India's solar manufacturing push meets Texas capacity, Indonesian cells and a memory squeeze

Commercial output on the two new lines began on 29 October 2026, pv magazine reported on 3 November. Each line adds 800 MW of annual module capacity, taking the Brookshire plant to 3.2 GW. Waaree Energies' board approved the expansion in April 2025.

For readers tracking India's technology manufacturing policy, the Texas move is a signal about where Indian manufacturers now put capital. It is less an American factory story. Waaree is an Indian company building abroad to serve a protected market. Its subsidiary is also expanding an Arizona module plant from 1 GW to 1.6 GW. The company says this should bring total US module capacity to 4.8 GW in the Indian fiscal year running from 1 April 2026 to 31 March 2027, pv magazine reported. That is a lot of new capacity aimed at one buyer.

Why an Indian module maker is building in Texas

Waaree's own explanation points at trade policy. The company said the expansion comes amid changes in US trade policy, including measures under Section 232. Those measures affect the economics of imported solar modules, according to pv magazine's account. Sunil Rathi, a director of Waaree Group, is quoted in that report saying shifting trade policy has strengthened the case for local manufacturing.

"With Brookshire now at 3.2 GW and Arizona expanding, we are serving American customers from American factories, and our US order book continues to grow every quarter," Rathi said.

The financial picture in the same report is more concrete than the strategy language. Revenue from Waaree's US manufacturing operations rose to INR 13 billion, about $146 million, in the second quarter of fiscal 2027. That was up from INR 7.32 billion a year earlier. The US domestic business accounted for 43.8% of the company's INR 592 billion order book, against roughly 36% at the end of the preceding quarter. Consolidated revenue from operations rose 23.9% to INR 75.18 billion in the same quarter.

Waaree is not only building abroad. It plans to lift solar cell capacity in India from 5.4 GW to 15.4 GW in fiscal 2027. It also plans to commission a 10 GW ingot and wafer plant in fiscal 2028, pv magazine reported. Those upstream steps matter because cells and wafers are the parts of the chain India has imported most heavily.

Chinese cells, made in Indonesia

The same day brought a reminder that the cell layer is still contested. Yingfa Ruineng, a Chinese cell manufacturer, unveiled a 7.5 GW plant in Indonesia producing n-type TOPCon cells, pv magazine reported on 3 November. It is the company's first overseas manufacturing base, alongside three in China. The four sites together have a combined effective capacity of 51.5 GW, including 45.5 GW of n-type TOPCon, according to company figures cited in the report.

The Indonesian plant is pitched as a regional supply hub for Southeast Asia and the Middle East. It is also pitched as support for Indonesia's 100 GW solar programme. The company's statement, quoted by pv magazine, frames local cell production as a way for module makers to shorten supply routes. It also helps them meet Indonesia's local content requirement, known as TKDN, subject to certification and regulatory rules.

That logic is familiar to anyone following India's production-linked incentive schemes. Local content rules pull assembly closer to the buyer. Cell and wafer capacity follows only when the economics or the rules force it. Yingfa Ruineng ranked second for global solar cell shipments in the first half of 2026, behind Tongwei, according to the research firm InfoLink, as cited by pv magazine.

The memory bill arrives

Not every manufacturing cost story is about solar. Tom's Hardware reported on 3 November that Apple faces what its CEO, John Ternus, called "unprecedented" memory price increases. Apple expects to pay more for memory in the December quarter, citing Reuters. The same report says Apple's manufacturing bill has climbed by $33 billion in just a few months.

"What we're seeing with memory pricing right now is unprecedented. And as we said earlier in the September quarter, we did see higher memory costs as compared to the June quarter, which were partly offset by the benefit of carrying inventory and lower costs on certain non-memory components. Now for the December quarter, we expect even higher memory costs, partly offset by a few factors."

Apple has absorbed part of the shock by using memory bought at lower prices. It has also cut costs elsewhere and raised prices on some products, including Mac and iPad at the end of June and iPhone in September, according to the Tom's Hardware account of Ternus's remarks. The company expects memory purchases to become even more expensive into 2027.

The driver named in that report is AI data centres consuming memory and storage supply. For India, the read-across is direct. Any device assembly or electronics manufacturing incentive that depends on imported memory now carries a cost line that is moving against the plan.

Chips that design chips

The demand side of that equation was on display in Shanghai. At a science forum on Friday 30 October, Chen Yunji, co-founder of AI chipmaker Cambricon Technologies, predicted that AI and chip design would converge until the design process is automated, the South China Morning Post reported.

"In the [next] 10 years, the most important development ... is the RSI (recursive self-improvement) techniques for artificial intelligence, where AI models improve AI models, and AI chips improve AI chips," Chen said, according to the Post.

That is a prediction, not a procurement plan. But it sits alongside the memory squeeze as part of the same supply story. More AI compute means more demand for the components that go into everything else.

India's networks, and the uplink problem

Inside India, the constraint may be radio rather than factories. Light Reading reported on 29 October that daily multimodal AI use in India is projected to rise from 22% now to 43% by 2030. The global forecast is 30%, citing Ericsson ConsumerLab research. That traffic is uplink-heavy. India's 5G networks were built mainly for downlink volume on mid-band spectrum, particularly 3.5 GHz, according to Arun Menon of MTN Consulting, quoted in the same article.

"The networks are ready for AI but they have to be optimized as per the particular requirements of the specific AI application," Nitin Bansal, Ericsson's managing director for India and head of networks for South East Asia, Oceania and India, told Light Reading.

Siddhant Cally, a senior analyst at Counterpoint Research, told the outlet that consistently responsive experience at scale will need better uplink performance. It will also need more site fibre and smarter spectrum use, not just incremental radio upgrades. Reliance Jio has built a 5G standalone network. Airtel has only recently begun moving to standalone, Light Reading reported.

What the policy file actually shows

Set these stories side by side and the shape of India's manufacturing push looks less like a single national programme. It looks more like a set of bets at different layers. Module assembly is scaling, in Texas and in India. Cell and wafer capacity is being added at home and abroad by Chinese and Indian firms. Components that India does not make, memory above all, are getting more expensive. Networks need uplink investment before AI traffic arrives in volume.

Two older items fill in the background. India launched its NVS-03 navigation satellite on 30 October aboard a GSLV Mark II, Space.com reported. It is the third next-generation satellite for the NavIC constellation. ISRO's launch brochure describes NavIC as a regional system serving India and about 1,500 km beyond its landmass. The constellation currently has three fully working satellites and needs four for position, navigation and timing services. That is why the commissioning of NVS-03 will be watched closely.

IEEE Spectrum, in a 28 October feature, described Beta Technologies as one of the top three US electric aircraft makers. The company aims first at regional cargo and medical transport before the passenger market. That is a US story, but it is the kind of capital-intensive hardware programme that Indian states have been courting.

One caution about the freshest numbers. The Waaree capacity figures come from the company and were reported by pv magazine. The revenue and order book figures come from the same report. The $33 billion manufacturing bill figure for Apple comes from Tom's Hardware's account of Reuters reporting, not from a filing. The Ericsson projection is survey-based research, not measured traffic. Each is useful, and each has a different weight.

The common thread is timing. Waaree's Texas lines started on 29 October. Yingfa Ruineng's Indonesian plant was presented on 3 November. Apple's memory warning covers the December quarter and extends into 2027. India's solar and electronics ambitions will be tested against all three at once. The cheapest lever, imported components, is the one getting more expensive.