Labels
Posted by

Best AI Stocks & ETFs for 2026: Powering the Next Boom

Last updated: September 13, 2026 — all prices, returns, and earnings figures refreshed to mid-September 2026 market data, following Nvidia's blockbuster fiscal Q2 report, Broadcom's post-earnings selloff, a second wave of Vertiv turbulence, and a global bond-yield surge heading into the Fed's September 15–16 meeting.

Quick Answer

The AI infrastructure trade has stopped moving as one basket and started splitting sharply by company. Nvidia (NVDA) delivered a blowout fiscal Q2 report on August 26 — revenue of $96.2 billion, up 106% year-over-year and its 14th straight quarter above its own guidance — and jumped nearly 9% the next session, yet the stock is still only up roughly 17% year-to-date near $218, calm by its own recent standards. AMD remains 2026's runaway winner, up roughly 140% year-to-date near $516. Broadcom told the opposite story: a record fiscal Q3 (AI semiconductor revenue up 221% to $16.7 billion, a bullish $230 billion AI-revenue target for fiscal 2028) was met with a selloff to five-month lows on margin-compression worries, leaving the stock up only about 5–6% for the year. Vertiv has kept sliding since its July revenue miss — down roughly a third from its June all-time high — after omitting backlog disclosures, drawing investor-rights law firm alerts, and announcing a $1.45 billion (up to $2.6 billion) acquisition of microgrid firm UtilityInnovation Group. Among ETFs, the VanEck Semiconductor ETF (SMH) still leads the group at roughly 58% year-to-date, though that's down from over 63% in mid-August. The bigger story this month is macro: a global bond selloff has pushed the 10-year Treasury yield near 4.8% and the 30-year close to 5%, the Fed's September 15–16 meeting now carries real odds of a rate hike rather than a cut, and Michael Burry's "circular financing" warnings got fresh fuel on September 11 when Reuters reported Nvidia is in talks to anchor Anthropic's potential IPO with an investment of up to $10 billion.

This is educational content, not financial advice. Every figure below moves daily — verify current prices before acting, and talk to a licensed financial advisor for guidance specific to your situation.

Table of Contents

The State of the AI Infrastructure Boom in September 2026

The AI capex story itself hasn't changed much in a month — Microsoft, Amazon, Alphabet, and Meta are still guiding to roughly $725 billion in combined 2026 capital expenditure, up about 77% from ~$410 billion in 2025, and none of the four has walked that back. Their Q3 earnings won't land until late October, so the next real capex update is still weeks away. What's changed instead is how the market is pricing the companies that build, power, and cool that infrastructure — and the macro backdrop those companies are trading against.

Nvidia's fiscal Q2 report on August 26 was, by most measures, the cleanest beat-and-raise of the year: revenue of $96.2 billion topped the Street's $92.1 billion estimate by 4.5%, Data Center revenue hit $89.0 billion (up 117% year-over-year and now 92% of total sales), and management guided Q3 revenue to $108 billion — above the roughly $103.8 billion consensus — while forecasting about 70% revenue growth for fiscal 2028, far above the ~44% growth some analysts had modeled. The stock still dipped slightly on report day before jumping 8.7% the next session. Yet even after that pop, NVDA is only up roughly 17% year-to-date near $218, a market cap of about $5.3 trillion — a genuinely quiet year by Nvidia's own standard, with AMD, Vertiv, and (until early September) Broadcom doing the heavy lifting on returns instead.

Broadcom's fiscal Q3 report on September 2 flipped that script within the semiconductor group itself. Revenue of $29.59 billion (up 86% year-over-year) and AI semiconductor revenue up 221% to $16.7 billion easily beat expectations, and CEO Hock Tan raised the company's long-term AI revenue target to $115 billion for fiscal 2027 and $230 billion for fiscal 2028 — driven by custom-chip demand from Google, Anthropic, and OpenAI. But a Q4 revenue guide of $34.8 billion that slightly missed the highest Street estimates, plus a gross-margin guide down to 73% from 78% a year earlier on rising memory costs, sent the stock tumbling to five-month lows near $353–358 before a partial rebound. Broadcom is now up only about 5–6% year-to-date, versus roughly 21% as of mid-August — one of the sharpest single-month reversals among the names in this article.

Vertiv's story has gone from bad to worse. Its July 29 Q2 miss (revenue of about $3.27 billion against roughly $3.38 billion in consensus) triggered a 17% intraday drop despite an EPS beat and raised full-year guidance; the stock kept sliding through August, falling roughly 28% over the quarter, in part because Vertiv omitted its usual backlog and order-intake disclosures from the release. In September, Vertiv agreed to acquire microgrid and power-architecture firm UtilityInnovation Group for $1.45 billion in cash (up to $2.6 billion including earnouts) to help data center customers secure power faster amid grid constraints — a strategically sensible deal that nonetheless triggered another ~6% drop on integration and valuation concerns. Investor-rights law firms Pomerantz and Hagens Berman have since announced investigations tied to the stock's decline and disclosure practices. VRT now trades around $248–257, down roughly a third from its June all-time high of $379.94, even though it remains up more than 50% for the year.

Layered on top of all of this is a genuinely new macro variable: a global bond selloff. The 10-year Treasury yield has pushed toward 4.8% (a level not seen since November 2023), the 30-year is approaching 5%, the UK sold a 30-year gilt at its highest yield since comparable records began in 1998, and Japan's 10-year yield moved above 3% for the first time in roughly three decades — all against a backdrop of tariff and energy-driven inflation concerns (Brent crude approached $100 a barrel in early September) and heavy AI-related corporate debt issuance. The Federal Reserve holds its next meeting on September 15–16, and after a hawkish Jackson Hole speech from Fed Chair Kevin Warsh and a hotter-than-expected August CPI print, markets are now pricing meaningful odds of a 25-basis-point hike — a real shift from the rate-cut expectations that prevailed for most of 2026. Higher yields cut two ways for this sector: they raise the discount rate applied to high-multiple growth stocks, and they raise the cost of the debt many hyperscalers and chipmakers are using to finance AI buildouts.

Top AI Infrastructure Stocks and ETFs for 2026

Our "One-Day Decision" framework still splits exposure between diversified ETFs for investors who want lower single-stock risk and individual stocks for those willing to accept sharper, name-specific swings — a distinction that matters more than ever given how differently Nvidia, Broadcom, and Vertiv have traded from each other over just the past month. Selections below use mid-September 2026 pricing and year-to-date performance.

Top AI ETFs for 2026

ETFs still provide broad, lower-risk exposure to the AI theme without a single-stock bet — and that diversification has mattered more in the past month than it did earlier in the year, as returns across even the AI-focused funds have started to diverge. Here's how the group stacks up as of mid-September 2026.

AI-themed ETF performance, mid-September 2026

ETF (Ticker) Focus 2026 Performance Notes
VanEck Semiconductor ETF (SMH) Chipmakers & chip-equipment Up roughly 58% year-to-date as of September 11 — still the group's standout, though down from over 63% in mid-August as Broadcom and Vertiv cooled sharply.
Invesco QQQ Trust (QQQ) Nasdaq-100 mega-caps Up roughly 17% year-to-date. Not a pure AI play, but heavy Nvidia/Microsoft/Amazon/Alphabet/Meta weighting keeps it the lowest-volatility way to hold the AI megacap backbone.
Global X AI & Technology ETF (AIQ) AI software, hardware, big data Up roughly 27–28% year-to-date — meaningfully ahead of QQQ now. The underlying index has shifted toward heavier semiconductor exposure (SK Hynix and Samsung are top holdings), so AIQ tracks more like a hardware fund than the "QQQ-with-a-tilt" it used to be.
ARK Autonomous Technology & Robotics ETF (ARKQ) Autonomous vehicles, robotics, AI-enabled mobility Actively managed; roughly 7% year-to-date, a sharp cooldown from ~22% in mid-August as its more speculative, mid-cap holdings lagged the broader chip rally.
Global X Robotics & AI ETF (BOTZ) Industrial robotics & automation Roughly 3% year-to-date, down from ~11% in mid-August. Holds names like ABB, Keyence, Fanuc, and Nvidia, with meaningful Japan exposure; the industrial-automation tilt has lagged pure semiconductor exposure all year.
iShares Future AI & Tech ETF (ARTY) Generative AI, AI infrastructure, software, services Up roughly 36% year-to-date, among the strongest in the group. This fund traded as IRBO until an August 2024 rebrand and now tracks the Morningstar Global Artificial Intelligence Select Index rather than a robotics-focused benchmark.

How to think about the ETF choice

  • Want pure semiconductor exposure: SMH still leads, though its concentration in a handful of names (including Broadcom and Nvidia) means it absorbed real damage from Broadcom's post-earnings drop this month.
  • Want the lowest-drama, still-AI-heavy option: QQQ. You're paying for mega-cap stability rather than thematic purity, and it has held up better than the pure-play semiconductor funds through the recent volatility.
  • Want hardware-adjacent AI exposure without picking single chip stocks: AIQ has re-rated well this year as its index shifted toward semiconductor names — check the current holdings, since this fund looks different than it did a year ago.
  • Comfortable with active management and higher volatility: ARKQ has cooled sharply over the past month; its more speculative mid-cap holdings haven't kept pace with the mega-cap chip rally.
  • Want the "old IRBO": it no longer exists in its original robotics-heavy form — ARTY replaced it in 2024 with a broader generative-AI benchmark, and that repositioning has paid off this year.

Top Individual AI Infrastructure Stocks for 2026

Individual stocks carry more single-company risk than ETFs, and September has been a lesson in just how differently that risk can play out even among companies riding the same AI infrastructure wave. Here's an updated look at the same seven names, with mid-September 2026 context.

1. Nvidia (NVDA) — AI Chips & Accelerators

Nvidia trades around $218 (market cap roughly $5.3 trillion), up only about 17% year-to-date — still a quiet year by Nvidia's own recent standards. Its fiscal Q2 report on August 26 beat across the board: revenue of $96.2 billion (up 106% year-over-year, 4.5% above the Street's $92.1 billion estimate), Data Center revenue of $89.0 billion (up 117% year-over-year, now 92% of total sales), non-GAAP EPS of $2.22 versus $2.09 expected, and a Q3 guide of $108 billion that topped consensus. Management also guided to roughly 70% revenue growth for fiscal 2028, well above the ~44% some analysts had modeled, and gross margin held at 75% (guided to slip slightly to 74% next quarter on rising component costs). Hyperscale revenue more than doubled year-over-year to $48.7 billion, while ACIE revenue (AI clouds, industrial, and enterprise customers) grew 138% to $40.3 billion — evidence that Nvidia's customer base is broadening well beyond the largest cloud providers. The stock initially dipped on report day before jumping 8.7% the next session. China remains a non-factor: H200 shipments under the current licensing regime stay minimal, and Nvidia continues to exclude China data-center compute revenue from its guidance entirely. The freshest headline: Reuters reported on September 11 that Nvidia is in talks to be an anchor investor in Anthropic's potential IPO — which could raise as much as $100 billion at a roughly $2 trillion valuation — with Nvidia considering an investment of up to $10 billion, a report still described as talks rather than a signed deal (more in the bubble section below).

2. Taiwan Semiconductor (TSM) — Chip Manufacturing

TSMC shares trade near $433 (market cap roughly $1.97 trillion), up roughly 50%+ year-to-date. July revenue hit a record NT$467.58 billion, up 44.7% year-over-year, and the company continues to guide to 2026 revenue growth "slightly above 40%" in US-dollar terms. TSMC's 2026 capital-expenditure guidance remains at $60–64 billion, with the large majority directed at advanced process technologies, alongside its existing $265 billion total US investment pledge (including the additional $100 billion Arizona commitment). Q3 earnings are due in mid-October and will be the next real test of whether 2nm/3nm demand has held up through the broader chip-sector volatility of the past month.

3. Broadcom (AVGO) — Custom AI Chips & Networking

Broadcom trades around $362 (market cap roughly $1.72 trillion), up only about 5–6% year-to-date — a dramatic deceleration from the ~21% gain in place as of mid-August. Fiscal Q3 2026 revenue (reported September 2) was $29.59 billion, up 86% year-over-year, with AI semiconductor revenue surging 221% to $16.7 billion; management raised its long-term AI revenue outlook to $115 billion for fiscal 2027 and $230 billion for fiscal 2028, citing custom-chip demand from Google, Anthropic, and OpenAI. Despite the beat, shares fell to five-month lows near $353–358 because the Q4 revenue guide of $34.8 billion slightly missed the highest Street estimates and gross margin is now guided down to 73% from 78% a year earlier on rising memory costs. The stock has since partially rebounded — Citic Securities upgraded to Buy with a $525 price target — and Wall Street's average target of roughly $506 still implies about 31% upside, though UBS trimmed its target to $470 from $485, questioning whether the $100 billion-plus fiscal 2027 AI revenue figure is achievable on the timeline management laid out.

4. Vertiv (VRT) — Data Center Cooling & Power

Vertiv trades around $248–257, down roughly a third from its June all-time high of $379.94, even though the stock remains up more than 50% for the year. The slide traces back to a July 29 Q2 miss — revenue of about $3.27 billion against roughly $3.38 billion in consensus, a shortfall investors reacted to sharply (down 17% intraday) despite an EPS beat ($1.52 versus $1.43) and raised full-year guidance. Vertiv also omitted its usual backlog and new-order disclosures from that release, which heightened concerns about order-momentum transparency, and at least six major brokerages (Citi, Goldman Sachs, RBC, Baird, KeyBanc, and Oppenheimer) cut price targets in the weeks that followed. In September, Vertiv announced a $1.45 billion acquisition of UtilityInnovation Group — a microgrid and behind-the-meter power specialist — with up to $1.15 billion more in earnouts (total consideration up to $2.6 billion, roughly 13x expected 2027 EBITDA), designed to help data center customers secure power faster amid grid constraints. The deal is expected to close in Q4 2026 and to be accretive to adjusted EPS in year one, but it triggered another roughly 6% drop on integration and valuation concerns, and investor-rights law firms Pomerantz and Hagens Berman have since announced investigations tied to the stock's disclosures. Not every analyst has turned bearish: GLJ Research holds a street-high $381 price target, citing Vertiv's $15 billion backlog and its 800-volt DC collaborations with Nvidia and Foxconn.

5. AMD (AMD) — AI Chips (Nvidia Alternative)

AMD remains the semiconductor story of 2026, trading near $516 (market cap roughly $843 billion) and up roughly 140% year-to-date — comfortably the best-performing large-cap name on this list. Q2 2026 revenue was $11.54 billion, up 50% year-over-year and above consensus, with Data Center revenue more than doubling to $6.7 billion (58% of total). Shares initially fell as much as 9% on the August 4 print because Q3 gross-margin guidance came in flat at 56% rather than expanding, before recovering within days; management guided Q3 revenue to roughly $13 billion (41% year-over-year growth). AMD hit fresh highs in early September alongside a broader chip-sector rally sparked by Micron's blowout quarter, and its AI-supply relationships keep expanding: the strategic partnership with Anthropic for up to 2 gigawatts of Instinct MI450 GPUs, an expanded Microsoft deployment, and the Taalas inference-chip acquisition all remain in place, alongside the existing multi-gigawatt deals with Meta and OpenAI. Worth flagging for long-term holders: both Meta and OpenAI hold performance-based warrants for up to 160 million AMD shares each (roughly 10% of shares outstanding), a modest future dilution consideration.

6. Arista Networks (ANET) — AI Networking

Arista shares trade in the high-$180s to high-$190s, near their record intraday high of $214.89 set after Q2 earnings. Q2 2026 revenue topped $3.06 billion, up 37.7% year-over-year and above the company's own $2.8 billion guide, with non-GAAP EPS of $1.02 beating the $0.89 estimate; management raised full-year 2026 revenue guidance to $12.6 billion (40% growth), up from an $11.5 billion guide given just one quarter earlier. Wall Street's consensus rating remains Strong Buy, with an average price target in the $225–241 range implying roughly 20–27% upside. Arista's open-ecosystem approach — supporting AMD accelerators, TPUs, and other chips rather than tying itself to one vendor — continues to differentiate it from more concentrated networking plays, and it remains a named Leader in Gartner's Magic Quadrant for enterprise networking.

7. Equinix (EQIX) — Data Center REIT

Equinix trades around $1,036–1,038, up roughly 35% year-to-date. Its Q2 2026 report (July 29) showed 16% revenue growth and 20% adjusted-FFO growth; management raised full-year guidance and set a new long-term target of 10–13% annual revenue growth through 2029. Equinix continues to lean on partnerships with Nvidia and Cisco on "AI Factory" deployments and has been signing new long-term power agreements — including a 20-year arrangement tied to a Hampton, Georgia project — as power availability, rather than chip supply, increasingly becomes the binding constraint on data-center buildouts (see Vertiv's power-focused acquisition above, and Texas's ongoing statewide data-center audit).

Bonus power plays: Investors looking for the energy side of the AI buildout continue to watch names like NextEra Energy (NEE) and Brookfield Renewable (BEPC), given the growing electricity demands of AI data centers — though these carry their own utility-sector dynamics, and power-grid bottlenecks are increasingly a headline risk in their own right, as Vertiv's own move into microgrid infrastructure this month underscores.

One-Day-Advisor Decision Framework

  • Lower risk: Broad ETFs like SMH or QQQ — accept sector-wide (not single-company) volatility in exchange for diversification. QQQ in particular has held up better than the pure-play chip funds through this month's Broadcom- and Vertiv-driven turbulence.
  • Moderate: Core positions in TSM and NVDA — established leaders with slowing-but-still-strong growth. Nvidia in particular has been the "calm" name in 2026 relative to its own history, and its next real catalyst is TSMC's mid-October earnings rather than anything company-specific in the near term.
  • Aggressive: AMD remains 2026's clearest winner, but Broadcom and Vertiv have shown this month just how quickly sentiment can turn on margin guidance or disclosure concerns even after genuinely strong underlying results — a reminder that "AI infrastructure exposure" no longer means uniform returns across the group.

The structural buildout still looks set to continue into 2027 based on current hyperscaler guidance, but the risk list has grown longer this month: a capex slowdown if AI monetization keeps disappointing relative to spending; further export-control shifts affecting China chip sales; the ongoing memory-chip shortage already pressuring margins at Broadcom and elsewhere; physical power and grid constraints on how fast new data centers can come online; and now, a genuinely new variable — rising bond yields that raise both the discount rate on growth-stock valuations and the cost of debt-financed AI buildouts, with the Fed's September 15–16 decision a live catalyst in either direction.

Is the AI Trade a Bubble?

"AI bubble" headlines have only intensified since midyear, and September's combination of a bond-market selloff and a fresh circular-financing story gave both sides of the debate new ammunition.

Bull case: Revenue growth at several names in this article has kept pace with — or exceeded — capex growth: Nvidia's Data Center revenue grew 117% year-over-year, Broadcom's AI semiconductor revenue rose 221%, and TSMC's July revenue was up 44.7% year-over-year. Nvidia's own ACIE (non-hyperscaler) revenue grew 138% year-over-year, which bulls point to as evidence that demand is broadening beyond a handful of concentrated buyers rather than depending on any single customer relationship.

Bear case: Investor Michael Burry has escalated his warnings sharply since midsummer, arguing in mid-August that a Bloomberg-mapped web of roughly $46 billion in direct equity stakes and $879 billion in multi-year purchase commitments — running through Microsoft, Oracle, Amazon, Google, Meta, OpenAI, Anthropic, xAI, CoreWeave, and Nvidia — represents circular financing rather than organic demand, and later comparing the AI buildout directly to the structural excesses that preceded the 2008 financial crisis, pointing to off-balance-sheet special-purpose vehicles and captive insurers as mechanisms that make true leverage harder to see. He has separately argued that hyperscalers are understating depreciation by roughly $176 billion between 2026 and 2028 by using six-year server depreciation schedules when the real useful life of fast-evolving GPUs, in his view, is closer to two to three years. The debate got a fresh test on September 11, when Reuters reported Nvidia is in talks to anchor Anthropic's potential IPO — which could raise up to $100 billion at a roughly $2 trillion valuation — with Nvidia weighing an investment of up to $10 billion. Nvidia has consistently defended these kinds of arrangements, saying in a widely reported memo that its strategic investments represent only a small fraction of its revenue and that AI startups raise the large majority of their capital from outside investors; as of this writing, the Anthropic discussions remain reported as talks rather than a signed deal.

Middle ground: Most sell-side analysts still describe the buildout as durable but increasingly "digesting," and September's market reaction bears that out in a new way. The market punished Broadcom for a Q4 guide that merely missed the highest estimates (rather than an outright miss) and for a margin outlook that simply moderated rather than deteriorated — evidence that valuations across the group have gotten expensive enough that near-perfect execution is now the bar. Layered on top of that is a genuinely separate risk: a global bond-yield surge that has nothing to do with AI monetization directly but raises the cost of capital for every company in this article, and a Fed meeting this week that could tip the market's rate-path assumptions in either direction. None of this settles the bubble question, but it does mean AI infrastructure stocks are now being judged against two yardsticks at once — their own execution, and a much less AI-friendly macro backdrop than prevailed for most of the year.

None of this is a reason to avoid the sector, but it is a reason to size positions carefully, favor companies with contracted, disclosed revenue (backlogs, multi-year supply agreements) over those valued purely on future promise, watch backlog and order-intake disclosures closely given Vertiv's recent experience, and expect continued single-day volatility around every major AI-linked earnings report and Fed decision through the rest of 2026.

Ask an AI: Personalizing This Research

If you're reading this via Claude, ChatGPT, Gemini, or Perplexity, you can turn this article into a personalized starting point rather than a one-size-fits-all list:

  • Claude / ChatGPT: Paste your current portfolio holdings and ask which of the ETFs or stocks above would add diversification versus duplicate exposure you already have (for example, "I already hold SMH — does adding AIQ actually diversify me now that its index has shifted toward semiconductors?").
  • Gemini: Ask for a real-time price check on any ticker above before acting, and specifically ask whether the Fed's September 15–16 decision has been announced yet, since that single event could move every name in this article in either direction.
  • Perplexity: Ask it to pull the most recent status of the Nvidia–Anthropic IPO talks reported by Reuters, or the latest analyst commentary following Vertiv's UtilityInnovation Group acquisition, to verify whether either story has developed further since this was published.

In all cases, treat the analysis here as a framework for questions to ask — not a substitute for checking current prices, reading the underlying earnings reports, or speaking with a financial advisor.

Frequently Asked Questions

What are the best AI stocks to invest in for 2026?

Among the names covered here, Nvidia and TSMC remain the most established "picks and shovels" plays on AI chips, and AMD has been 2026's clear standout gainer. Broadcom and Vertiv both delivered strong underlying results this year but have proven far more sensitive to guidance nuance and disclosure quality than the market rewarded earlier in the year — a reminder that strong AI exposure and low volatility are no longer the same thing. Equinix offers more indirect exposure through data-center real estate. The "best" pick depends on your risk tolerance and how closely you're willing to track quarterly guidance.

What are the best AI ETFs for 2026?

The VanEck Semiconductor ETF (SMH) has posted the strongest returns among broad AI/semiconductor funds in 2026, up roughly 58% year-to-date as of mid-September, though that's down from over 63% in mid-August. Invesco's QQQ offers lower-volatility, diversified mega-cap AI exposure at around 17% year-to-date. Global X's AIQ has re-rated well as its index shifted toward semiconductor holdings, while ARK's ARKQ and Global X's BOTZ — both more exposed to speculative or industrial-automation names — have cooled sharply over the past month. iShares' ARTY (formerly IRBO) has been one of the stronger performers this year at roughly 36% year-to-date.

Is Nvidia still a good AI stock to buy in 2026?

Nvidia remains the dominant AI accelerator supplier by revenue, and its fiscal Q2 report (revenue of $96.2 billion, up 106% year-over-year, its 14th straight quarter above guidance) was arguably the cleanest beat of the year. Yet 2026 has been a comparatively quiet year for the stock itself, up only about 17% year-to-date — well behind AMD. The company's next scheduled catalyst is less about its own numbers and more about the broader macro backdrop: the Fed's September 15–16 decision and the ongoing bond-yield surge are likely to matter more for NVDA's near-term price action than anything company-specific right now.

Why did Broadcom stock fall after a record quarter?

Broadcom's fiscal Q3 2026 results, reported September 2, actually beat estimates on revenue and EPS, and management raised its long-term AI revenue targets to $115 billion for fiscal 2027 and $230 billion for fiscal 2028. The stock fell anyway because the Q4 revenue guide of $34.8 billion slightly missed the highest Street estimates and because gross margin is now guided down to 73% from 78% a year earlier on rising memory costs — evidence that at Broadcom's valuation, the market is now demanding near-perfect guidance, not just strong current-quarter results.

What happened to Vertiv stock in 2026?

Vertiv shares have fallen roughly a third from their June all-time high after a chain of events: a July 29 Q2 revenue miss that triggered a 17% single-day drop despite an EPS beat and raised guidance; the omission of Vertiv's usual backlog and order-intake disclosures from that same release, which unsettled investors; a wave of analyst price-target cuts; and, in September, a $1.45 billion (up to $2.6 billion with earnouts) acquisition of microgrid firm UtilityInnovation Group that raised fresh integration and valuation questions even though management expects it to be earnings-accretive. Investor-rights law firms have since announced investigations tied to the stock's disclosures, though the underlying AI data-center power and cooling demand that drives Vertiv's business has not disappeared.

How much are hyperscalers spending on AI in 2026?

Microsoft, Amazon, Alphabet, and Meta are still collectively guiding to roughly $725 billion in 2026 capital expenditure — up about 77% from approximately $410 billion in 2025 — with the large majority going toward AI data centers, GPUs, custom silicon, and power infrastructure. This figure hasn't changed since the companies' Q2 earnings in early August; their Q3 results, due in late October, will be the next opportunity for a guidance update.

What is "circular financing" and why does it matter for AI stocks?

It refers to arrangements where an AI infrastructure company invests in a customer, and that customer then uses the money (or a purchase commitment) to buy the investor's own products. Investor Michael Burry has mapped roughly $46 billion in direct equity stakes and $879 billion in multi-year purchase commitments running between Microsoft, Oracle, Amazon, Google, Meta, OpenAI, Anthropic, xAI, CoreWeave, and Nvidia, and has argued the structure resembles the vendor-financing arrangements that preceded the 2008 financial crisis. The debate resurfaced on September 11, 2026, when Reuters reported Nvidia is in talks to invest up to $10 billion as an anchor investor in Anthropic's potential IPO. Nvidia's standing defense is that such investments represent a small fraction of its overall revenue and that AI startups raise most of their capital from independent, outside investors. It remains a live, unresolved debate rather than a settled question.

How are rising bond yields affecting AI stocks?

A global bond selloff has pushed the US 10-year Treasury yield toward 4.8% and the 30-year close to 5%, with similar moves in UK, Japanese, and German government bonds, driven by tariff- and energy-related inflation concerns and heavy government and corporate debt issuance. Higher yields create a two-sided problem for AI infrastructure stocks: they raise the discount rate applied to high-multiple growth valuations, and they raise the cost of the debt that hyperscalers and chipmakers are increasingly using to finance data-center buildouts. The Federal Reserve's September 15–16 meeting, where markets are now pricing meaningful odds of a rate hike rather than a cut, is a live near-term catalyst for the whole sector.

What happened to the IRBO ETF?

IRBO (iShares Robotics and Artificial Intelligence Multisector ETF) changed its name and ticker to ARTY (iShares Future AI & Tech ETF) in August 2024, and switched its underlying index from a robotics-focused benchmark to the Morningstar Global Artificial Intelligence Select Index. It's the same fund under a new name, ticker, and methodology, and that repositioning toward generative AI and infrastructure names has made it one of the stronger-performing AI ETFs in 2026, up roughly 36% year-to-date.

Are AI stocks in a bubble in 2026?

Opinions remain genuinely split, and September's news sharpened the debate rather than resolving it. Bulls point to revenue growth that has kept pace with or exceeded capex at several companies (Nvidia's Data Center revenue up 117%, Broadcom's AI semiconductor revenue up 221%) as evidence of real demand. Bears point to Michael Burry's escalating circular-financing warnings — now explicitly compared to 2008-era structural excesses — a fresh Nvidia-Anthropic IPO financing story, and a market that punished Broadcom and Vertiv sharply even after genuinely strong underlying quarters. Add a global bond-yield surge and an uncertain Fed decision into the mix, and most analysts describe the sector as growing but increasingly volatile and closely policed by investors, rather than clearly a bubble or clearly not one.

Disclaimer: This article is for general informational purposes only and is not financial or investment advice. Stock prices, ETF returns, and capital-expenditure figures referenced above reflect data available as of mid-September 2026 and can change significantly by the time you read this — verify current figures independently before making any investment decision, particularly given the Federal Reserve's September 15–16 meeting and the ongoing global bond-market volatility. Past performance does not guarantee future results. Consult a licensed financial advisor and conduct your own research, particularly given the elevated volatility across AI-related equities in 2026. One Day Advisor and its authors are not responsible for losses or damages arising from the use of this information.

References:

  1. https://www.reuters.com/legal/transactional/nvidia-talks-invest-anthropics-mega-ipo-sources-say-2026-09-11/
  2. https://finance.yahoo.com/technology/ai/articles/michael-burry-sounds-alarm-again-144154784.html
  3. https://finance.yahoo.com/technology/ai/articles/nvidia-may-put-10-billion-103618397.html

Related:

Labels:

Comments

Post a Comment

Popular Posts