has expanded rapidly as startups, banks, telecom operators, public agencies and manufacturing groups test generative AI, automation and data analytics. Large language models and AI inference services require dependable access to graphics processing units, high-density power systems, liquid cooling, resilient connectivity and local data storage.

Visakhapatnam’s selection reflects Andhra Pradesh’s attempt to position itself as a technology and logistics corridor after years of competition from established hubs such as Bengaluru, Hyderabad, Chennai, Pune and the National Capital Region. The port city already has an industrial base, naval presence, road and rail connectivity and a coastline suited to cable infrastructure. The state government is banking on the Google project to attract hardware suppliers, energy providers, construction contractors, cloud specialists and AI startups.

Employment claims around the project have been ambitious, with state leaders citing the possibility of hundreds of thousands of direct and indirect jobs through construction, operations, supply chains and associated services. The more immediate employment impact is likely to come from civil works, electrical systems, fibre deployment, facility management, security and engineering support. Higher-skilled roles in cloud architecture, AI operations and chip-level systems may depend on how quickly local training pipelines are aligned with industry needs.

Power and water use will be closely watched. Gigawatt-scale data centres require enormous electricity supply and sophisticated cooling systems, making renewable energy procurement, grid stability and water stewardship critical. Google has said the project will be tied to a long-term clean energy strategy and community programmes covering water, skills, maritime trade modernisation and women-led businesses. Delivering on those commitments will be essential for public acceptance, especially as data centres worldwide face scrutiny over their environmental footprint.