Short-term crypto forecasting often looks less like weather prediction and more like refereeing a fight between macro, ETF flows, and product upgrades.
Captain Altcoin published a head-to-head, two AI forecasting systems, Grok and DeepSeek AI, were asked where Bitcoin (BTC) and Cardano (ADA) would land by the end of September 2025. Below are the published forecast bands. Important: Captain Altcoin provided the numeric outputs, but neither model’s training data, backtests, probability weights, nor implementation details were supplied with those outputs. Treat them as conditional scenario maps, not sealed forecasts.
How the models differ (high level)
Grok prioritizes technical confirmation and macro signals, such as support/resistance, multi-day price structure, and risk-off cues. DeepSeek leans harder on near-term catalysts and the compounding effect of positive events, like successful protocol upgrades, sustained ETF momentum, and regulatory progress. That difference in emphasis explains why both AIs can agree the setup looks constructive but still reach materially different price bands.
“Both AI models agree that Bitcoin and Cardano have constructive setups heading into September, but they reach very different conclusions on how much upside investors should expect.”, Captain Altcoin
Published forecasts (target: end of September 2025)
Bitcoin (BTC)
- Grok – Base case: $62, 000 to $66, 000.
- Grok – Bullish breakout: if buyers sustain a push through $65, 500-$66, 000, a run to $68, 000 or $72, 000 is possible.
- Grok – Bearish: if the $63, 000 handle is decisively lost, price could slide toward $58, 000 or $62, 000.
- DeepSeek – Base case: $60, 000 to $68, 000.
- DeepSeek – Bullish: $72, 000 or $78, 000.
- DeepSeek – Bearish: $52, 000 to $58, 000, conditional on ETF inflows drying up, a souring economy, and BTC losing the $63, 000 floor.
DeepSeek explicitly lists the assumption that “ETF inflows average above $100 million per day” for its bullish outcome; that figure is an assumption reported by Captain Altcoin and is not independently verified here.
Cardano (ADA)
- Grok – Base case: $0.14 to $0.18.
- Grok – Bullish: $0.20 or $0.25.
- Grok – Bearish: $0.12 or $0.14.
- DeepSeek – Base case: $0.15 to $0.22.
- DeepSeek – Bullish: $0.28 or $0.38.
- DeepSeek – Bearish: $0.09 to $0.13 if upgrades fail to bring users/adoption.
Why I lean toward Grok’s short-window cautions
For a two-month horizon, ETF flows and macro shocks dominate and are often lumpy. CoinDesk reported single-day spot-ETF inflows above $1 billion and noted that BlackRock’s IBIT passed $80 billion AUM unusually fast (CoinDesk, Jul 11, 2025). Those episodic moves can flip momentum within weeks. Grok’s focus on technical confirmation, waiting for decisive breakouts or breakdowns around key handles like $63, 000-$66, 000, better reflects that flow-driven reality for short horizons.
For ADA, protocol upgrades matter in a tangible way. DMarketForces documented the Van Rossem hard fork activation and a roughly 5% intraday reaction to about $0.1663 tied to reduced Plutus smart-contract costs and planned infrastructure handovers (DMarketForces). That makes DeepSeek’s catalyst-heavy upside plausible, but turning upgrade headlines into sustained price gains historically requires measurable upticks in developer activity, dApp usage, and transaction economics, not just good press. Given those dependencies, Grok’s mid-range ($0.20-$0.25) feels more likely to hold in a short window than DeepSeek’s $0.28-$0.38, unless on-chain usage accelerates.
Concrete catalysts that will decide which scenario wins
- Spot Bitcoin ETF flows: sustained inflows or sudden outflows materially change net demand. CoinDesk’s Jul 11, 2025 coverage shows flows can be episodic and large, undercutting steady-average assumptions.
- Federal Reserve policy (July 29-30, 2025 FOMC): changes in rate guidance or the macro outlook can move liquidity and risk appetite; the FOMC statement is typically released at the meeting close (federalreserve.gov).
- U.S. legislative and regulatory signals: progress on bills that clarify custody or ETF rules (the CLARITY Act was referenced by the models) would ease institutional uncertainty if it advances.
- Cardano network upgrades: the Van Rossem hard fork (reported by DMarketForces) and subsequent Plutus cost improvements, Pogun initiative milestones, and infrastructure handovers are the primary on-chain drivers for ADA.
- Market structure and sentiment: futures basis, funding rates, and realized volatility can amplify or mute ETF-driven moves.
“ETF demand remains the biggest pillar supporting Bitcoin.”, Captain Altcoin
Limitations and things the models did not disclose
Captain Altcoin published the numeric bands and some scenario assumptions, but crucial model provenance is missing: who operates “Grok” and “DeepSeek AI” in this context, what data feeds they used (on-chain metrics, order books, ETF flow APIs, macro indicators), whether outputs are probabilistic or deterministic, and how past performance looks. Without those details, treat the bands as conditional hypotheses tied to explicit assumptions (for example, DeepSeek’s $100M/day inflow assumption).
Some jargon also lacks a public definition in the coverage: terms such as “DOG Mode” were mentioned without explanation; when you see unexplained model jargon, ask for the exact definition and the data behind it before relying on it for trading or planning.
Practical monitoring checklist (what to watch, daily/weekly)
- Net daily spot-ETF flows and AUM changes (watch for spikes that can shift momentum; CoinDesk shows these are episodic).
- FOMC statement and press conference language for July 29-30, 2025 (federalreserve.gov).
- Cardano official changelogs and IOG/Foundation posts for Van Rossem activation notes, Plutus cost metrics, and infrastructure handover progress (use primary sources where possible).
- On-chain adoption metrics for ADA: weekly active addresses, Plutus script executions, and gas-cost trends.
- Market structure signals for BTC: futures open interest, basis, funding rates, and on-exchange balances.
How to use these AI outputs, three practical steps
- Validate the assumptions: confirm ETF-flow averages, upgrade activation timestamps, and the calendar of regulatory events before treating a model’s bullish case as likely.
- Wait for confirmation: for short windows, require decisive technical confirmation, such as multi-day closes above breakout levels or sustained inflows, before increasing allocation.
- Size for scenarios: treat AI bands as scenario buckets, size positions so the worst case in your plan matches your risk tolerance if a bearish scenario materializes.
Key takeaways, quick Q&A
-
How high could Bitcoin go by the end of September 2025?
Grok’s base is $62k, $66k with upside to $68k or $72k after a confirmed breakout; DeepSeek’s base is $60k, $68k with bullish targets at $72k or $78k. Both outcomes depend heavily on ETF inflows and macro conditions.
-
Is DeepSeek’s “$100M/day” ETF-inflow assumption realistic?
It’s an explicit assumption reported by DeepSeek/ Captain Altcoin. CoinDesk (Jul 11, 2025) shows ETF inflows can be much larger on some days (> $1B) but are episodic, so a steady $100M/day average is possible but not guaranteed, verify against daily flow reports before treating it as a baseline.
-
What could push Cardano toward the high end of these ranges?
Successful Van Rossem activation that lowers Plutus execution costs, visible increases in dApp activity and Plutus executions, smooth infrastructure handovers, and a supportive macro and liquidity backdrop could lift ADA into DeepSeek’s higher bands; without measurable adoption gains, upside is more likely to be limited and short-lived.
-
Can ADA reach $10 soon?
Captain Altcoin’s FAQ framed $10 as technically possible but highly ambitious because it implies a very large market capitalization; most mainstream models see double digits as a long-term outcome, not a short-term one.
-
Which forecast should investors use?
Use the AI outputs for scenario planning, not as a single trade trigger. For a weeks-to-months horizon, the model emphasizing technical confirmation and macro caution (Grok) is more defensible; the catalyst-heavy model (DeepSeek) is plausible if multiple favorable events align, but that alignment is a lower-probability stack.
Final note
AI models can structure messy inputs into testable scenarios, but their value depends on transparency. If you rely on these forecasts, ask for the raw outputs, the probability weights attached to each band, definitions for any jargon, and historical backtests. Combine AI scenarios with live ETF flow data (CoinDesk and ETF providers), the Fed calendar (federalreserve.gov), and primary Cardano sources (IOG/Foundation changelogs) before sizing positions. Forecasts are maps, keep one eye on the compass and another on the clock.