Bloomberg Puts AI Build Costs Into an Energy Market Framework as Regulation Risk Mounts
Analysts compare AI infrastructure spending to energy capital cycles, as UK policy warnings and storage stock volatility reinforce the parallel.
Analysts on Bloomberg Intelligence drew an explicit comparison between AI infrastructure spending and the economics of building out energy markets, describing the AI build as expensive and structurally analogous to capital cycles that have periodically destabilised energy sectors. The parallel is not flattering to the AI thesis. NYMEX Henry Hub front-month gas traded at $2.73 per MMBtu on Monday (2026-07-27), and ICE Brent crude front-month held at $90.38 per barrel, both offering live benchmarks for what energy infrastructure economics look like at scale, and what large-scale power supply costs.7,6
Energy infrastructure investment carries specific failure modes: demand overestimation in early phases, returns concentrated among the best-positioned projects, and prolonged cost overruns during interconnection and permitting. OilPrice.com reported on June 14, 2026 (2026-06-14) that the AI data centre buildout is already showing those signs, with projects lacking strong sponsors, favourable grid locations, or utility partnerships facing delays and longer interconnection timelines.3
Fluence Energy (NASDAQ: FLNC) provided the sharpest equity illustration of the sector's volatility. Shares closed at $24.16 on May 8, 2026 (2026-05-08), up 98.2% in a single week after the company disclosed master supply agreements with two hyperscalers and revealed a record $5.6 billion order backlog. The speed of that move showed how starved the market was for evidence that AI power demand was converting into signed contracts.2
The surge had a short shelf life. FLNC shares remain down roughly 39% year to date, with market capitalisation at $3.21 billion and a price-to-sales ratio of 1.23 times trailing revenue. A violent re-rating on a single catalyst followed by a persistent discount is a pattern energy infrastructure equities have repeated across multiple cycles.2,1
Fluence's Q1 2026 figures showed adjusted EBITDA of $2.0 million, the fourth consecutive quarter in positive territory, with non-GAAP gross margin expanding to 52%. Operational stability at modest absolute scale, against a $5.6 billion execution obligation, is where the earnings story currently sits.2
The regulatory dimension was made explicit on July 9, 2026 (2026-07-09), when UK Conservative Party leader Kemi Badenoch told a Westminster audience that high energy prices and regulatory barriers would deter AI firms from choosing Britain, saying they would simply not "come here." The speech placed the UK's energy cost structure in direct competition with other jurisdictions for data centre investment. Power tariffs and permitting timelines have become primary site-selection variables for AI operators choosing between locations.5
Asia is running into related constraints. DLA Piper data published on June 30, 2026 (2026-06-30) showed regulatory uncertainty and execution risk are slowing battery energy storage investment across the region. China ranked as the world's third most attractive BESS market, cited by 14% of survey respondents, behind the US at 25% and the UK at 19%. For companies building global AI power supply chains, country-level regulatory exposure now acts as a measurable cost embedded in project returns.4
ICE Endex TTF front-month gas held at €63.76 per megawatt-hour on Monday (2026-07-27), unchanged. European power costs at that level keep the continent a difficult destination for large-scale AI data centre deployment. Badenoch's speech placed that dynamic in an explicitly competitive context, framing energy cost as a jurisdictional variable that shapes where AI capital ultimately lands.5
Energy capital cycles have historically bifurcated between best-in-class projects (those with grid access, credit-worthy counterparties, and favourable permits) and the rest. OilPrice.com's June 14, 2026 (2026-06-14) analysis flagged exactly that pattern emerging in AI buildout, with stronger sponsors advancing while weaker ones accumulate delays. Fluence's backlog, concentrated in hyperscaler commitments, sits toward the favoured end of that distribution. The wider storage and power supply universe does not have the same cushion.3,2
The conversion rate of Fluence's $5.6 billion backlog over the next two quarters is the most direct test of whether that positioning holds. On-schedule execution at Q1 margins would support the re-rating case. Any slippage on timeline or gross margin would make the energy infrastructure cautionary tale the more accurate analogy.2,3