NextFin News - Broadcom raised its artificial-intelligence chip revenue forecast to about $115 billion for fiscal 2027, up from a prior target of more than $100 billion, and said it can see that figure doubling to roughly $230 billion by fiscal 2028. On the same day, Qatar paused plans to revive output at the world's largest liquefied natural gas export complex after renewed fighting in the Strait of Hormuz, while the United Arab Emirates moved ahead with an integrated LNG trading platform — a market-share contest that is redrawing the map of global gas supply.
The two developments sit in different industries but answer the same question: how durable is the infrastructure buildout of the late 2020s? Broadcom's answer is that AI demand is compounding faster than consensus expected. The gas market's answer is that the energy infrastructure powering that buildout is fracturing along geopolitical fault lines, and that fracture may be permanent.
The Numbers Behind Broadcom's AI Surge
Broadcom reported fiscal third-quarter results after the close on September 2 that underscored the speed of the AI spending cycle. AI semiconductor revenue reached $16.7 billion, up 221% from a year earlier, lifting total revenue to $29.59 billion — ahead of the $29.36 billion analysts expected. Adjusted earnings came in at $3.32 a share, versus $3.24 expected.
The market's reaction, however, was not celebratory. Fourth-quarter revenue guidance of about $34.8 billion fell just short of the $35.03 billion average estimate compiled by LSEG, and the stock had already surrendered roughly a quarter of its value from its June peak before the report. The options market had priced an implied move of about ±9% into the September 2 print, and the shares have been trading with the volatility of a company whose investors are positioned for perfection.
"Line of sight to achieve AI revenue from chips in excess of $100 billion in 2027," chief executive Hock Tan told investors, before the company went further and lifted the fiscal 2027 AI chip revenue forecast to about $115 billion, with a path to approximately $230 billion in fiscal 2028.
The detail that matters most is not the headline growth rate but the visibility behind it. Tan said Broadcom can see more than 10 gigawatts of AI infrastructure deployments for Anthropic, over 5 gigawatts for OpenAI, and 3 gigawatts for Meta. "That is committed capacity, not aspiration, and it closes most of the gap to what the market wanted," said Patrick Moorhead, chief executive of analyst firm Moor Insights & Strategy.
That distinction — committed capacity versus aspiration — is the fulcrum of the entire AI chip thesis. A gigawatt-scale data center does not order accelerators quarter by quarter; it locks in chip supply years ahead, because the power, cooling, and building timelines stretch well beyond a single procurement cycle. When a chip supplier can point to named hyperscalers and named gigawatt figures, the revenue is no longer a bet on next quarter's capital expenditure budget. It is a bet on infrastructure that is already under construction.
Why the Gas Fight Is Structural, Not Cyclical
The disruption in the LNG market is often framed as a cyclical supply shock: fighting in the Gulf, tankers attacked, prices spike, then everything reverts once the shooting stops. That framing is wrong. This is a structural reordering of global gas trade, and the evidence is geographic.
More than 110 billion cubic meters of LNG passed through the Strait of Hormuz in 2025. About 93% of Qatar's LNG exports and 96% of the UAE's transited the strait — together representing almost one-fifth of global LNG trade. Unlike crude oil, which can be piped around chokepoints, LNG has no "Hormuz bypass": the gas must be liquefied at the source and shipped through the strait. When Qatar pauses its ramp-up, those volumes do not shift to a different route. They leave the market, and buyers rewire their supply chains.
The International Energy Agency estimates the disruption will cause a cumulative loss of roughly 140 billion cubic meters of LNG supply between 2026 and 2030. That is not a quarterly blip; it is a five-year supply hole. And into that hole are stepping the exporters with what amounts to geopolitical insurance. The United States, Canada, Nigeria, and Malaysia have all lifted output, with North America accounting for more than 85% of the increase in global LNG supply expected in 2026 — the fastest growth pace since 2019. Two new projects, Commonwealth LNG in May and Delphin FLNG 1 in June, moved to final investment decisions precisely because buyers are paying a premium for supply that does not transit a war zone.
This is where the Qatar-versus-UAE dynamic becomes a market-share fight rather than a temporary outage. Qatar's historical advantage has been cost: it produces LNG more cheaply than almost any competitor, and its North Field expansion was designed to lock in decades of dominance. But low cost does not matter if the cargo cannot leave the terminal. ADNOC's newly announced integrated LNG platform — combining the marketing activities of ADNOC Gas and XRG with ADNOC's trading capabilities through the Abu Dhabi Global Market — is a bid to convert Qatar's vulnerability into opportunity. It offers buyers a Gulf supplier whose cargoes can be marketed and traded with the flexibility of a paper portfolio, backed by a shipping fleet that is itself expanding: ADNOC's shipping arm recently ordered four more LNG carriers, bringing its newbuild program to 18 vessels.
A report published in August by the Center for Strategic and International Studies put the stakes plainly: the longer the disruption lasts, the more other suppliers will capture Qatari market share, both through existing trade flows and through new investment decisions in more stable jurisdictions. Market share lost in a five-year supply shock is not recovered when the shooting stops; it is written into long-term contracts that outlast the crisis.
The Counter-Thesis: Demand May Not Cooperate
Both bullish narratives — the AI chip supercycle and the structural gas deficit — rest on one shared assumption: that demand will keep rising fast enough to absorb the premium being built into prices. The strongest case against that assumption comes from the demand side of the gas equation, and it deserves weight.
Despite the scale of the Gulf shock, prices have not risen to the levels seen in 2022. The explanation, according to energy research published by KAPSARC, is that Asian buyers have switched back to coal and curtailed industrial gas use. European storage has also drawn down as LNG imports declined. If coal substitution holds — and if the regasification terminals and gas turbines that Southeast Asian importers need fail to arrive on time — then LNG demand growth will undershoot the buoyant forecasts underpinning the structural-deficit narrative. In that world, today's price premium is a temporary risk charge that unwinds once supply routes normalize, and the "structural" call was really a cyclical spike in disguise.
The same logic applies, with a twist, to AI chips. The bear case voiced by strategists who have warned of an AI capital-expenditure bubble is that hyperscalers are overbuilding: they are ordering chips against revenue models that assume AI services will monetize faster than they may. If Anthropic, OpenAI, and Meta slow their buildouts even modestly in 2027, Broadcom's $230 billion fiscal 2028 figure — which rests on capacity commitments made today — would be the first number to come under pressure. Broadcom's own guidance acknowledges the risk implicitly: it guided fourth-quarter revenue below the consensus estimate even as it raised the multi-year AI outlook, a signal that near-term execution is not frictionless.
The falsifying signals are specific. For the gas thesis: if Chinese and Indian LNG imports do not recover to their pre-disruption trend within four quarters of the Qatar ramp-up resuming, the structural-deficit call is wrong. For the AI chip thesis: if any of the three named hyperscalers cuts its disclosed AI infrastructure capital expenditure guidance by more than 15% in the next two quarters, the structural-growth thesis for custom AI chips weakens materially. A judgment without a falsifying condition is a wish; both of these can be checked against public data.
The Link Between Silicon and Molecules
The second-order connection between Broadcom's forecast and the LNG fight is power. A gigawatt of AI data-center capacity draws roughly the electricity of a mid-sized city, and in the United States — the engine of both the AI buildout and the LNG supply response — a large share of that power still comes from natural gas. The same North American gas boom that is filling the Gulf supply gap is also what makes U.S. data centers cheaper to run than their European or Asian counterparts.
This creates a feedback loop with distributional consequences. Regions with secure, cheap gas — North America, and to a lesser extent exporters with non-Hormuz routes — gain a dual advantage: they can export LNG at a security premium and host AI infrastructure at a power-cost discount. Regions dependent on Hormuz-transiting cargoes — much of Asia and Europe — face the inverse: higher, more volatile energy costs that erode the economics of both industrial gas use and power-intensive computing.
The mechanism is already visible in the price data. The European benchmark Dutch TTF natural gas price for October 2026 delivery reached an intraday high of €70.85 per megawatt-hour early this week, according to ICE data, and was up more than 20% over the past month and more than 120% from a year earlier. In March, European gas futures surged as much as 48% in a single session before paring gains to close 22% higher — the highest levels since 2023. Meanwhile, the correlation between European and Asian spot LNG benchmarks rose to a record 0.955 in 2025, meaning a shock in the Gulf now transmits to both markets almost simultaneously. A buyer in Tokyo and a buyer in Rotterdam are, increasingly, bidding for the same destination-flexible cargoes — and when Qatar sits out, they bid against each other.
What to Watch: Scenarios Across Time Horizons
The base case is that both trends persist in parallel: AI chip demand compounds at the rate Broadcom describes, supported by committed hyperscaler capacity, while the LNG market remains structurally tighter than the pre-2026 consensus of a looming glut. Before the conflict, the industry was debating whether a wave of more than 320 billion cubic meters of LNG export capacity that had reached final investment decisions would trigger oversupply and push Asian and European spot prices from $11 per million British thermal units down to $6. That debate has been suspended, not settled.
The upside case for Broadcom is that the $230 billion fiscal 2028 figure proves conservative if AI inference demand exceeds current deployment plans. The downside case is a hyperscaler capital expenditure pause that leaves custom-chip orders unfulfilled and turns today's backlog into tomorrow's write-down. For gas, the upside case is a prolonged Hormuz disruption that pushes North American exporters into a multi-year supercycle; the downside case is a settlement that restores Qatari volumes faster than expected, flooding a market that has priced in scarcity.
Split by horizon, the picture diverges. In the short term, both stories are driven by sentiment and positioning — Broadcom's stock by options-implied volatility around earnings, gas prices by headlines from the Gulf. Over the medium term, fundamentals dominate: chip revenue conversion and LNG cargo cancellation notices are the metrics that matter. Over the long term, the structural calls decide — whether custom AI chips become a durable layer of computing infrastructure, and whether the Hormuz chokepoint permanently reroutes global gas trade toward the Americas.
The specific signals to track: Broadcom's next quarterly disclosure of hyperscaler committed capacity and any revision to the $115 billion fiscal 2027 AI revenue forecast; LNG force majeure and cargo cancellation notices from QatarEnergy; the pace of North American final investment decisions outside the Gulf; and Chinese and Indian LNG import volumes relative to their pre-disruption trend. If Broadcom's guidance holds and Qatar's pause extends into late 2026, the twin infrastructure themes — silicon and molecules — will continue to move together.
Broadcom is betting the AI boom will compound for years. The gas market is betting the Gulf will stay fractured. Both bets rest on the same assumption — that the infrastructure buildout of the 2020s cannot be undone by a single chokepoint. One of them will be wrong.
Explore more exclusive insights at nextfin.ai.
