1 Rising Cryptocurrency I'll Be Buying in September, and 1 I'll Definitely Avoid

Bittensor rose 21% in the 30 days through Sept. 10, 2026, and its August V440 upgrade redirects newly issued TAO toward subnets demonstrating stronger demand, improving incentives for productive AI-service marketplaces. However, customer revenue was estimated at only $3 million-$15 million annually versus roughly $316 million of annual TAO issuance, underscoring that ecosystem activity remains heavily token-subsidized. Pump.fun gained 41% over the same period and has burned $354 million of tokens since launch, including more than $27 million in August, but faces intensifying competition as Robinhood Chain's Pons generated $96 million in 30-day fees—nearly double Pump.fun's level.

Analysis

TAO’s investable question is not AI demand but whether the emissions-routing change creates a credible transition from token-funded rewards to external service revenue. Concentrating emissions can improve leading subnet liquidity and miner participation, but it also creates reflexivity: token-market depth, rather than customer economics, may continue determining capital allocation. Over the next 1-3 months, TAO can outperform on scarcity/reallocation narrative; over 6-18 months, the thesis requires measurable growth in paid inference/compute demand and declining emissions-to-revenue dependence.

PUMP’s burn mechanism is economically equivalent to a variable buyback funded by highly cyclical transaction activity. That can support price during risk-on meme volume, but it has little downside protection if launch activity migrates or broader Solana speculation cools; buybacks fall precisely when holders need them most. The more interesting listed-equity read-through is HOOD: if its chain captures durable retail token issuance and trading flow, it expands the company’s crypto engagement funnel and potentially lowers dependence on third-party venues—but current evidence is insufficient to underwrite a material earnings revision.

Consensus may overvalue TAO’s apparent "AI" exposure relative to listed AI infrastructure. Unlike NVDA, whose economics are anchored in enterprise capex and backlog, TAO remains exposed to crypto liquidity, validator incentives, and token-price feedback loops. Conversely, the market may underappreciate that a successful migration toward paid services would reduce the network’s effective dilution burden and could merit a higher-quality crypto multiple; the key falsifier is that customer revenue fails to accelerate while emissions remain the dominant source of subnet economics.

Key Decisions for Investors

• Watch, do not chase TAO after the recent momentum. Initiate only on a 15-25% pullback or after independently verifiable quarterly evidence that external subnet revenue is growing faster than token emissions; size as a high-volatility satellite position with a 6-12 month horizon.
• If long TAO, hedge beta through a partial short in a broad crypto proxy such as BTC or SOL rather than treating it as a pure AI allocation. The desired return driver is improving network monetization versus crypto beta; exit if paid-service metrics remain immaterial after two reporting cycles.
• Avoid PUMP as a strategic long despite nominal supply reduction. A tactical long is justified only if fee share and burn rate stabilize against new venues for at least 30 days; otherwise the risk is a rapid compression in both transaction revenue and the multiple assigned to burns.
• Maintain HOOD on an earnings-revision watchlist for 1-3 months: upgrade only if chain activity translates into disclosed incremental crypto assets, trading revenue, or funded-account engagement. Absent those disclosures, treat ecosystem activity as sentiment-positive rather than a modeled EPS catalyst.
• For AI exposure, prefer NVDA over TAO for core capital. TAO offers convex upside if monetization inflects, but NVDA has materially clearer revenue capture; a TAO/NVDA relative-value long is premature until TAO demonstrates customer-funded economics.