"The World's 10 Most Valuable Companies in 2036: Three Newcomers Will Break In"
Ten years is a long time in markets, and it is the right horizon for this question, because a decade is about how long it takes the top of the list to turn over. The baseline, from CompaniesMarketCap's September 8, 2026 ranking, is: NVIDIA, Apple, Alphabet, Microsoft, Amazon, TSMC, SpaceX, Broadcom, Saudi Aramco, and Meta. The historical base rate is what makes a prediction possible at all. Comparing published decade snapshots, five of the ten names changed between 2015 and 2026 — Apple, Alphabet, Microsoft, Amazon, and Meta survived — and between 2005 and 2026, only Microsoft survived. Against that background, predicting three newcomers by 2036 is not bold. It is below the historical rate.
The first reason the list will still be dominated by the AI stack is that the spending behind it is still accelerating. The four hyperscalers are on course for roughly $733 billion in capital expenditure in 2026, up about 79% from 2025. Inference costs have collapsed — a GPT-4-equivalent that cost $30 per million tokens in March 2023 costs about $0.40 in 2026 — and IDC projects AI compute hardware growing from $56 billion in 2023 to $702 billion in 2029. TSMC is the winner-agnostic toll road on that buildout: its August 2026 revenue grew 53.3% year on year, and it fabricates the chips of nearly every contender. NVIDIA stays near the top but drops from first place, because custom silicon is taking share — custom ASICs are projected at roughly 28% of the AI server market in 2026, with NVIDIA's share falling from around 70%. The company that sells the picks and shovels to everyone keeps its place; the company that owns one architecture, however dominant, does not keep the crown forever.
The newcomers arrive by three different roads. OpenAI is seeking a funding round at roughly $1.4 trillion pre-money, with an IPO pushed to 2027 — a private company already valued at top-ten scale. Anthropic raised at a $965 billion post-money valuation in May 2026, filed a confidential S-1 in June 2026, and is reported to be targeting a November 2026 IPO at up to around $2 trillion; that last figure is a press report, not a priced offering, and should be read as such. Eli Lilly is the non-AI entrant, and the most instructive: tirzepatide is on course for more than $45 billion in 2026 sales, its oral GLP-1 pill was FDA-approved in April 2026, and Lilly already sits at number 15 at roughly $1 trillion. The closest miss is SpaceX, number 7 today after its June 2026 IPO: a roughly $2 trillion valuation on roughly $23 billion of revenue, with Starlink at $11.4 billion in 2025 revenue, leaves it one speculative leap — orbital AI compute — short of a durable top-ten seat.
The dropouts follow from the same logic. Meta spends like a hyperscaler without owning a cloud toll road, so its capital expenditure has no captive customer to bill. Saudi Aramco faces oil cyclicality and the energy transition, and it is the only current top-ten name with a negative five-year return in the October 2026 data. SpaceX's exclusion is the judgment call most likely to be wrong: if orbital data centers become real, it has the launch monopoly to own them, and it would displace the weakest incumbent instead. My predicted 2036 list, in order, is: Alphabet, TSMC, NVIDIA, Microsoft, Amazon, OpenAI, Apple, Broadcom, Anthropic, and Eli Lilly.
The formal prediction is deliberately narrower than that list, because a ranked list of ten is ten predictions wearing a trench coat, and only the composition claim is cleanly decidable: at December 31, 2036, at least three of the world's top ten publicly traded companies by market cap will not have been in the top ten in October 2026, and at least one of those newcomers will be a company that was still privately held in October 2026. This is a prediction about composition, not about price targets — no 2036 market-cap values are claimed, and no prediction market prices a 2036 top-ten list, so this is my own judgment. The honest uncertainties are substantial: an AI capital expenditure bust, with Moody's noting hyperscaler spending above $700 billion in 2026 against twelve to twenty-four months for revenue to arrive; an antitrust breakup, which courts have so far declined but a decade of remedies could still deliver; a Taiwan shock, given TSMC makes roughly 90% of the most advanced chips; and the unnameable platform shift, which replaced half the list in each of the last two transitions. But the base rate is on the prediction's side: in every decade on record, the top ten has turned over by at least half, and three newcomers would be a slow decade, not a fast one.