Amazon opens Alexa+ in India with Hindi and English support

Emerald AI, Google and NVIDIA have launched the AI Energy Management Alliance, or AEMA, to coordinate work between computing companies and the power industry. Josh Parker described the initiative in NVIDIA’s announcement.

The premise is that a data centre need not draw a fixed amount of electricity regardless of grid conditions. It could shift computing workloads, discharge storage or use generation paired with the facility when the wider system is under strain.

Measuring flexibility before connection

AEMA’s principles call for clear obligations before a facility connects: how it handles brief disturbances, reduces demand and responds to emergencies. The alliance wants standard measures for response speed, duration and predictability, alongside operational data sharing.

It also proposes faster, risk-adjusted connection pathways for facilities making verifiable flexibility commitments. Connection costs would reflect their effects on the grid, including any avoided upgrades, rather than favouring a particular technology.

NVIDIA argues that this approach could make better use of existing capacity and shorten connection timelines. Those are intended benefits, not results established by the alliance’s launch. AEMA will work with utilities on connection arrangements and advocate for policies recognising responsive demand.

For data-centre developers and utilities, the proposal links connection terms to measurable operating commitments. A facility seeking faster access would need to demonstrate how it can reduce grid demand, not simply promise to be flexible.

Hang Ten Systems told TechCrunch’s Jagmeet Singh that it had closed the additional investment five weeks after its initial $32 million seed round. Xora, Temasek’s early-stage investment platform, led the new financing, with earlier lead investor Mayfield also participating.

Founded in May by former Infosys CEO Vishal Sikka, the Palo Alto business advises large companies on AI and helps build and modernise their software. It typically targets enterprises with more than $10 billion in annual revenue.

Smaller teams, with final checks elsewhere

Sikka says Hang Ten has secured multiple seven-figure contracts. Its customers include Fresenius Kabi, Saudi Aramco and Siemens Energy. The 21 enterprises he describes as its pipeline combine existing customers and late-stage prospects, rather than 21 signed customers.

Co-founder Sanjay Rajagopalan says engineers use an internal framework called Hobie, which packages reusable AI skills for regulated industries and complex software work. He claims some projects can use two to four people where about 30 might previously have been needed; customers or independent third parties still handle final quality checks and certification.

The company promises a tenfold improvement in cost, speed or a combination of both. That is its commercial claim, not an independently established result. The funding will support hiring, including planned expansion into Europe and India.

Large companies gain another funded supplier for software modernisation, with a model built around smaller delivery teams. Buyers still have a direct role in validating and certifying the systems, even where AI reduces the work of producing code.