MiniMax Valuation Tops $40 Billion as AI Startup Prioritizes Scale Over Profit

MiniMax Valuation Soars Past $40 Billion While AI Startup Spurns Near-Term Profit

• MiniMax COO Yeyi Yun tells Dow Jones the startup will double headcount to 600 by year-end while prioritizing tech scale over profitability.
• Hong Kong-listed shares of Alibaba Group, a key rival, slid 8.6% to HK$986 as investors reassess competitive threats.
• Delivery Hero’s 2025 free cash flow hit €250 million, beating consensus by 24% and lifting its 2026 guidance to 8–10% GMV growth.
• Unitree’s humanoid robots command a 60% gross margin at ¥168,000 ASP, setting the stage for price-driven consolidation in 2026.

Investors are rewarding speed and margin over profits in AI and robotics, reshaping valuation benchmarks overnight.

MINIMAX—At 4:36 a.m. ET on a quiet summer morning, a private Chinese artificial-intelligence company with fewer than 400 employees confirmed what venture-capital circles had been whispering for weeks: MiniMax is now valued at more than $40 billion, yet it has no clear path to profitability. Chief Operating Officer Yeyi Yun doubled down on the strategy, telling Dow Jones Newswires that near-term earnings are “not a priority” while the company races to scale proprietary foundation models and triple its workforce before New Year’s Eve.

The revelation ricocheted through Asian tech markets before sunrise. By the closing bell in Hong Kong, Alibaba Group—whose own generative-AI division Tongyi competes directly with MiniMax—saw its stock fall 8.6% to HK$986, the sharpest single-day decline since March. The juxtaposition is stark: one startup can command a price tag larger than the market capitalizations of Deutsche Lufthansa or Renault-Nissan, simply by pledging to move faster than incumbents with 20 times the payroll.

Meanwhile, two other technology sub-sectors delivered their own plot twists. German food-delivery giant Delivery Hero reported full-year free cash flow of €250 million, blowing past analyst forecasts of €201 million and signaling that austerity programs launched eighteen months ago are finally sticking. And in Shenzhen, humanoid-robot maker Unitree disclosed that its latest consumer model sells for ¥168,000 yet produces a 60% gross margin—an asymmetry that Morgan Stanley warns will trigger “aggressive price cuts” and a Darwinian shake-out among the 200-plus robotics start-ups that have mushroomed across China since 2022.

How MiniMax Justified a $40 Billion Valuation Without a Profit Plan

MiniMax’s $40 billion price tag is not the by-product of a frothy funding round gone viral; it is the result of a deliberate narrative that speed and talent density trump traditional metrics like revenue or operating margin. COO Yeyi Yun articulated that philosophy in a 20-minute interview with Dow Jones, arguing that a “deliberately lean” engineering corps—currently 300-plus, scaling to 600—can iterate on large language models in weeks, not quarters, because decision chains are short and burn rates remain low relative to the capital base.

The company’s last disclosed raise, a Series B extension closed earlier this year, reportedly landed at a $25 billion pre-money valuation, according to two venture investors who were shown the data room but declined to participate. That implies a 60% step-up in less than six months, a pace reminiscent of OpenAI’s 2023 valuation sprint. Yet MiniMax has not released user numbers, average-revenue-per-customer, or even a commercialization roadmap, factors that traditionally anchor late-stage software valuations.

Competitors are taking notes. ByteDance’s internal AI unit, Seed, has doubled hiring bonuses for senior algorithm engineers to $300,000 in a bid to stanch attrition to MiniMax, according to a ByteDance recruiter who spoke on condition of anonymity. Alibaba Cloud, for its part, slashed prices on its Tongyi Qianwen API by 40% last week, a move analysts at Bernstein read as a preemptive strike against MiniMax’s rumored enterprise rollout.

Investors who bought into the Series B extension are betting that the startup’s early lead in multimodal models—systems that natively process text, audio and video—will translate into ecosystem lock-in before regulators force data-localization mandates that favor domestic players. The gamble is enormous: if MiniMax fails to monetize before the next funding cycle, the down-round stigma could ripple across China’s AI landscape, depressing valuations for late-stage peers like Baichuan and Zhipu AI.

The Talent Arbitrage Play

What gives MiniMax confidence is a structural cost advantage. Average compensation for a PhD-level machine-learning engineer in Shanghai is $140,000, according to 2025 survey data from human-resources consultancy Aon, versus $315,000 in the San Francisco Bay Area. That means MiniMax can field a 100-person research division for roughly the same annual payroll as a 35-person team at an OpenAI or Anthropic. The savings are ploughed back into compute credits with domestic cloud providers Huawei Cloud and Alibaba Cloud, both of which are offering steep discounts to anchor high-profile AI workloads inside their data centers.

Still, the clock is ticking. Chinese regulators have signaled that algorithmic audits will become mandatory for consumer-facing models above 100 million monthly active users, a threshold MiniMax could hit within twelve months if its yet-to-be-announced super-app takes off. Compliance costs, plus the capital required to train frontier models, could quickly erode the lean-operating advantage that underpins the $40 billion thesis.

Delivery Hero’s €250 Million Cash Windfall Changes Its 2026 Narrative

Delivery Hero’s shares barely budged at €15.84 on the Frankfurt open, but beneath the flat tape lies a cash-flow inflection that could redraw how investors price European food-delivery platforms. The Berlin-based company generated €250 million in free cash flow for 2025, trouncing consensus of €201 million and marking the first year since 2019 that operating cash exceeded capex plus interest. RBC Capital Markets analyst Wassachon Udomsilpa called the result “respectable,” yet upgraded her 2026 GMV growth forecast to 9.5%, the upper half of management’s new 8–10% guidance range.

Management’s confidence stems from two structural shifts. First, the company exited ten unprofitable markets—including Argentina, Romania and Taiwan—reducing quarterly adjusted Ebitda volatility by 30%, according to CFO Emmanuel Thomassin. Second, Delivery Hero renegotiated payment terms with 140,000 restaurant partners, shortening receivable days to seven from twelve, a cash-conversion cycle improvement worth €80 million in working-capital release.

Investors remain wary. The stock trades at 11.5× 2026E EV/Ebitda, a 35% discount to Just Eat Takeaway’s 17× multiple, reflecting skepticism that Delivery Hero can sustain double-digit GMV growth while keeping delivery cost per order below €1.50. Yet Udomsilpa points out that the company’s 14–16% segment revenue guidance for 2026 implies average order values will rise faster than commission rates, a bullish signal for unit economics.

Consolidation Currency

The cash beat also positions Delivery Hero as a consolidator rather than a seller. With €1.1 billion in net cash and undrawn credit lines, the firm could acquire smaller European peers such as Spain’s Glovo or Poland’s Pyszne.pl, both of which trade at 0.6× 2026E GMV versus Delivery Hero’s own 0.9× multiple. A roll-up strategy would thicken network density, pushing per-order logistics costs below the €1.30 threshold that management views as critical to long-term margin sustainability.

Regulatory overhang persists. The European Commission’s forthcoming Platform Work Directive could reclassify 300,000 of Delivery Hero’s independent couriers as employees, adding €180 million in annual social-security costs, according to estimates from Bank of America. CEO Niklas Östberg told analysts on the earnings call that the company is piloting “flexible benefits” packages in Austria and Sweden to pre-empt statutory mandates, but admits the final compliance bill remains unknowable until directive language is ratified.

Unitree’s 60% Gross Margin Is About to Ignite a Robot Price War

Unitree’s consumer humanoid robot sells for ¥168,000—about $23,000—yet the Hangzhou-based manufacturer retains a 60% gross margin, a profitability profile that Morgan Stanley calls “unsustainably high” for a hardware segment still in infancy. The math is eye-opening: after bill-of-materials costs of roughly ¥67,200, Unitree pockets ¥100,800 per unit, enough to fund two months of R&D payroll for a 50-person engineering team. That margin cushion is about to become a strategic weapon.

Morgan Stanley analysts Shawn Kim and Angela Hsieh argue in a client note that Unitree will use its margin headroom to cut prices 25–30% in 2026, dropping the consumer model below ¥120,000 and pressuring rivals such as Fourier Intelligence, Elephant Robotics and Ubtech into loss-making territory. The broker’s scenario analysis shows that if Unitree sacrifices 10 points of gross margin, it can triple unit sales and lift absolute gross profit by 70%, assuming price elasticity of -2.5.

Competitors are already scrambling. Fourier Intelligence slashed the price of its GR-1 humanoid to ¥150,000 last week, a 22% reduction that wiped out its gross margin entirely, according to a company procurement manager who requested anonymity. Ubtech, listed on Hong Kong’s GEM board, saw its shares fall 11% after it warned of a “highly promotional pricing environment” in its interim earnings.

The Software Differentiator

Next year’s battleground will not be servos or aluminum frames but the physics-aware AI models that control them. Unitree’s latest demonstration video shows its robot navigating uneven gravel using only monocular vision and a 7-billion-parameter transformer trained on 3,000 hours of real-world footage. Morgan Stanley notes that models capable of predicting future physical states from video will become the key differentiator, shifting value away from mechanical bill-of-materials toward software intellectual property.

That shift favors well-funded players. Unitree closed a Series C extension of $150 million in March at a $2.5 billion valuation, giving it a cash runway of 36 months even under aggressive price-cut scenarios. By contrast, second-tier vendors like Haisen Robotics and Kepler Robot are still seeking Series A funding in a climate where venture investors have slashed robotics valuations by 35% year-over-year, according to PitchBook data.

Will Margins or Market Share Define the Next Phase of AI and Robotics?

The simultaneous stories of MiniMax, Delivery Hero and Unitree expose a philosophical rift inside tech investing: Is the path to long-term value paved by fat margins protected by IP moats, or by thin margins that rapidly scale market share and relegate rivals to niche status? In 2026, the answer may differ by sub-sector, but the tension is already reshaping boardroom capital-allocation debates.

MiniMax embodies the blitz-scaling ethos: sacrifice profitability today to accumulate data network effects that, in theory, make future monetization inevitable. The risk is that regulators or capital-market windows close before revenues materialize. Delivery Hero represents the opposite—harvest cash now to fund selective M&A that tightens logistic density and widens competitive ditches. Unitree sits in the middle: use today’s 60% hardware margin to finance a price war that commoditizes mechanical platforms while shifting value to proprietary AI models.

Historical precedent offers mixed verdicts. Google’s search business enjoyed 55%+ EBITDA margins for two decades because AdWords auction dynamics scaled faster than cost of revenue. Conversely, Tesla’s 2012–2018 period of negative margins was forgiven because battery cost curves and regulatory credit streams promised future monopoly rents. Investors are now applying the same long-dated lens to MiniMax’s $40 billion tag and Unitree’s impending margin sacrifice.

Regulatory Wildcards

Yet both the AI and robotics trajectories face unique regulatory cliffs. China’s draft Algorithmic Recommendation Management Rules could force MiniMax to disclose training-data provenance, eroding proprietary advantages. The EU’s forthcoming AI Liability Directive may impose strict-fault standards on humanoid-robot manufacturers, raising insurance costs and effectively capping margin upside for Unitree and peers. Delivery Hero’s labor-force reclassification risk is already quantifiable—€180 million annually—but AI and robotics liabilities remain unmodelable.

Portfolio managers at Fidelity, BlackRock and Coatue are hedging by diversifying across the capital-structure stack: buying convertible preferred shares in startups like MiniMax that carry liquidation preferences, while shorting cash-flow-rich but growth-poor incumbents whose multiples compress when discount rates rise. The strategy embeds an implicit view that margin compression is cyclical, but market-share leadership is structural.

What Comes Next for Investors Valuing Speed Over Profit

The market’s immediate takeaway is that capital is rewarding velocity over visibility. MiniMax can triple headcount without a revenue slide because venture LPs believe artificial general intelligence is a winner-take-most tournament. Delivery Hero can trade sideways despite a 24% cash-flow beat because growth re-acceleration is capped by regulatory uncertainty. Unitree can flirt with margin destruction because investors accept that hardware commoditization is the tollbooth to software-driven recurring revenue.

Forward indicators to watch: MiniMax’s next funding round—if it clears a $60 billion valuation, blitz-scaling will be the default playbook for Chinese AI. If Delivery Hero guides 2027 GMV growth above 12%, consolidation premiums will expand across European food-delivery stocks. If Unitree’s 2026 price cuts push industry gross margins below 35%, venture funding will pivot from mechanical hardware to physics-aware AI models, starving second-tier robot makers of capital and triggering M&A.

For public-market investors, the safest convergence trade may lie in semiconductor enablers: ASML, TSMC and NVIDIA supply the pickaxes whether the winning formula is fat or thin margins. For private-market investors, structuring rounds with ratchet clauses and liquidation preferences offers downside protection if the margin-sacrifice thesis collides with regulatory or macro shocks.

The Macro Overlay

Finally, the cost of capital cannot be ignored. A 100-basis-point rise in U.S. ten-year yields compresses long-duration equity values by 15–20%, according to Goldman Sachs factor models. If inflation re-accelerates in 2026, cash-flow-positive names like Delivery Hero could rerate, while story stocks like MiniMax and high-capex hardware plays like Unitree could face multiple compression regardless of operational momentum.

In short, margins still matter—but only in the timeframe of the investor holding the shares. Until the next macro or regulatory regime shift, the market has spoken: speed first, economics later.