Should hyperscalers do more about AI emissions?

Should hyperscalers do more about AI emissions?

Legal & General [LGGNY] backed shareholder resolutions urging Alphabet [GOOGL], Amazon [AMZN] and Meta [META] to explain how their artificial intelligence (AI) expansion can align with emissions targets, Bloomberg reported. Although all three proposals failed, L&G singled out Meta, saying the social media giant “could improve its disclosures on how it balances the need for ‘speed to power’ against the need for clean and reliable power,” as well as how it sees “the role of gas as grid-connected power and on-site generation within its decarbonisation pathway.”

Data centres guzzle gas 

Gas-fired power capacity being developed specifically for US data centres has nearly doubled to more than 189GW since the start of 2026, according to Global Energy Monitor, as tech companies turn to private, behind-the-meter gas. “Increasingly, the US gas power buildout is getting tied directly to the data center buildout – you can’t talk about one without the other,” Jenny Martos, a research analyst at Global Energy Monitor, told Wired.

Is the solution to build them in space?

SpaceX [SPCX] plans to launch its first AI satellites in Q4 2027, using a space-optimised Vera Rubin NVL72 system from Nvidia [NVDA], with the orbital computing network reaching “significant scale” in 2028, according to a post from Elon Musk on X. The project highlights how fast the firm is aiming to make space-based data centres a reality. Musk has said they will be cheaper and easier to build than terrestrial ones. 

Nvidia’s powered land portfolio

In another sign that energy and AI are becoming increasingly symbiotic, the behemoth chipmaker is extending its energy investment spree beyond nuclear and grid software into “powered land”, taking a minority stake in Cloverleaf Infrastructure to speed up the build-out of AI data centres. Nvidia’s growing web of energy investments, from SB Energy to Crusoe, represents a move to clear a key bottleneck in the AI race: land to build data centres on, and the power to run them.

Can this memory player diversify revenue towards AI?

Silicon Motion’s [SIMO] Q2 2026 revenue surged 127% y/y to $451m, driven by explosive growth in its Ferri product line, which now comprises 30% of revenue, up from just 4% a year earlier. The company is pushing into AI infrastructure with its new MonTitan SSD reference design. SIMO shares have surged over 180% year-to-date on the AI story; Aureon unpacks how, while the stock is the cheapest of its peer group on forward earnings, its bull case depends on the memory supercycle persisting.

Cybertrucks: now even pricier

On Tuesday, Tesla [TSLA] raised the price of two variants of its Cybertruck pickup by $5,000; according to the company website, its Dual Motor version now costs nearly $75,000, whereas the Premium version costs nearly $85,000. The company has not provided a reason for the price increase for the vehicle, which has struggled to find buyers in the US’ competitive truck market. 

Three cancer research stocks to watch after Moderna’s breakthrough

Moderna’s [MRNA] melanoma vaccine breakthrough sent its shares soaring and lifted sequencing and mRNA peers in its wake. Aureon examines three ways to play the personalised cancer treatment boom: Moderna’s direct mRNA exposure, Illumina’s [ILMN] ‘picks and shovels’ sequencing play and BioNTech’s [BNTX] diversified oncology pipeline.