June 2026 marks a period of intense valuation recalibration for the AI large model sector. MiniMax, once considered one of the "AI Twin Stars" of the Hong Kong stock market, has seen a particularly pronounced decline during this industry-wide correction.
According to Gate’s stock trading page, as of June 15, 2026, MiniMax is trading at $432, up 1.7%. For comparison, its all-time high in March this year was $1,330. This sharp drop from peak levels raises questions about the underlying market consensus. Can the company’s fundamentals support a recovery in its share price?
What Do Pricing Strategy Fluctuations Reveal About the Competitive Landscape?
A mid-June report from JPMorgan downgraded MiniMax’s rating from "Overweight" to "Neutral," slashing its target price from HK$1,100 to HK$400—a reduction of about 64%. The report’s rationale was not a slowdown in revenue growth; in fact, JPMorgan raised its revenue forecasts for MiniMax by 34%-74% for 2026-2027. Instead, the downgrade centers on a deeper issue: pricing power.
The immediate trigger for this adjustment was a pricing move by MiniMax. On June 1, the company launched its flagship M3 model at roughly twice the price of its predecessor, M2.7. Yet just a week later, MiniMax permanently cut the M3 price by 50%, bringing it back in line with M2.7. JPMorgan’s analysis suggests that, with AI demand still outstripping inference supply, no developer would willingly give up premium pricing. The rapid price drop reflects that M3’s intelligence improvements may not have fully justified the initial launch price.
In contrast, Zhipu AI has doubled its API prices this year while maintaining sales growth. This difference highlights a core theme: as industry use cases converge around APIs, code generation, intelligent agents, and enterprise workflows, leading model capabilities become increasingly vital. Pricing power is shifting from broad product coverage to the strength of the model itself.
How Does Lock-Up Expiry Pressure Affect Valuation Repricing?
Expectations around the unlocking of restricted shares are another key factor weighing on MiniMax’s stock price. Public information shows that on July 9, 2026, MiniMax will see its first major lock-up expiry, involving about 146 million shares. Approximately 46.44% of shares will transition from locked to tradable status, increasing supply nearly tenfold compared to the current free float.
Over one-third of these shares are held by financial investors, who typically have clear exit timelines and have already enjoyed substantial gains since the IPO. The lock-up expiry window often aligns with strong profit-taking motivation.
CICC’s analysis further highlights structural differences in lock-up pressure: while both MiniMax and Zhipu AI face upcoming unlocks, Zhipu’s unlocked shares represent only 11.6% of its capital, mostly held by state-backed investors with higher holding stability. MiniMax’s unlock, by contrast, represents a larger share of Hong Kong-listed equity and is more heavily concentrated among financial investors, posing a more direct supply-side shock.
Additionally, the scarcity premium for global AI large model IPOs is gradually being diluted. OpenAI and Anthropic have both quietly filed for IPOs, StepStar is preparing a Hong Kong prospectus, and Moonshot has launched a new financing round. As more comparable companies enter the market, the valuation premium for existing AI stocks faces systemic compression.
Is the Product Pricing Turmoil a Short-Term Event or a Structural Risk?
On June 1, MiniMax not only launched the M3 model but also switched its billing method from per-use to per-token. The minimum package price rose from RMB 29 to RMB 49, with no advance notice to users. Developers found that actual token consumption for equivalent tasks far exceeded expectations, with cost increases as high as 257%. One user calculated that previously, consuming 3-5 billion tokens per month cost just RMB 49; after the change, the same usage would cost about RMB 175.
This incident quickly escalated into a trust crisis. That evening, the company issued an apology, admitting "we failed to communicate this adjustment adequately in advance," and offered compensation measures such as preserving unlimited quota rights for existing users.
Technically, per-token billing is the industry standard, adopted by Anthropic, OpenAI, and other global peers. The issue, however, lies in execution: abrupt changes without sufficient buffer periods or clear communication can cause significant disruption, especially in a price-sensitive developer market.
More fundamentally, the pricing controversy exposes structural tensions in MiniMax’s business model. Over 70% of revenue comes from overseas consumer products (mainly the virtual companion app Talkie), yet gross margin for this segment is just 4.7%. Given the high costs of model R&D, profitability here is extremely limited. As compute costs continue to rise, the previous low-price strategy becomes unsustainable, making price increases almost inevitable. However, how MiniMax raises prices and maintains user trust will directly impact perceptions of its platform transformation capabilities.
Do Industry Capital Flows Still Support AI Large Model Valuations?
On a broader scale, the overall capital landscape for AI remains fundamentally unchanged. As of June 2, 2026, southbound funds have net purchased about HK$280.8 billion year-to-date, mainly flowing into high-dividend defensive sectors and AI technology growth stocks. Globally, Microsoft, Google, and the other top four cloud providers have maintained their combined $700 billion AI capex plans for 2026. The World Semiconductor Trade Statistics organization forecasts a roughly 90% year-on-year increase in the global semiconductor market, with the market size set to surpass $1.5 trillion.
However, capital flows are evolving. As industry use cases become clearer, investment logic is shifting from speculative themes to performance certainty. Investors are avoiding crowded sectors and focusing on stocks with reasonable valuations and clear growth paths. This shift increases the risk of valuation compression for companies lacking pricing power.
On June 15, BofA Securities initiated coverage of MiniMax with a "Buy" rating and a target price of HK$500. They believe that as performance gaps between leading models narrow, sustainable differentiation will increasingly depend on compute access, product distribution, and commercialization execution—not just marginal benchmark test outperformance. MiniMax, they argue, holds certain advantages in these areas. BofA forecasts MiniMax’s revenue will grow from $79 million in 2025 to $2.8 billion in 2028.
Can Fundamentals Support a Long-Term Valuation Recovery?
The key to whether MiniMax’s valuation can rebound lies in the pace and direction of its fundamental development. In 2025, total revenue reached $79.038 million, up 158.9% year-on-year, with over 70% from international markets. Gross profit jumped from $3.738 million in 2024 to $20.079 million, with gross margin rising from 12.2% to 25.4%. By year-end 2025, MiniMax had served over 236 million users across more than 200 countries and regions, with 214,000 enterprise customers and developers.
Growth momentum continued into 2026. By February, ARR (annual recurring revenue) exceeded $150 million. Daily token consumption for the M2 series text models grew more than sixfold compared to December 2025, with Coding Plan token usage up more than tenfold. On the expense side, there was structural optimization: sales and distribution costs fell 40.3% year-on-year, while R&D spending rose 33.8%, though this increase was significantly lower than revenue growth.
However, profitability remains a challenge. MiniMax posted a $1.87 billion loss for the year, with about $1.6 billion attributed to "fair value loss on financial liabilities." Adjusted net loss still stood at $250 million. Achieving profitability will require more time and continued improvements in model efficiency.
Looking ahead, Yuanta Securities highlights several upcoming catalysts: the planned launch of the Conch new model on June 26, lock-up expiry on July 8, potential inclusion in Stock Connect on August 6, and the possible release of the full M3 version in the second half of the year. These events could influence market sentiment and valuation trends in the short term.
How Will Structural Industry Divergence Impact Individual Stocks?
The AI large model sector is undergoing a structural shift from "scale narratives" to "value realization." Investors are losing patience with high-investment, low-output models, and the market is separating companies into two groups: those with sustainable pricing power able to launch leading models and convert them into commercial value, and those forced to compete on price, anchoring to industry low points and facing systemic valuation resets.
In the Hong Kong stock market, this polarization continues. Capital is flowing heavily into a handful of AI concept stocks, while funds exit other sectors. As AI investment logic shifts toward performance certainty, investors are avoiding crowded trades and favoring assets with reasonable valuations and clear growth trajectories.
Over longer industry cycles, history shows that disruptive technology sectors never lack for deep corrections. The recent global AI sell-off was triggered by short-term macro and trading factors—better-than-expected US non-farm payrolls boosting Treasury yields, and profit-taking after excessive gains—not by fundamental industry turning points such as major cloud providers slashing compute budgets or mass AI order cancellations. From this perspective, the sector’s underlying growth logic remains intact, but divergence between individual stocks will become more pronounced.
Conclusion
MiniMax’s sharp pullback from historic highs is the result of multiple overlapping factors. Pricing strategy adjustments have raised doubts about its pricing power, while lock-up expiry expectations have intensified supply-side pressure in the short term. Missteps in communicating product billing changes have further affected market perceptions of the company’s platform capabilities. These issues are amplified against the backdrop of a systemic revaluation in the AI sector, creating a multi-dimensional impact.
On the fundamentals side, MiniMax continues to deliver rapid revenue growth, improving gross margins, and expanding its overseas market share. Key variables for future performance include whether the M3 model’s market acceptance can support prices returning to reasonable premium levels, how the July lock-up shock is absorbed by the market, and whether new model launches and product iterations can demonstrate differentiated competitiveness. The stark contrast between BofA’s buy rating and JPMorgan’s downgrade underscores growing institutional divergence on MiniMax’s valuation.
For investors focused on the AI sector, the core challenge is not simply deciding whether the "AI bubble has burst." Instead, it’s identifying which companies possess sustainable pricing power and commercialization execution amid structural industry divergence.
FAQ
1. What are the main drivers behind MiniMax’s recent share price correction?
This correction is the result of multiple factors: M3’s initial launch price was set too high and quickly reduced, raising doubts about pricing ability; July’s large-scale lock-up expiry is creating supply-side pressure; miscommunication during billing model changes has triggered a trust crisis; and in the context of the AI sector’s systemic revaluation, companies lacking pricing power face greater risks of valuation compression.
2. Is MiniMax’s financial foundation healthy?
In 2025, revenue rose 158.9% year-on-year to $79.038 million, with gross margin improving from 12.2% to 25.4% and R&D efficiency steadily increasing. By February 2026, ARR exceeded $150 million, and business growth continued. Adjusted net loss was $250 million, with the profitability turning point yet to arrive, though losses are narrowing. The company secured about $1.1 billion in cash reserves post-IPO, so short-term liquidity pressure is limited.
3. How significant is the impact of July’s large-scale lock-up expiry on the share price?
The expiry involves about 146 million shares, representing roughly 46.44% of issued shares, with supply set to increase nearly tenfold. Financial investors hold a large portion and are motivated to realize gains. Based on past high-ratio unlocks in tech stocks, there is some short- to medium-term selling pressure, but the market has already priced in part of this expectation.
4. What are MiniMax’s core differences compared to peers?
The main differences lie in pricing power and the pace of leading model iteration. Unlike peers who have maintained rising API prices, MiniMax’s rapid price adjustment after the M3 launch has been interpreted as aligning with DeepSeek’s price range, weakening its premium narrative. This is the core logic behind the current valuation discount.
5. What factors could catalyze a share price recovery?
Short-term catalysts to watch include: whether the Conch new model launching on June 26 outperforms expectations in terms of performance or user acceptance; how quickly and to what extent the market absorbs the July 8 lock-up expiry; potential inclusion in Stock Connect on August 6, which could bring incremental capital allocation; and the ultimate market performance of the full M3 version expected later this year.
