Speculative-predictions piece (id 5, Step 3 gate): the evidence base is thin — minimal owner signal, no weekly pulse, several access failures. Every bet below is capped at Medium confidence, each carries a falsifiable prove-it-wrong trigger, and the intro you're reading is the required label. Three honest speculative bets beat five padded ones. Time horizon: through March 2027. Revisit date for all three: April 2027.
No prior predictions exist — this is the first edition, so there's no scoreboard to open with. What follows is direction, not noise: the threads in the evidence with visible causal logic, stated as bets.
Bet 1: The filter becomes the headline number
The bet: By March 2027, at least one major purifier brand's product page or flagship retailer listing will lead with an annual filter-cost figure alongside the unit price.
Why: the causal thread is the corpus's most-recurring complaint. Sans's $156–160/yr 3-month cycle is the most-cited drawback in its reviews; Molekule's ~$350/yr is the defining complaint in its coverage; HouseFresh already publishes a $145.94/yr combined max-use running-cost estimate for the Mighty2. The filter invoice is already the buying conversation in reviews — the product pages just haven't caught up. Fire season makes replacement indicators light up in living rooms across the West, and the first brand to publish the number turns a weakness into a trust signal.
Confidence: Medium — capped per the speculative gate. The thread is real (reviewer complaint recurrence + HouseFresh's precedent), but it's one tester's practice, not an industry movement, and brands have strong incentives to keep the blade pricing quiet.
What would prove it wrong: April 2027 arrives with no major brand or flagship listing leading on annual filter cost — product pages still burying the replacement price below the fold.
Bet 2: True CO2 sensing goes downmarket
The bet: By March 2027, a sub-$150 monitor or a purifier under $300 will ship with a true NDIR CO2 sensor as a headline feature — not an index, not an estimate. This is a forecast about a marketed feature, not evidence that NDIR is the only technology that can measure CO2 or that the label guarantees accuracy.
Why: the AirGradient ONE kit at $138 already proved the price point is reachable, and the 2026 Changchun study gave the platform a snippet-derived reported accuracy signal for the tested platform, not validation of every sensor or all uses. Meanwhile the gaps are glaring and widely documented: the September 23, 2026 dossiers record no CO2 sensor for PurpleAir Classic, Flex and Zen; the full current line was not verified. Those dated dossiers also list no CO2 sensor for IKEA VINDSTYRKA and only PM2.5 for Temtop S1. Post-COVID ventilation awareness isn't fading, and "no CO2" is becoming a conspicuous absence in spec tables. Someone fills it from below.
Confidence: Medium — capped. The price-point proof exists (ONE kit) and the demand signal is plausible, but the owner-demand evidence is thin — no directly observed buyer chorus asking for CO2 — and supply chains move slower than logic.
What would prove it wrong: the sub-$150 monitor shelf in April 2027 is still PM-only (Temtop S1 class unchanged), and no purifier under $300 advertises true NDIR CO2.
Bet 3: The paywall arrives for your air data
The bet: By March 2027, at least one monitor brand in this corpus will move a currently-free data feature — history, API access, or multi-device dashboards — behind a paid tier, or publish explicit subscription terms where none existed.
Why: the thread is already visible. PurpleAir's API is already priced (details UNVERIFIED). AirGradient's dashboard is "free for now" per its own stale comparison page — "for now" doing heavy lifting. A single community source claims Airthings tiers limit data access (single-sourced, UNVERIFIED — flagged, not asserted). The hardware economics push this way: monitors are low-margin boxes, and the recurring revenue has to come from somewhere — filters for purifiers, data for monitors. The Coway rental post (directly observed, May 2025) shows the industry already thinks in subscriptions.
Confidence: Medium — capped, and the thinnest of the three. The "free for now" language and the priced API are real signals, but the Airthings claim is single-sourced and the AirGradient terms are stale. This is a real thread, thinly evidenced — exactly what the speculative cap is for.
What would prove it wrong: April 2027 finds AirGradient's dashboard still free with no published paid tier, Airthings publishing explicit free-forever data terms, and PurpleAir's API pricing unchanged — the paywall fails to materialize.
The through-line
All three bets point the same way: the air quality business is migrating from selling boxes to selling ongoing relationships — filter subscriptions, data tiers, cost-of-ownership transparency as a competitive weapon. The box was never the business; the bets are about when the industry admits it.
What could blow all three up: a major independent testing program — the measured large-class and specialty-class head-to-heads the documented sample says don't exist — landing in 2027 and resetting the conversation to performance instead of pricing models. Measurement is the exogenous shock this market hasn't had. If it arrives, the bets above get rewritten around it.