Social Capital founder Chamath Palihapitiya estimates that 70 to 75 cents of every dollar spent on AI right now comes from AI companies that are unprofitable and funded by venture capital, since 1% of companies account for 90% of AI spending and 80% of that comes from unprofitable VC-backed firms. Much of that money goes to renting compute from Microsoft, Amazon and Alphabet. Steve Eisman says about 70% of the hyperscalers' AI revenue comes from Anthropic and OpenAI, and Citadel founder Ken Griffin has said he thinks AI is a bubble. Amazon's capital spending rose from $52.7 billion in 2023 to $132 billion in 2025, and Alphabet's from $32.3 billion to $91.5 billion. The author argues that the bubble risk sits with the AI labs, while Microsoft, Amazon and Alphabet are already public and profitable, and says next week's earnings reports will test whether cloud growth holds up.
Summary written by Upside AI from the original article. Read the source for full details.
Read the original on Yahoo Finance ↗
Keywords
artificial intelligenceAI investmentventure capitalcloud computingNVDA
Open a keyword to see other news on the same topic.
Related tickers and likely impact
NVDA
▲ Likely positiveNvidia
Relevance: low
| Market | US stocks |
|---|---|
| Sector | Information Technology |
| Industry | Semiconductors |
| Market cap | $5.2T |
| Close on 10/23/2026 | 211.00 |
| Change that session | -0.57% |
MSFT
• NeutralMicrosoft
Relevance: low
| Market | US stocks |
|---|---|
| Sector | Information Technology |
| Industry | Systems Software |
| Market cap | $4.0T |
| Close on 10/23/2026 | 524.53 |
| Change that session | +0.66% |
Impact is scored by AI from the headline and may be wrong. It is not investment advice.
Related news
- Neoclouds secure record funds as Nvidia’s Huang says more investment required
Venture investors put $6.6 billion into 58 AI neocloud deals so far this year, according to PitchBook data, more money and deals than any prior year and 51% above last year’s total of $4.4 billion. Neoclouds are companies that sell remote access to GPUs for AI workloads. In an interview with CNBC, Nvidia CEO Jensen Huang said demand for neoclouds is “off the chart” and that more investment is still needed.
- Amazon cloud is nearly sold out through 2027. Can its AI spending keep up with demand?
Amazon Web Services (AWS) has sold out most of its cloud capacity for 2027 and a substantial portion of 2028, according to CEO Andy Jassy speaking on a podcast. Despite planned capital expenditure of about $220 billion in 2026, Amazon is struggling to meet demand from AI labs and enterprise customers, and much of its 2026 capacity is already booked. Jassy described AI demand as a “barbell”: AI labs are consuming computing resources at an unprecedented scale, while the middle of the market, where everyday business software has yet to widely adopt AI, remains largely untapped. He said inference is still about an order of magnitude too expensive to be used pervasively, suggesting costs need to fall roughly tenfold. Amazon's Trainium chips are positioned as a lower-cost option, offering 30% to 40% better price-performance than comparable processors, and Amazon has disclosed more than $225 billion in future Trainium revenue commitments, including deals involving Anthropic and OpenAI.
- Nvidia CEO Jensen Huang says neocloud demand is 'off the charts' but capital is scarce
Nvidia CEO Jensen Huang told CNBC on Thursday, on the sidelines of the company's GTC conference in Berlin, that demand for neocloud firms is “off the charts” and their capabilities are “extraordinary.” He said the bottleneck is the working capital needed to fuel growth, as AI demand across many sectors has created a scarcity of capital, with demand for capital far exceeding its availability. He added that this gap will be solved.
- NVIDIA Is Finally “Too Big to Fail.” Here’s Why Meta and Google Aren’t.
Semafor's Liz Hoffman argues that NVIDIA is so embedded in the AI ecosystem that the U.S. government would be more likely to rescue it in a serious financial crisis than Meta or Google, though she acknowledges this is a theoretical exercise and no rescue is guaranteed. NVIDIA generated nearly $97 billion in free cash flow in the fiscal year ended January 2026, funding both strategic investments in companies that depend on its infrastructure and shareholder returns. By contrast, Meta and Alphabet spent roughly 60% and 55% of their operating cash flow on capital expenditures in 2025, mainly for their own infrastructure. The article also notes NVIDIA accounts for roughly 8% of the S&P 500, so a significant deterioration in its financial position could affect a broad group of index investors.
- This Nvidia-Backed Cloud Computing Firm Is Reportedly In Talks To Sign A $45B Compute Deal With Anthropic
According to Bloomberg, Anthropic is in advanced talks to acquire about $45 billion of computing power from Nvidia-backed Nscale, centered on a second building at Nscale's Monarch campus in West Virginia that is slated to deliver 460 megawatts using Nvidia's Vera Rubin chips. Final terms are not set, and an agreement could come within days; both companies declined to comment. The site was first intended for Microsoft, which withdrew after signing a preliminary letter of intent. Anthropic already has infrastructure commitments with Nscale worth about $44.6 billion, according to filings tied to Nscale's upcoming public offering. Nscale closed a $3.36 billion pre-IPO round led by Third Point and is targeting a $30 billion valuation for its debut.
- The Trillion-Dollar AI Bet: What It Means for Hyperscaler Investors
The five major hyperscalers, Alphabet, Microsoft, Amazon, Meta and Oracle, are expected to spend almost $780 billion on capex in 2026, up 87% year over year, and to exceed $1 trillion annually from 2027 through 2030. Financial risk is rising: Alphabet's long-term debt reached $98.2 billion, up 568% from four years ago, and Meta's was $83.7 billion; Alphabet posted negative free cash flow of $5.9 billion in Q2, Meta's free cash flow fell 91% year over year, and both have paused share buybacks. The article stresses that no one knows what returns will follow, and past spending booms such as railroads and telecom led to overbuilt capacity even though the infrastructure later enabled new businesses. Bain & Company estimates AI infrastructure spending could reach $1.5 trillion annually by 2031, and that the AI economy would need $6 trillion in yearly revenue to justify these outlays, with about two-thirds expected to come from innovation that does not yet exist.
Related markets
Latest news
- Bloom Energy Raised Its 2026 Forecast for the Third Time. Is the Stock a Buy More Than a Third Below Its Record?
Bloom Energy reported record third-quarter revenue of $1.24 billion, up 138% year over year, and raised its 2026 revenue forecast to $4.4 billion–$4.6 billion for the third time this year. Despite the results, shares fell about 9% to $217.77 on Oct. 22 and dropped to roughly $203 after hours, still about 37% below the June 22 record close of $345.85. The weak spot was cash: operations used $27 million in the third quarter, contract assets (revenue booked but not yet billed) rose to $793 million, and free cash flow was negative $181 million. The author says the stock now looks cheaper relative to the raised earnings forecast and could be considered for gradual investment, but wants to see contract assets convert to cash before increasing the position more aggressively.
- Apple Joins Nvidia in the $5 Trillion Club. Here's Which Stock I Think Can Get to $10 Trillion First.
Apple has joined the $5 trillion market-cap club after recent stock gains, while Nvidia remains slightly larger at $5.13 trillion. The author expects Apple to reach $10 trillion first, arguing it is a lifestyle brand with strong customer loyalty: iPhone owners account for about half of Apple's sales, 87% of current iPhone owners say their next smartphone will also be an iPhone, and Apple-made devices in use reached a record 2.5 billion. Nvidia trades at about 14 times next year's expected profit of $15.91 per share, which may reflect a headwind in its AI business, since data centers account for roughly 90% of its revenue and competitors such as AMD and Qualcomm are rising. The article notes that either company could reach the milestone first and that both can perform well.
- Inflation Is Running Hotter Than Expected. That Could Be Bullish for Bitcoin.
Consumer prices rose 3.7% in the 12 months through September 2026, and Bitcoin fell about $1,000 after the hotter-than-expected October 14 report, as higher inflation raises the odds of Federal Reserve rate hikes. The article argues Bitcoin is not necessarily well suited to hedge rising costs: producer prices jumped 6.3% over 12 months, and squeezed consumers may invest less in riskier assets like Bitcoin. However, J.P. Morgan Asset Management views Bitcoin as a bet that governments will keep creating money, and U.S. M2 money supply grew 5.7% in the year through August 2026. If the Fed cuts rates while prices keep climbing, Bitcoin's fixed supply could serve as an inflation refuge, and the author says the next few quarters will test this thesis.
- AI bear slams Nvidia's GPU pitch: 'There's not scarcity other than that which is created by Jensen Huang'
Ed Zitron, CEO of PR firm EZPR and a critic of the AI spending boom, says Nvidia (NVDA) is creating the scarcity it sells against. He pointed to a concentrated customer base, with five companies accounting for about 70% of Nvidia's accounts receivable in the first half, and accused Jensen Huang of channel stuffing by pushing GPUs into the channel. He questioned what more data centers deliver, citing a Bloomberg report that Microsoft (MSFT) has about 2 gigawatts of AI compute online out of 12 gigawatts reported. He said OpenAI and Anthropic have become 'load-bearing' for several large companies and, asked what should happen if they falter, replied 'Let them die.' He argued that a rescue would not hold up the market, since it depends on GPU sales continuing to rise at higher prices.
- Vertiv (VRT) Stock Looks Below Fair Value After A Huge Run
The article says Vertiv (VRT) shares have risen strongly on the data center and AI infrastructure wave, with a five-year return of about 660%. A discounted cash flow (DCF) model estimates latest twelve-month free cash flow at about $3.0 billion and an intrinsic value meaningfully above the $194.32 share price. New AI factory partnerships with NVIDIA and Dassault Systèmes, plus a sizeable order backlog tied to AI data center projects, are cited as factors that could affect how fast cash arrives and how much investment is needed. The piece also notes a P/E of 43.2x and asks whether the current price is fully supported by the cash flows implied in the DCF estimate.
- Ex-dividend calendar for Oct 26–30: One bank's 50% rights issue ratio stands out
This week, 10 companies will pay cash dividends, ranging from nearly 3.5% to 20%. MSR set Nov 2 as the last registration date for its 20% advance dividend of the 2/2026 round, with an ex-date of Oct 30 and an estimated payout of about 2,210 billion VND on Nov 6. MSR's shareholders approved an advance of 50% for this round, about 5,500 billion VND, which together with the 10% advance from round 1/2026 brings its total 2026 dividend to 60%, roughly 6,600 billion VND. Techcombank (TCB) set Oct 30, 2026 as its last registration date, with an expected ex-rights date of Oct 29, to issue 3,543,120,207 shares at a 50% ratio, so holders of 100 shares receive 50 new ones. The issue, worth over 35,431 billion VND at par value, is funded by undistributed retained earnings from the audited 2025 financial statements.
Get the briefing on Telegram or Zalo
Join the Upside Telegram channel or Zalo group for the Vietnam stocks and gold briefing at 8:00 and 19:00 (Vietnam time) on trading days, plus a weekly wrap on Saturday morning.