Priced for Tomorrow
The AI laboratories are valued at close to a trillion dollars each, on revenues of a few tens of billions. What they earn, what they spend, what they have promised, and what the analysts on both sides say it adds up to.
On 28 May 2026 Anthropic announced that it had raised $65 billion “valuing the company at $965 billion post-money”. Fifteen months earlier, in March 2025, it had raised $3.5 billion at $61.5 billion.[1] OpenAI had closed its own round two months before, “with $122 billion in committed capital at a post money valuation of $852 billion”, and by October was reported to be in talks at about $1.4 trillion; Anthropic, according to press reports of a draft prospectus, hopes to list its shares after the American midterm elections at about $2 trillion.[2][3] Those are the prices of two companies that, in their most recent reported quarters, recognised revenue of $6.7 billion and $16.8 billion.
Behind them stand the companies that sell them computing and chips. Nvidia, Microsoft, Alphabet, Amazon, Meta and Oracle are worth more than $18 trillion together on the stock market and plan to spend more than $800 billion this year, mostly on data centres. Whether those prices make sense is the most argued question in finance. The bank governors who watch financial stability have warned about it; the chief executives of the companies involved have used the word “bubble” themselves; and a smaller but serious group of analysts argues that the numbers, unlike those of 2000, are backed by real earnings.
This essay sets out what is known: what the companies are valued at, what they actually earn and lose, what they have promised to spend, and what each side of the argument says. It takes no side. One disclosure belongs at the start: it was written with a tool made by Anthropic, one of the companies discussed, which is treated like any other.
From $29 billion to $852 billion
A valuation in a private round is the price at which a few investors buy a slice of a company, multiplied up to the whole. It is not what the company would fetch if sold, and the money behind it is not always cash. When OpenAI closed its round on 31 March it described $122 billion of “committed capital”; according to Bloomberg Law, $35 billion of Amazon’s $50 billion depends on OpenAI going public or reaching a technical milestone, and other investors pay in instalments.[2] With that caution, the prices have risen faster than for any group of companies in memory (Figure 1). OpenAI was valued at about $29 billion when Microsoft invested in January 2023, at $157 billion in October 2024, $300 billion in March 2025, $500 billion in an employee share sale that October and $852 billion in March 2026.[4] Anthropic went from $61.5 billion to $183 billion in September 2025, $380 billion in February 2026 and $965 billion in May, the first time it was priced above OpenAI.[1] Elon Musk’s xAI was folded into SpaceX in February 2026 at a reported $250 billion, and SpaceX listed its shares in June.[5]
The smaller laboratories are priced on the same scale. Mistral, in France, raised €3 billion in September 2026 at more than €21 billion; Safe Superintelligence, which sells nothing yet, was valued at $32 billion; Thinking Machines Lab was valued at $12 billion within months of its founding and has been in talks at $40 billion or more; Cursor, a tool for programmers, was bought by SpaceX in August at an implied $60 billion.[6]
What they actually earn
The laboratories mostly talk about revenue in one way and account for it in another. The figure in announcements is the “run-rate”: the latest month’s revenue, or sometimes the latest week’s, multiplied to a year. Anthropic’s rose, by its own statements, from “approximately $1 billion” at the start of 2025 to over $5 billion in August, $14 billion in February 2026 and $47 billion in May.[1] OpenAI said in March 2026 that it was “now generating $2B in revenue per month”.[2] Run-rates are not audited, flatter a fast-growing business, and are not defined the same way by everyone: in October the Financial Times reported that OpenAI’s annualised revenue was near $50 billion, not the $70 billion it had signalled, a gap attributed to counting revenue shared with cloud partners gross or net in order to match Anthropic’s figure.[7]
The figures that would appear in accounts, revenue recognised in a period, are smaller, and for the private companies they come from leaks. OpenAI’s revenue was about $3.7 billion in 2024 and $13.07 billion in 2025, according to audited documents reported by Ed Zitron and confirmed by the Financial Times; the Wall Street Journal reported $5.7 billion in the first quarter of 2026 and $6.7 billion in the second. Anthropic recognised $4.6 billion in 2025, according to the draft prospectus reported by Reuters, then about $4.7 billion, $11.5 billion and $16.8 billion in the first three quarters of 2026; in the second quarter it overtook OpenAI for the first time (Figure 2).[3][8] SpaceX’s artificial-intelligence unit, the former xAI, is the only one with audited public numbers: $3.2 billion in 2025, most of it advertising and subscriptions on X, and $7.6 billion in the third quarter of 2026, driven, it says, by “new cloud services agreements, the acquisition of Cursor, and increased Grok adoption”; one of those agreements has Anthropic renting its chips at $1.25 billion a month.[5]
The growth is real by any standard. Anthropic’s third-quarter revenue was more than three times its revenue for the whole of 2025. ChatGPT had more than 900 million weekly users and more than 50 million subscribers in March, which means that about one user in eighteen pays; OpenAI says businesses now make up more than 40 per cent of its revenue.[2] Two features make the revenue less solid than the growth suggests. It is concentrated: about a fifth of Anthropic’s revenue in the first half of 2026 came from two customers, according to the Wall Street Journal. And part of it is circular: the laboratories sell to companies that are also their investors and their suppliers, a point the section on the builders returns to.[8]
Losses, and what kind
OpenAI’s operating loss in 2025, the gap between revenue and the cost of running the business, was about $20.9 billion, more than one and a half times its revenue; in the first half of 2026 it was about $21.6 billion, on revenue of $12.4 billion.[8] Anthropic’s operating loss in 2025 was more than $8 billion on $4.6 billion of revenue; this year it has told investors of operating profits in the first two quarters and positive free cash flow in the third, on adjusted measures that are not yet in any public filing.[3] The headline net losses are much larger, $38.5 billion for OpenAI in 2025 and $80.7 billion for Anthropic in the first half of 2026, but most of that is an accounting effect: the convertible instruments the companies issued to investors become more valuable as the valuation rises, and the increase is booked as a loss. A rising share price makes the reported loss bigger.[3][8]
The larger numbers are the promises. In November 2025 OpenAI’s chief executive, Sam Altman, put its commitments for computing at about $1.4 trillion over eight years; contracts that can be traced add up to something less, including an “incremental $250 billion of Azure services” from Microsoft, about $300 billion from Oracle over five years and $22 billion from CoreWeave.[9] Anthropic’s draft prospectus reportedly lists about $450 billion of cloud agreements over the next decade.[3] Some of these can be cancelled: SpaceX’s prospectus says its $1.25 billion-a-month agreement with Anthropic “may be terminated by either party” after the first three months.[5] Set against recognised revenue of tens of billions, the commitments are what makes the question serious: they are paid out of revenue that does not yet exist.
Thirty times sales
The usual way to compare prices is to divide the value of a company by its revenue or its profit. The laboratories have no profits to divide by, so the comparison is with revenue (Figure 3). At its last priced round, OpenAI was valued at about 32 times its second-quarter revenue multiplied by four; Anthropic, at 14 times its third quarter, or 21 times the run-rate it announced with the round; xAI, at its merger, at 78 times a year of revenue that was mostly X’s advertising. For listed companies, Nvidia trades at about 17 times its sales of the past year, Microsoft at 12, Alphabet at 9, Amazon at under 4, and Palantir, the most expensive large name, at 75.[10]
What the multiples mean depends on growth. A company whose revenue triples in a year is cheap at 30 times this year’s sales if it keeps tripling, and very dear if it stops. The listed giants are priced more soberly than the laboratories: Nvidia trades at about 18 times the profits analysts expect for the coming year, against the roughly 200 times earnings Cisco reached at the peak of the internet boom in March 2000.[10] The market as a whole is less sober. The ten largest companies in the S&P 500, most of them tied to AI, make up about 39 per cent of the index, close to the record of 40.7 per cent at the end of 2025; the cyclically adjusted price-earnings ratio, which compares prices with ten years of profits, is about 40.5, against 44.2 at its peak in December 1999.[10] Anyone who owns an index fund owns a large part of this bet.
The money behind the money
The laboratories spend; the large listed companies build. Amazon, Alphabet, Microsoft and Meta spent about $410 billion on capital investment in 2025, mostly data centres, and plan roughly $715 to $770 billion for 2026; with Oracle, the total passes $800 billion (Figure 4).[11] For the first time, the spending exceeds what the businesses generate. Alphabet spent $44.9 billion on property and equipment in the second quarter of 2026, more than the $39.1 billion its operations brought in; Oracle’s free cash flow was negative $23.7 billion over its last financial year; Amazon’s was negative over the past twelve months.[11] The difference is borrowed or raised: the five issued about $220 billion of bonds in 2026 to mid-August; Meta financed a data centre in Louisiana through a separate vehicle that sold $27.3 billion of bonds; Oracle sold $20 billion of new shares in a single quarter, and the price of insuring its debt against default has nearly quadrupled in a year.[11]
The revenue that would repay the building has not yet arrived, but much of it has been signed for. Microsoft’s contracted future revenue reached $678 billion in June 2026; in January it said that about 45 per cent of that backlog came from OpenAI. Oracle reported $664 billion of contracted revenue, more than eight years of its current sales, much of it understood to be OpenAI’s, though Oracle has never named the customer.[12] Those backlogs are only as good as the customer’s ability to pay, and the customer’s ability to pay depends on raising more money from, among others, the same suppliers.
That circle is what worries even sympathetic analysts. In September 2025 Nvidia announced that it intended to invest “up to $100 billion” in OpenAI as OpenAI deployed Nvidia’s systems; the agreement was never signed, and Nvidia’s chief executive said in March 2026 that it was “probably not in the cards”, after Nvidia had put $30 billion into OpenAI’s round instead. AMD gave OpenAI warrants for up to 160 million of its own shares, vesting as OpenAI buys AMD chips. Microsoft and Nvidia committed up to $15 billion to Anthropic as Anthropic committed to buy $30 billion of Microsoft’s cloud; Amazon booked $53.4 billion of income in a single quarter, mostly from the rising value of its stake in Anthropic, which is also one of its largest cloud customers.[13] The International Monetary Fund counted 14 such arrangements since 2023, with about $300 billion of equity and $650 billion of debt “in exchange for $1.4 trillion in future revenue commitments”.[14]
A $600 billion question that became a $20 trillion one
The sceptical case starts from arithmetic. In June 2024 David Cahn, a partner at the venture firm Sequoia, took Nvidia’s sales, doubled them for the rest of the data centre and doubled again for the margin the customers would need, and concluded that “AI’s $200B question is now AI’s $600B question”: the annual revenue the industry would have to earn to justify one year’s spending.[15] In October 2025, applying the same arithmetic to the industry’s plans measured in gigawatts, he reached “AI’s $20 trillion question”, and added: “the only way to justify that is AGI.”[16] The consultancy Bain estimated that “two trillion dollars in annual revenue” would be needed by 2030 to pay for the computing the industry expects to need, and that “even with AI-related savings, the world is still $800 billion short”; J.P. Morgan’s asset managers put the revenue needed for a 10 per cent return on current investment at $650 billion a year, about $35 a month for every iPhone user.[17][18]
The second strand doubts the value to customers. Jim Covello, head of equity research at Goldman Sachs, asked in June 2024: “What $1tn problem will AI solve? Replacing low-wage jobs with tremendously costly technology is basically the polar opposite of the prior technology transitions I’ve witnessed”.[19] The MIT economist Daron Acemoglu estimated the gain to productivity at “no more than a 0.71% increase in total factor productivity over 10 years”.[20] A report from an MIT group in July 2025 found that “95% of organizations are getting zero return”, a figure quoted far more often than its method, which was 52 interviews and 153 survey answers on whether specific pilot projects showed a measurable effect on profit and loss within six months; the same report found workers at over 90 per cent of the firms using AI tools on their own.[21] Covello’s point about jobs is one this blog examined in The Machine and the Hour.[a]
The third strand is about accounting. A chip that is obsolete in three years but is written off over six makes profits look larger. The investor Michael Burry called extending the useful lives of servers “one of the more common frauds of the modern era” and estimated $176 billion of understated depreciation between 2026 and 2028; “fraud” is his allegation, and the changes are disclosed estimates.[22] The companies have moved in opposite directions: Amazon shortened the life of some servers from six years to five from 2025, citing “an increased pace of technology development, particularly in the area of artificial intelligence”, while Meta lengthened most of its servers to five and a half years, which added $2.59 billion to its 2025 profit.[23] HSBC calculated that OpenAI faces a funding gap of about $207 billion by 2030 even if its revenue grows as planned.[24]
The official warnings have grown more pointed. The Bank of England’s Financial Policy Committee wrote in October 2025 that equity valuations “appear stretched, particularly for technology companies focused on Artificial Intelligence” and that “the risk of a sharp market correction has increased”.[25] In August 2026 Andrew Bailey, as chair of the Financial Stability Board, warned the G20 of “stretched asset valuations (particularly AI-related investments)”, and in October the IMF put an AI-driven correction first among the rising risks to the financial system.[14][26] In Bank of America’s survey of fund managers in October 2025, a record 54 per cent said AI stocks were in a bubble.[27]
This time there are profits
The defence starts from the difference with 2000. Many of the internet companies of that year had no profits and little revenue; the companies carrying the AI boom are among the most profitable businesses in history, and most of their share prices have risen with their earnings. Peter Oppenheimer, Goldman Sachs’s chief global equity strategist, wrote in October 2025 that “fundamental growth and robust earnings have thus far, been the primary drivers of technology sector appreciation, rather than irrational speculation”; measured against growth, the sector’s price-earnings ratio stood at 1.6, against 3.7 at the peak of the internet bubble, and Goldman’s analysts concluded that it is “not in a bubble (at least not yet)”.[16] Nvidia’s data-centre sales were $89 billion in a single quarter, up 117 per cent in a year, and its shares trade at under 18 times expected profits.[28] “There has been a lot of talk about an AI bubble,” its chief executive, Jensen Huang, said in November 2025. “From our vantage point, we see something very different.”[29]
The second argument is demand. The companies say they are short of computing, not of customers; Microsoft’s chief executive described chips “sitting in inventory that I can’t plug in” for lack of power and buildings.[29] Adoption is spreading: the share of American businesses paying for an AI product, measured from card transactions by the fintech Ramp, rose to 55.9 per cent in September 2026, although the Census Bureau’s survey of all firms, which asks a different question, finds 17 to 20 per cent using AI.[30] The cost of running a model to a given standard has fallen by between 9 and 900 times a year, depending on the task, which widens the uses that pay.[31] The same fall cuts the other way, since revenue per unit falls with it, and growth has to come from much greater volume.
The third argument is that even a bubble may be worth having. Jeff Bezos called it “a kind of industrial bubble”, as opposed to a financial one: “every experiment gets funded”, the good and the bad, and “the benefits to society from AI are going to be gigantic.”[32] Railways and fibre-optic cables were overbuilt, ruined their investors, and served the economy for decades. The analyst Ben Thompson, who argues that “bubbles have real benefits”, adds the qualification that matters here: chips wear out in a few years, unlike rails and fibre, so the lasting legacy is more likely to be power stations and chip factories than the data centres themselves.[33] Mark Zuckerberg put the bet from the spender’s side: “If we end up misspending a couple of hundred billion dollars, I think that that is going to be very unfortunate, obviously. But what I’d say is I actually think the risk is higher on the other side.”[34]
Real technology, uncertain timing
The most striking feature of the debate is how many of the people inside the industry stand in the middle. “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes,” Sam Altman told reporters in August 2025, adding: “Is AI the most important thing to happen in a very long time? My opinion is also yes.”[35] Sundar Pichai, Alphabet’s chief executive, told the BBC that the moment was “both rational and there are elements of irrationality”, and that “no company is going to be immune, including us.”[36] Dario Amodei, Anthropic’s chief executive, said in December 2025 that “there’s an inherent risk when the timing of the economic value is uncertain”, that some competitors were “YOLO-ing”, and that he did not know whether his company’s revenue a year later would be “20 billion” or “50”.[37] By the summer of 2026 Anthropic’s run-rate had passed the second.
The historian of technology most often cited in this debate, Carlota Perez, described how each technological revolution goes through a speculative “installation” phase, ending in a crash, before a long period of deployment. Commentators often place AI in such a phase. Perez herself has argued that AI is not a new revolution but a powerful part of the digital one that began in the 1970s, which would put it in the deployment phase, where the returns are supposed to come.[38] The two readings point to different futures: a crash followed by a golden age, or a long, uneven payoff without a single break.
A wobble, so far
The argument has had its tests. On 27 January 2025, after the Chinese laboratory DeepSeek released a cheap, capable model, Nvidia lost about $590 billion of market value in a day, the largest one-day loss for any company; it recovered within months.[39] In June and July 2026 South Korea’s stock market, dominated by the memory-chip makers Samsung and SK Hynix, fell by between a third and two-fifths in a few weeks.[40] This month the American market had its worst week since April 2025: the semiconductor index fell 11.4 per cent in the week to 16 October, after the report on OpenAI’s revenue, a surprise in inflation and a sharp rise in the cost of insuring the debt of CoreWeave, a company that rents out Nvidia chips and owes about $52 billion; the twenty largest AI and chip stocks lost more than $1 trillion of value in the week.[40][11] Both main indices were still up about 12 per cent for the year. It was a correction in the most exposed shares, not a crash, and it says little about which side is right.
What would decide it
The disagreement is not really about whether AI is useful; both sides accept that it is. It is about timing and distribution: whether revenue grows fast enough, and reaches the companies that spent, before the bills for the buildings, the chips and the debt fall due. Several things will show it, most of them within a year. The first is audited numbers: Anthropic’s prospectus, if it lists after the midterms, will be the first full public account of a frontier laboratory’s finances, and OpenAI’s chief executive has said its own listing will not come in 2026.[3] The second is whether recognised revenue keeps closing the gap with spending, which in 2026 runs at several times the laboratories’ combined sales. The third is margins: whether the price of using the models falls faster than the cost of serving them. The fourth is the backlogs: whether the hundreds of billions in contracts held by Microsoft, Oracle and the cloud companies turn into payments or into renegotiations.
If they hold, today’s prices may look like those of Amazon in 1999, expensive and right. If they do not, the losses will not stay with a few venture funds. They will fall on the shareholders of the largest listed companies, which means on most pension and index funds; on the lenders to data centres, including the private credit funds the Financial Stability Board is watching; and on the suppliers of chips and memory in Taiwan and Korea. Either way, the size of the bet is now large enough that its outcome is no longer only a matter for the people who placed it.
This piece was written collaboratively with Claude Opus 5.5 (Anthropic): human specification and editorial direction; machine research, analysis and drafting. Anthropic, the maker of the tool used, is one of the companies discussed; its figures and its chief executive are treated as any other source, and its own claims of profitability are reported as unverified. Research agents gathered and checked the sources; quotations were read in the original or in a full-text copy, except where noted: the remarks of Altman, Bezos, Zuckerberg, Pichai, Amodei, Huang and Nadella and Burry’s post are quoted as reported. OpenAI’s and Anthropic’s recognised revenue, losses and compute commitments come from press reports of leaked documents, not public filings; SpaceX’s are audited. Market values and multiples are from the close of 20 October 2026. The figures and multiples come from a script, scripts/ai_valuation.py; quarterly revenue multiplied by four is an annualisation, not a forecast.
Cover: computed for this essay (scripts/ai_valuation_cover.py).
Authored by: Luis Matos Ferreira — Physicist, Developer, Writer
Ideas this essay uses without stopping to explain them, and where this blog does. The letters in the text point here; the links go straight to the section.
- The Race to Become the Default — what the AI developers are competing for besides the best model.
- The Machine and the Hour — what AI has done so far to jobs, pay and hours.
- Productive by Definition — who counts as productive, and how the accounts decide it.
- The Tollkeepers — when an income is a reward and when it is a toll.
- Anthropic, “Anthropic raises Series E at $61.5B post-money valuation”, 3 March 2025; “Series F at $183B”, 2 September 2025; “Series G”, 12 February 2026; “Series H”, 28 May 2026.
- OpenAI, “OpenAI raises $122 billion to accelerate the next phase of AI”, 31 March 2026 (read in the Internet Archive copy); “Scaling AI for everyone”, 27 February 2026; Bloomberg Law on the conditions of Amazon’s commitment, and Bloomberg and Reuters, 19 October 2026, on new talks, as reported.
- Reuters, Wall Street Journal and Fortune (29 September 2026) on Anthropic’s draft registration statement; Financial Times, Reuters and Bloomberg, 14–15 October 2026, on its third-quarter revenue; Bloomberg on its adjusted margins; Fortune, September 2026, on OpenAI’s listing; all as reported.
- OpenAI, “New funding to scale the benefits of AI”, 2 October 2024, and “New funding to build towards AGI”, 31 March 2025; Bloomberg and Reuters on the January 2023 and October 2025 valuations, as reported.
- Space Exploration Technologies Corp., prospectus (Form 424B4), 12 June 2026, segment information and cloud services agreements; third-quarter 2026 results, Form 8-K, 20 October 2026; Bloomberg on the merger terms, as reported.
- Orrick, deal note on Mistral AI’s Series D, September 2026; The Information and Bloomberg on Safe Superintelligence and Thinking Machines Lab, as reported; SpaceX, third-quarter 2026 results (Cursor).
- Financial Times, 8 October 2026, as reported by Dow Jones via MarketScreener.
- Financial Times and Quartz, June 2026, on OpenAI’s 2025 accounts as obtained by E. Zitron; Wall Street Journal on OpenAI’s and Anthropic’s 2026 quarters and customer concentration; all as reported.
- S. Altman, post on X, 6 November 2025; Microsoft, Form 10-Q for the quarter to 30 September 2025; Wall Street Journal (September 2025) on Oracle; CoreWeave announcements.
- StockAnalysis and companiesmarketcap, close of 20 October 2026; on Cisco in 2000, AJ Bell and Benedict Evans, as reported; on index concentration and the cyclically adjusted price-earnings ratio, market trackers drawing on S&P and R. Shiller’s data, as reported.
- Alphabet, second-quarter 2026 results; Microsoft, fourth-quarter fiscal 2026 results; Oracle, first-quarter fiscal 2027 results; Amazon and Meta second-quarter results and guidance, CNBC; BNP Paribas via Reuters on bond issuance; on Meta’s Hyperion financing and Oracle’s and CoreWeave’s credit spreads, Bloomberg, Reuters and CNBC; as reported where not linked.
- Microsoft, fourth-quarter fiscal 2026 results; Microsoft’s second-quarter fiscal 2026 call on OpenAI’s share of the backlog, as reported; Oracle, first-quarter fiscal 2027 results.
- CNBC, 22 September 2025, on Nvidia’s letter of intent; Fortune and CNBC, February–March 2026, on its replacement; AMD, Form 8-K, October 2025; Anthropic, “Microsoft, NVIDIA and Anthropic announce strategic partnerships”, 18 November 2025; Amazon, second-quarter 2026 results, as reported by CNBC.
- IMF, “Global financial markets face rising risks from AI, bond yields and cyberattacks”, October 2026 (Global Financial Stability Report, chapter 1).
- D. Cahn, “AI’s $600B Question”, Sequoia Capital, 20 June 2024.
- Goldman Sachs, Top of Mind 143, “AI: in a bubble?”, 22 October 2025 (interview with D. Cahn; P. Oppenheimer).
- Bain & Company, Global Technology Report 2025, press release, 23 September 2025.
- S. Aliaga, “How is AI being monetized?”, J.P. Morgan Asset Management, 9 January 2026.
- Goldman Sachs, Top of Mind 129, “Gen AI: too much spend, too little benefit?”, 25 June 2024 (interview with J. Covello).
- D. Acemoglu, “The Simple Macroeconomics of AI”, 5 April 2024.
- A. Challapally, C. Pease, R. Raskar and P. Chari, The GenAI Divide: State of AI in Business 2025, MIT NANDA, July 2025 (preliminary findings).
- M. Burry, post on X, 10 November 2025, as reported by CNBC and StreetInsider.
- Amazon.com, Form 10-K for 2024; Meta Platforms, Form 10-K for 2025.
- HSBC Global Research (N. Cote-Colisson), November 2025, as reported by The Register and the Financial Times.
- Bank of England, Financial Policy Committee record, October 2025.
- Financial Stability Board, “FSB Chair’s letter to G20”, 31 August 2026.
- Bank of America, Global Fund Manager Survey, October 2025, as reported by Bloomberg and Fortune.
- Nvidia, “NVIDIA announces financial results for second quarter fiscal 2027”, 26 August 2026.
- J. Huang, Nvidia earnings call, 19 November 2025; S. Nadella, BG2 podcast, November 2025; as reported.
- Ramp AI Index, via FRED (RAMPAIALL), September 2026; US Census Bureau, “AI use among businesses”, May 2026.
- Epoch AI, “LLM inference prices have fallen rapidly but unequally across tasks”, 12 March 2025.
- J. Bezos, Italian Tech Week, Turin, 3 October 2025, as reported by Bloomberg and Data Center Knowledge.
- B. Thompson, “The Benefits of Bubbles”, Stratechery, November 2025.
- M. Zuckerberg, Access podcast, September 2025, as reported by Sherwood News.
- S. Altman, remarks to reporters, August 2025, first reported by The Verge, as quoted by Entrepreneur.
- S. Pichai, interview with the BBC, November 2025, as reported by Ars Technica.
- D. Amodei, New York Times DealBook Summit, 3 December 2025, as reported by TechCrunch.
- C. Perez, Technological Revolutions and Financial Capital, Edward Elgar, 2002; “What Is AI’s Place in History?”, Project Syndicate, March 2024, as summarised by I. Wladawsky-Berger.
- Reuters, 27–28 January 2025, on Nvidia’s loss of market value, as reported.
- Korea Times, 28 July 2026, and Wolf Street on the KOSPI; CNBC, CNN and Reuters, 15–20 October 2026, on the October sell-off; as reported.
Comentários
Enviar um comentário